Fix sandbox detection for user shell commands (#6094)

This PR introduces the ability for `core` to send `warnings` as it can
send `errors. It also sends a warning on compaction.

<img width="811" height="187" alt="image"
src="https://github.com/user-attachments/assets/0947a42d-b720-420d-b7fd-115f8a65a46a"
/>

.codespellrc

@@ -1,6 +1,6 @@
 [codespell]
 # Ref: https://github.com/codespell-project/codespell#using-a-config-file
-skip = .git*,vendor,*-lock.yaml,*.lock,.codespellrc,*test.ts,*.jsonl
+skip = .git*,vendor,*-lock.yaml,*.lock,.codespellrc,*test.ts,*.jsonl,frame*.txt
 check-hidden = true
 ignore-regex = ^\s*"image/\S+": ".*|\b(afterAll)\b
 ignore-words-list = ratatui,ser

.github/ISSUE_TEMPLATE/2-bug-report.yml

@@ -20,6 +20,14 @@ body:
     attributes:
       label: What version of Codex is running?
       description: Copy the output of `codex --version`
+    validations:
+      required: true
+  - type: input
+    id: plan
+    attributes:
+      label: What subscription do you have?
+    validations:
+      required: true
   - type: input
     id: model
     attributes:
@@ -32,11 +40,18 @@ body:
       description: |
         For MacOS and Linux: copy the output of `uname -mprs`
         For Windows: copy the output of `"$([Environment]::OSVersion | ForEach-Object VersionString) $(if ([Environment]::Is64BitOperatingSystem) { "x64" } else { "x86" })"` in the PowerShell console
+  - type: textarea
+    id: actual
+    attributes:
+      label: What issue are you seeing?
+      description: Please include the full error messages and prompts with PII redacted. If possible, please provide text instead of a screenshot. 
+    validations:
+      required: true
   - type: textarea
     id: steps
     attributes:
       label: What steps can reproduce the bug?
-      description: Explain the bug and provide a code snippet that can reproduce it.
+      description: Explain the bug and provide a code snippet that can reproduce it. Please include session id, token limit usage, context window usage if applicable.
     validations:
       required: true
   - type: textarea
@@ -44,11 +59,6 @@ body:
     attributes:
       label: What is the expected behavior?
       description: If possible, please provide text instead of a screenshot.
-  - type: textarea
-    id: actual
-    attributes:
-      label: What do you see instead?
-      description: If possible, please provide text instead of a screenshot.
   - type: textarea
     id: notes
     attributes:

.github/ISSUE_TEMPLATE/4-feature-request.yml

@@ -2,7 +2,6 @@ name: 🎁 Feature Request
 description: Propose a new feature for Codex
 labels:
   - enhancement
-  - needs triage
 body:
   - type: markdown
     attributes:
@@ -19,11 +18,6 @@ body:
       label: What feature would you like to see?
     validations:
       required: true
-  - type: textarea
-    id: author
-    attributes:
-      label: Are you interested in implementing this feature?
-      description: Please wait for acknowledgement before implementing or opening a PR.
   - type: textarea
     id: notes
     attributes:

.github/ISSUE_TEMPLATE/5-vs-code-extension.yml

@@ -14,11 +14,21 @@ body:
     id: version
     attributes:
       label: What version of the VS Code extension are you using?
+    validations:
+      required: true
+  - type: input
+    id: plan
+    attributes:
+      label: What subscription do you have?
+    validations:
+      required: true
   - type: input
     id: ide
     attributes:
       label: Which IDE are you using?
       description: Like `VS Code`, `Cursor`, `Windsurf`, etc.
+    validations:
+      required: true
   - type: input
     id: platform
     attributes:
@@ -26,11 +36,18 @@ body:
       description: |
         For MacOS and Linux: copy the output of `uname -mprs`
         For Windows: copy the output of `"$([Environment]::OSVersion | ForEach-Object VersionString) $(if ([Environment]::Is64BitOperatingSystem) { "x64" } else { "x86" })"` in the PowerShell console
+  - type: textarea
+    id: actual
+    attributes:
+      label: What issue are you seeing?
+      description: Please include the full error messages and prompts with PII redacted. If possible, please provide text instead of a screenshot. 
+    validations:
+      required: true
   - type: textarea
     id: steps
     attributes:
       label: What steps can reproduce the bug?
-      description: Explain the bug and provide a code snippet that can reproduce it.
+      description: Explain the bug and provide a code snippet that can reproduce it. Please include session id, token limit usage, context window usage if applicable.
     validations:
       required: true
   - type: textarea
@@ -38,11 +55,6 @@ body:
     attributes:
       label: What is the expected behavior?
       description: If possible, please provide text instead of a screenshot.
-  - type: textarea
-    id: actual
-    attributes:
-      label: What do you see instead?
-      description: If possible, please provide text instead of a screenshot.
   - type: textarea
     id: notes
     attributes:

.github/dotslash-config.json

@@ -27,6 +27,34 @@
           "path": "codex.exe"
         }
       }
+    },
+    "codex-responses-api-proxy": {
+      "platforms": {
+        "macos-aarch64": {
+          "regex": "^codex-responses-api-proxy-aarch64-apple-darwin\\.zst$",
+          "path": "codex-responses-api-proxy"
+        },
+        "macos-x86_64": {
+          "regex": "^codex-responses-api-proxy-x86_64-apple-darwin\\.zst$",
+          "path": "codex-responses-api-proxy"
+        },
+        "linux-x86_64": {
+          "regex": "^codex-responses-api-proxy-x86_64-unknown-linux-musl\\.zst$",
+          "path": "codex-responses-api-proxy"
+        },
+        "linux-aarch64": {
+          "regex": "^codex-responses-api-proxy-aarch64-unknown-linux-musl\\.zst$",
+          "path": "codex-responses-api-proxy"
+        },
+        "windows-x86_64": {
+          "regex": "^codex-responses-api-proxy-x86_64-pc-windows-msvc\\.exe\\.zst$",
+          "path": "codex-responses-api-proxy.exe"
+        },
+        "windows-aarch64": {
+          "regex": "^codex-responses-api-proxy-aarch64-pc-windows-msvc\\.exe\\.zst$",
+          "path": "codex-responses-api-proxy.exe"
+        }
+      }
     }
   }
 }

.github/prompts/issue-deduplicator.txt

@@ -0,0 +1,18 @@
+You are an assistant that triages new GitHub issues by identifying potential duplicates.
+
+You will receive the following JSON files located in the current working directory:
+- `codex-current-issue.json`: JSON object describing the newly created issue (fields: number, title, body).
+- `codex-existing-issues.json`: JSON array of recent issues (each element includes number, title, body, createdAt).
+
+Instructions:
+- Load both files as JSON and review their contents carefully. The codex-existing-issues.json file is large, ensure you explore all of it.
+- Compare the current issue against the existing issues to find up to five that appear to describe the same underlying problem or request.
+- Only consider an issue a potential duplicate if there is a clear overlap in symptoms, feature requests, reproduction steps, or error messages.
+- Prioritize newer issues when similarity is comparable.
+- Ignore pull requests and issues whose similarity is tenuous.
+- When unsure, prefer returning fewer matches.
+
+Output requirements:
+- Respond with a JSON array of issue numbers (integers), ordered from most likely duplicate to least.
+- Include at most five numbers.
+- If you find no plausible duplicates, respond with `[]`.

.github/prompts/issue-labeler.txt

@@ -0,0 +1,26 @@
+You are an assistant that reviews GitHub issues for the repository.
+
+Your job is to choose the most appropriate existing labels for the issue described later in this prompt.
+Follow these rules:
+- Only pick labels out of the list below.
+- Prefer a small set of precise labels over many broad ones.
+- If none of the labels fit, respond with an empty JSON array: []
+- Output must be a JSON array of label names (strings) with no additional commentary.
+
+Labels to apply:
+1. bug — Reproducible defects in Codex products (CLI, VS Code extension, web, auth).
+2. enhancement — Feature requests or usability improvements that ask for new capabilities, better ergonomics, or quality-of-life tweaks.
+3. extension — VS Code (or other IDE) extension-specific issues.
+4. windows-os — Bugs or friction specific to Windows environments (PowerShell behavior, path handling, copy/paste, OS-specific auth or tooling failures).
+5. mcp — Topics involving Model Context Protocol servers/clients.
+6. codex-web — Issues targeting the Codex web UI/Cloud experience.
+8. azure — Problems or requests tied to Azure OpenAI deployments.
+9. documentation — Updates or corrections needed in docs/README/config references (broken links, missing examples, outdated keys, clarification requests).
+10. model-behavior — Undesirable LLM behavior: forgetting goals, refusing work, hallucinating environment details, quota misreports, or other reasoning/performance anomalies.
+
+Issue information is available in environment variables:
+
+ISSUE_NUMBER
+ISSUE_TITLE
+ISSUE_BODY
+REPO_FULL_NAME

.github/workflows/ci.yml

@@ -1,7 +1,7 @@
 name: ci
 
 on:
-  pull_request: { branches: [main] }
+  pull_request: {}
   push: { branches: [main] }
 
 jobs:
@@ -27,26 +27,29 @@ jobs:
       - name: Install dependencies
         run: pnpm install --frozen-lockfile
 
-      # build_npm_package.py requires DotSlash when staging releases.
+      # stage_npm_packages.py requires DotSlash when staging releases.
       - uses: facebook/install-dotslash@v2
 
       - name: Stage npm package
+        id: stage_npm_package
         env:
           GH_TOKEN: ${{ github.token }}
         run: |
           set -euo pipefail
           CODEX_VERSION=0.40.0
-          PACK_OUTPUT="${RUNNER_TEMP}/codex-npm.tgz"
-          python3 ./codex-cli/scripts/build_npm_package.py \
+          OUTPUT_DIR="${RUNNER_TEMP}"
+          python3 ./scripts/stage_npm_packages.py \
             --release-version "$CODEX_VERSION" \
-            --pack-output "$PACK_OUTPUT"
-          echo "PACK_OUTPUT=$PACK_OUTPUT" >> "$GITHUB_ENV"
+            --package codex \
+            --output-dir "$OUTPUT_DIR"
+          PACK_OUTPUT="${OUTPUT_DIR}/codex-npm-${CODEX_VERSION}.tgz"
+          echo "pack_output=$PACK_OUTPUT" >> "$GITHUB_OUTPUT"
 
       - name: Upload staged npm package artifact
         uses: actions/upload-artifact@v4
         with:
           name: codex-npm-staging
-          path: ${{ env.PACK_OUTPUT }}
+          path: ${{ steps.stage_npm_package.outputs.pack_output }}
 
       - name: Ensure root README.md contains only ASCII and certain Unicode code points
         run: ./scripts/asciicheck.py README.md
@@ -57,3 +60,6 @@ jobs:
         run: ./scripts/asciicheck.py codex-cli/README.md
       - name: Check codex-cli/README ToC
         run: python3 scripts/readme_toc.py codex-cli/README.md
+
+      - name: Prettier (run `pnpm run format:fix` to fix)
+        run: pnpm run format

.github/workflows/codespell.yml

@@ -25,4 +25,3 @@ jobs:
         uses: codespell-project/actions-codespell@406322ec52dd7b488e48c1c4b82e2a8b3a1bf630 # v2.1
         with:
           ignore_words_file: .codespellignore
-          skip: frame*.txt

.github/workflows/issue-deduplicator.yml

@@ -0,0 +1,140 @@
+name: Issue Deduplicator
+
+on:
+  issues:
+    types:
+      - opened
+      - labeled
+
+jobs:
+  gather-duplicates:
+    name: Identify potential duplicates
+    if: ${{ github.event.action == 'opened' || (github.event.action == 'labeled' && github.event.label.name == 'codex-deduplicate') }}
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+    outputs:
+      codex_output: ${{ steps.codex.outputs.final-message }}
+    steps:
+      - uses: actions/checkout@v5
+
+      - name: Prepare Codex inputs
+        env:
+          GH_TOKEN: ${{ github.token }}
+        run: |
+          set -eo pipefail
+
+          CURRENT_ISSUE_FILE=codex-current-issue.json
+          EXISTING_ISSUES_FILE=codex-existing-issues.json
+
+          gh issue list --repo "${{ github.repository }}" \
+            --json number,title,body,createdAt \
+            --limit 1000 \
+            --state all \
+            --search "sort:created-desc" \
+            | jq '.' \
+            > "$EXISTING_ISSUES_FILE"
+
+          gh issue view "${{ github.event.issue.number }}" \
+            --repo "${{ github.repository }}" \
+            --json number,title,body \
+            | jq '.' \
+            > "$CURRENT_ISSUE_FILE"
+
+      - id: codex
+        uses: openai/codex-action@main
+        with:
+          openai-api-key: ${{ secrets.CODEX_OPENAI_API_KEY }}
+          allow-users: "*"
+          model: gpt-5
+          prompt: |
+            You are an assistant that triages new GitHub issues by identifying potential duplicates.
+
+            You will receive the following JSON files located in the current working directory:
+            - `codex-current-issue.json`: JSON object describing the newly created issue (fields: number, title, body).
+            - `codex-existing-issues.json`: JSON array of recent issues (each element includes number, title, body, createdAt).
+
+            Instructions:
+            - Compare the current issue against the existing issues to find up to five that appear to describe the same underlying problem or request.
+            - Focus on the underlying intent and context of each issue—such as reported symptoms, feature requests, reproduction steps, or error messages—rather than relying solely on string similarity or synthetic metrics.
+            - After your analysis, validate your results in 1-2 lines explaining your decision to return the selected matches.
+            - When unsure, prefer returning fewer matches.
+            - Include at most five numbers.
+
+          output-schema: |
+            {
+              "type": "object",
+              "properties": {
+                "issues": {
+                  "type": "array",
+                  "items": {
+                    "type": "string"
+                  }
+                },
+                "reason": { "type": "string" }
+              },
+              "required": ["issues", "reason"],
+              "additionalProperties": false
+            }
+
+  comment-on-issue:
+    name: Comment with potential duplicates
+    needs: gather-duplicates
+    if: ${{ needs.gather-duplicates.result != 'skipped' }}
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+      issues: write
+    steps:
+      - name: Comment on issue
+        uses: actions/github-script@v8
+        env:
+          CODEX_OUTPUT: ${{ needs.gather-duplicates.outputs.codex_output }}
+        with:
+          github-token: ${{ github.token }}
+          script: |
+            const raw = process.env.CODEX_OUTPUT ?? '';
+            let parsed;
+            try {
+              parsed = JSON.parse(raw);
+            } catch (error) {
+              core.info(`Codex output was not valid JSON. Raw output: ${raw}`);
+              core.info(`Parse error: ${error.message}`);
+              return;
+            }
+
+            const issues = Array.isArray(parsed?.issues) ? parsed.issues : [];
+            const currentIssueNumber = String(context.payload.issue.number);
+
+            console.log(`Current issue number: ${currentIssueNumber}`);
+            console.log(issues);
+
+            const filteredIssues = issues.filter((value) => String(value) !== currentIssueNumber);
+
+            if (filteredIssues.length === 0) {
+              core.info('Codex reported no potential duplicates.');
+              return;
+            }
+
+            const lines = [
+              'Potential duplicates detected. Please review them and close your issue if it is a duplicate.',
+              '',
+              ...filteredIssues.map((value) => `- #${String(value)}`),
+              '',
+              '*Powered by [Codex Action](https://github.com/openai/codex-action)*'];
+
+            await github.rest.issues.createComment({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              issue_number: context.payload.issue.number,
+              body: lines.join("\n"),
+            });
+
+      - name: Remove codex-deduplicate label
+        if: ${{ always() && github.event.action == 'labeled' && github.event.label.name == 'codex-deduplicate' }}
+        env:
+          GH_TOKEN: ${{ github.token }}
+          GH_REPO: ${{ github.repository }}
+        run: |
+          gh issue edit "${{ github.event.issue.number }}" --remove-label codex-deduplicate || true
+          echo "Attempted to remove label: codex-deduplicate"

.github/workflows/issue-labeler.yml

@@ -0,0 +1,115 @@
+name: Issue Labeler
+
+on:
+  issues:
+    types:
+      - opened
+      - labeled
+
+jobs:
+  gather-labels:
+    name: Generate label suggestions
+    if: ${{ github.event.action == 'opened' || (github.event.action == 'labeled' && github.event.label.name == 'codex-label') }}
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+    outputs:
+      codex_output: ${{ steps.codex.outputs.final-message }}
+    steps:
+      - uses: actions/checkout@v5
+
+      - id: codex
+        uses: openai/codex-action@main
+        with:
+          openai-api-key: ${{ secrets.CODEX_OPENAI_API_KEY }}
+          allow-users: "*"
+          prompt: |
+            You are an assistant that reviews GitHub issues for the repository.
+
+            Your job is to choose the most appropriate existing labels for the issue described later in this prompt.
+            Follow these rules:
+            - Only pick labels out of the list below.
+            - Prefer a small set of precise labels over many broad ones.
+
+            Labels to apply:
+            1. bug — Reproducible defects in Codex products (CLI, VS Code extension, web, auth).
+            2. enhancement — Feature requests or usability improvements that ask for new capabilities, better ergonomics, or quality-of-life tweaks.
+            3. extension — VS Code (or other IDE) extension-specific issues.
+            4. windows-os — Bugs or friction specific to Windows environments (always when PowerShell is mentioned, path handling, copy/paste, OS-specific auth or tooling failures).
+            5. mcp — Topics involving Model Context Protocol servers/clients.
+            6. codex-web — Issues targeting the Codex web UI/Cloud experience.
+            8. azure — Problems or requests tied to Azure OpenAI deployments.
+            9. documentation — Updates or corrections needed in docs/README/config references (broken links, missing examples, outdated keys, clarification requests).
+            10. model-behavior — Undesirable LLM behavior: forgetting goals, refusing work, hallucinating environment details, quota misreports, or other reasoning/performance anomalies.
+
+            Issue number: ${{ github.event.issue.number }}
+
+            Issue title:
+            ${{ github.event.issue.title }}
+
+            Issue body:
+            ${{ github.event.issue.body }}
+
+            Repository full name:
+            ${{ github.repository }}
+
+          output-schema: |
+            {
+              "type": "object",
+              "properties": {
+                "labels": {
+                  "type": "array",
+                  "items": {
+                    "type": "string"
+                  }
+                }
+              },
+              "required": ["labels"],
+              "additionalProperties": false
+            }
+
+  apply-labels:
+    name: Apply labels from Codex output
+    needs: gather-labels
+    if: ${{ needs.gather-labels.result != 'skipped' }}
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+      issues: write
+    env:
+      GH_TOKEN: ${{ github.token }}
+      GH_REPO: ${{ github.repository }}
+      ISSUE_NUMBER: ${{ github.event.issue.number }}
+      CODEX_OUTPUT: ${{ needs.gather-labels.outputs.codex_output }}
+    steps:
+      - name: Apply labels
+        run: |
+          json=${CODEX_OUTPUT//$'\r'/}
+          if [ -z "$json" ]; then
+            echo "Codex produced no output. Skipping label application."
+            exit 0
+          fi
+
+          if ! printf '%s' "$json" | jq -e 'type == "object" and (.labels | type == "array")' >/dev/null 2>&1; then
+            echo "Codex output did not include a labels array. Raw output: $json"
+            exit 0
+          fi
+
+          labels=$(printf '%s' "$json" | jq -r '.labels[] | tostring')
+          if [ -z "$labels" ]; then
+            echo "Codex returned an empty array. Nothing to do."
+            exit 0
+          fi
+
+          cmd=(gh issue edit "$ISSUE_NUMBER")
+          while IFS= read -r label; do
+            cmd+=(--add-label "$label")
+          done <<< "$labels"
+
+          "${cmd[@]}" || true
+
+      - name: Remove codex-label trigger
+        if: ${{ always() && github.event.action == 'labeled' && github.event.label.name == 'codex-label' }}
+        run: |
+          gh issue edit "$ISSUE_NUMBER" --remove-label codex-label || true
+          echo "Attempted to remove label: codex-label"

.github/workflows/rust-ci.yml

@@ -9,7 +9,7 @@ on:
 # CI builds in debug (dev) for faster signal.
 
 jobs:
-  # --- Detect what changed (always runs) -------------------------------------
+  # --- Detect what changed to detect which tests to run (always runs) -------------------------------------
   changed:
     name: Detect changed areas
     runs-on: ubuntu-24.04
@@ -84,8 +84,8 @@ jobs:
         run: cargo shear
 
   # --- CI to validate on different os/targets --------------------------------
-  lint_build_test:
-    name: ${{ matrix.runner }} - ${{ matrix.target }}${{ matrix.profile == 'release' && ' (release)' || '' }}
+  lint_build:
+    name: Lint/Build — ${{ matrix.runner }} - ${{ matrix.target }}${{ matrix.profile == 'release' && ' (release)' || '' }}
     runs-on: ${{ matrix.runner }}
     timeout-minutes: 30
     needs: changed
@@ -94,6 +94,11 @@ jobs:
     defaults:
       run:
         working-directory: codex-rs
+    env:
+      # Speed up repeated builds across CI runs by caching compiled objects.
+      RUSTC_WRAPPER: sccache
+      CARGO_INCREMENTAL: "0"
+      SCCACHE_CACHE_SIZE: 10G
 
     strategy:
       fail-fast: false
@@ -148,20 +153,94 @@ jobs:
           targets: ${{ matrix.target }}
           components: clippy
 
-      - uses: actions/cache@v4
+      # Explicit cache restore: split cargo home vs target, so we can
+      # avoid caching the large target dir on the gnu-dev job.
+      - name: Restore cargo home cache
+        id: cache_cargo_home_restore
+        uses: actions/cache/restore@v4
         with:
           path: |
             ~/.cargo/bin/
             ~/.cargo/registry/index/
             ~/.cargo/registry/cache/
             ~/.cargo/git/db/
-            ${{ github.workspace }}/codex-rs/target/
-          key: cargo-${{ matrix.runner }}-${{ matrix.target }}-${{ matrix.profile }}-${{ hashFiles('**/Cargo.lock') }}
+          key: cargo-home-${{ matrix.runner }}-${{ matrix.target }}-${{ matrix.profile }}-${{ hashFiles('**/Cargo.lock') }}-${{ hashFiles('codex-rs/rust-toolchain.toml') }}
+          restore-keys: |
+            cargo-home-${{ matrix.runner }}-${{ matrix.target }}-${{ matrix.profile }}-
+
+      # Install and restore sccache cache
+      - name: Install sccache
+        uses: taiki-e/install-action@0c5db7f7f897c03b771660e91d065338615679f4 # v2
+        with:
+          tool: sccache
+          version: 0.7.5
+
+      - name: Configure sccache backend
+        shell: bash
+        run: |
+          set -euo pipefail
+          if [[ -n "${ACTIONS_CACHE_URL:-}" && -n "${ACTIONS_RUNTIME_TOKEN:-}" ]]; then
+            echo "SCCACHE_GHA_ENABLED=true" >> "$GITHUB_ENV"
+            echo "Using sccache GitHub backend"
+          else
+            echo "SCCACHE_GHA_ENABLED=false" >> "$GITHUB_ENV"
+            echo "SCCACHE_DIR=${{ github.workspace }}/.sccache" >> "$GITHUB_ENV"
+            echo "Using sccache local disk + actions/cache fallback"
+          fi
+
+      - name: Restore sccache cache (fallback)
+        if: ${{ env.SCCACHE_GHA_ENABLED != 'true' }}
+        id: cache_sccache_restore
+        uses: actions/cache/restore@v4
+        with:
+          path: ${{ github.workspace }}/.sccache/
+          key: sccache-${{ matrix.runner }}-${{ matrix.target }}-${{ matrix.profile }}-${{ hashFiles('**/Cargo.lock') }}-${{ github.run_id }}
+          restore-keys: |
+            sccache-${{ matrix.runner }}-${{ matrix.target }}-${{ matrix.profile }}-${{ hashFiles('**/Cargo.lock') }}-
+            sccache-${{ matrix.runner }}-${{ matrix.target }}-${{ matrix.profile }}-
+
+      - if: ${{ matrix.target == 'x86_64-unknown-linux-musl' || matrix.target == 'aarch64-unknown-linux-musl'}}
+        name: Prepare APT cache directories (musl)
+        shell: bash
+        run: |
+          set -euo pipefail
+          sudo mkdir -p /var/cache/apt/archives /var/lib/apt/lists
+          sudo chown -R "$USER:$USER" /var/cache/apt /var/lib/apt/lists
+
+      - if: ${{ matrix.target == 'x86_64-unknown-linux-musl' || matrix.target == 'aarch64-unknown-linux-musl'}}
+        name: Restore APT cache (musl)
+        id: cache_apt_restore
+        uses: actions/cache/restore@v4
+        with:
+          path: |
+            /var/cache/apt
+          key: apt-${{ matrix.runner }}-${{ matrix.target }}-v1
 
       - if: ${{ matrix.target == 'x86_64-unknown-linux-musl' || matrix.target == 'aarch64-unknown-linux-musl'}}
         name: Install musl build tools
+        env:
+          DEBIAN_FRONTEND: noninteractive
+        shell: bash
         run: |
-          sudo apt install -y musl-tools pkg-config && sudo rm -rf /var/lib/apt/lists/*
+          set -euo pipefail
+          sudo apt-get -y update -o Acquire::Retries=3
+          sudo apt-get -y install --no-install-recommends musl-tools pkg-config
+
+      - name: Install cargo-chef
+        if: ${{ matrix.profile == 'release' }}
+        uses: taiki-e/install-action@0c5db7f7f897c03b771660e91d065338615679f4 # v2
+        with:
+          tool: cargo-chef
+          version: 0.1.71
+
+      - name: Pre-warm dependency cache (cargo-chef)
+        if: ${{ matrix.profile == 'release' }}
+        shell: bash
+        run: |
+          set -euo pipefail
+          RECIPE="${RUNNER_TEMP}/chef-recipe.json"
+          cargo chef prepare --recipe-path "$RECIPE"
+          cargo chef cook --recipe-path "$RECIPE" --target ${{ matrix.target }} --release --all-features
 
       - name: cargo clippy
         id: clippy
@@ -180,6 +259,146 @@ jobs:
           find . -name Cargo.toml -mindepth 2 -maxdepth 2 -print0 \
             | xargs -0 -n1 -I{} bash -c 'cd "$(dirname "{}")" && cargo check --profile ${{ matrix.profile }}'
 
+      # Save caches explicitly; make non-fatal so cache packaging
+      # never fails the overall job. Only save when key wasn't hit.
+      - name: Save cargo home cache
+        if: always() && !cancelled() && steps.cache_cargo_home_restore.outputs.cache-hit != 'true'
+        continue-on-error: true
+        uses: actions/cache/save@v4
+        with:
+          path: |
+            ~/.cargo/bin/
+            ~/.cargo/registry/index/
+            ~/.cargo/registry/cache/
+            ~/.cargo/git/db/
+          key: cargo-home-${{ matrix.runner }}-${{ matrix.target }}-${{ matrix.profile }}-${{ hashFiles('**/Cargo.lock') }}-${{ hashFiles('codex-rs/rust-toolchain.toml') }}
+
+      - name: Save sccache cache (fallback)
+        if: always() && !cancelled() && env.SCCACHE_GHA_ENABLED != 'true'
+        continue-on-error: true
+        uses: actions/cache/save@v4
+        with:
+          path: ${{ github.workspace }}/.sccache/
+          key: sccache-${{ matrix.runner }}-${{ matrix.target }}-${{ matrix.profile }}-${{ hashFiles('**/Cargo.lock') }}-${{ github.run_id }}
+
+      - name: sccache stats
+        if: always()
+        continue-on-error: true
+        run: sccache --show-stats || true
+
+      - name: sccache summary
+        if: always()
+        shell: bash
+        run: |
+          {
+            echo "### sccache stats — ${{ matrix.target }} (${{ matrix.profile }})";
+            echo;
+            echo '```';
+            sccache --show-stats || true;
+            echo '```';
+          } >> "$GITHUB_STEP_SUMMARY"
+
+      - name: Save APT cache (musl)
+        if: always() && !cancelled() && (matrix.target == 'x86_64-unknown-linux-musl' || matrix.target == 'aarch64-unknown-linux-musl') && steps.cache_apt_restore.outputs.cache-hit != 'true'
+        continue-on-error: true
+        uses: actions/cache/save@v4
+        with:
+          path: |
+            /var/cache/apt
+          key: apt-${{ matrix.runner }}-${{ matrix.target }}-v1
+
+      # Fail the job if any of the previous steps failed.
+      - name: verify all steps passed
+        if: |
+          steps.clippy.outcome == 'failure' ||
+          steps.cargo_check_all_crates.outcome == 'failure'
+        run: |
+          echo "One or more checks failed (clippy or cargo_check_all_crates). See logs for details."
+          exit 1
+
+  tests:
+    name: Tests — ${{ matrix.runner }} - ${{ matrix.target }}
+    runs-on: ${{ matrix.runner }}
+    timeout-minutes: 30
+    needs: changed
+    if: ${{ needs.changed.outputs.codex == 'true' || needs.changed.outputs.workflows == 'true' || github.event_name == 'push' }}
+    defaults:
+      run:
+        working-directory: codex-rs
+    env:
+      RUSTC_WRAPPER: sccache
+      CARGO_INCREMENTAL: "0"
+      SCCACHE_CACHE_SIZE: 10G
+
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - runner: macos-14
+            target: aarch64-apple-darwin
+            profile: dev
+          - runner: ubuntu-24.04
+            target: x86_64-unknown-linux-gnu
+            profile: dev
+          - runner: ubuntu-24.04-arm
+            target: aarch64-unknown-linux-gnu
+            profile: dev
+          - runner: windows-latest
+            target: x86_64-pc-windows-msvc
+            profile: dev
+          - runner: windows-11-arm
+            target: aarch64-pc-windows-msvc
+            profile: dev
+
+    steps:
+      - uses: actions/checkout@v5
+      - uses: dtolnay/rust-toolchain@1.90
+        with:
+          targets: ${{ matrix.target }}
+
+      - name: Restore cargo home cache
+        id: cache_cargo_home_restore
+        uses: actions/cache/restore@v4
+        with:
+          path: |
+            ~/.cargo/bin/
+            ~/.cargo/registry/index/
+            ~/.cargo/registry/cache/
+            ~/.cargo/git/db/
+          key: cargo-home-${{ matrix.runner }}-${{ matrix.target }}-${{ matrix.profile }}-${{ hashFiles('**/Cargo.lock') }}-${{ hashFiles('codex-rs/rust-toolchain.toml') }}
+          restore-keys: |
+            cargo-home-${{ matrix.runner }}-${{ matrix.target }}-${{ matrix.profile }}-
+
+      - name: Install sccache
+        uses: taiki-e/install-action@0c5db7f7f897c03b771660e91d065338615679f4 # v2
+        with:
+          tool: sccache
+          version: 0.7.5
+
+      - name: Configure sccache backend
+        shell: bash
+        run: |
+          set -euo pipefail
+          if [[ -n "${ACTIONS_CACHE_URL:-}" && -n "${ACTIONS_RUNTIME_TOKEN:-}" ]]; then
+            echo "SCCACHE_GHA_ENABLED=true" >> "$GITHUB_ENV"
+            echo "Using sccache GitHub backend"
+          else
+            echo "SCCACHE_GHA_ENABLED=false" >> "$GITHUB_ENV"
+            echo "SCCACHE_DIR=${{ github.workspace }}/.sccache" >> "$GITHUB_ENV"
+            echo "Using sccache local disk + actions/cache fallback"
+          fi
+
+      - name: Restore sccache cache (fallback)
+        if: ${{ env.SCCACHE_GHA_ENABLED != 'true' }}
+        id: cache_sccache_restore
+        uses: actions/cache/restore@v4
+        with:
+          path: ${{ github.workspace }}/.sccache/
+          key: sccache-${{ matrix.runner }}-${{ matrix.target }}-${{ matrix.profile }}-${{ hashFiles('**/Cargo.lock') }}-${{ github.run_id }}
+          restore-keys: |
+            sccache-${{ matrix.runner }}-${{ matrix.target }}-${{ matrix.profile }}-${{ hashFiles('**/Cargo.lock') }}-
+            sccache-${{ matrix.runner }}-${{ matrix.target }}-${{ matrix.profile }}-
+
       - uses: taiki-e/install-action@0c5db7f7f897c03b771660e91d065338615679f4 # v2
         with:
           tool: nextest
@@ -187,27 +406,58 @@ jobs:
 
       - name: tests
         id: test
-        # Tests take too long for release builds to run them on every PR.
-        if: ${{ matrix.profile != 'release' }}
         continue-on-error: true
-        run: cargo nextest run --all-features --no-fail-fast --target ${{ matrix.target }}
+        run: cargo nextest run --all-features --no-fail-fast --target ${{ matrix.target }} --cargo-profile ci-test
         env:
           RUST_BACKTRACE: 1
 
-      # Fail the job if any of the previous steps failed.
-      - name: verify all steps passed
-        if: |
-          steps.clippy.outcome == 'failure' ||
-          steps.cargo_check_all_crates.outcome == 'failure' ||
-          steps.test.outcome == 'failure'
+      - name: Save cargo home cache
+        if: always() && !cancelled() && steps.cache_cargo_home_restore.outputs.cache-hit != 'true'
+        continue-on-error: true
+        uses: actions/cache/save@v4
+        with:
+          path: |
+            ~/.cargo/bin/
+            ~/.cargo/registry/index/
+            ~/.cargo/registry/cache/
+            ~/.cargo/git/db/
+          key: cargo-home-${{ matrix.runner }}-${{ matrix.target }}-${{ matrix.profile }}-${{ hashFiles('**/Cargo.lock') }}-${{ hashFiles('codex-rs/rust-toolchain.toml') }}
+
+      - name: Save sccache cache (fallback)
+        if: always() && !cancelled() && env.SCCACHE_GHA_ENABLED != 'true'
+        continue-on-error: true
+        uses: actions/cache/save@v4
+        with:
+          path: ${{ github.workspace }}/.sccache/
+          key: sccache-${{ matrix.runner }}-${{ matrix.target }}-${{ matrix.profile }}-${{ hashFiles('**/Cargo.lock') }}-${{ github.run_id }}
+
+      - name: sccache stats
+        if: always()
+        continue-on-error: true
+        run: sccache --show-stats || true
+
+      - name: sccache summary
+        if: always()
+        shell: bash
         run: |
-          echo "One or more checks failed (clippy, cargo_check_all_crates, or test). See logs for details."
+          {
+            echo "### sccache stats — ${{ matrix.target }} (tests)";
+            echo;
+            echo '```';
+            sccache --show-stats || true;
+            echo '```';
+          } >> "$GITHUB_STEP_SUMMARY"
+
+      - name: verify tests passed
+        if: steps.test.outcome == 'failure'
+        run: |
+          echo "Tests failed. See logs for details."
           exit 1
 
   # --- Gatherer job that you mark as the ONLY required status -----------------
   results:
     name: CI results (required)
-    needs: [changed, general, cargo_shear, lint_build_test]
+    needs: [changed, general, cargo_shear, lint_build, tests]
     if: always()
     runs-on: ubuntu-24.04
     steps:
@@ -216,7 +466,8 @@ jobs:
         run: |
           echo "general: ${{ needs.general.result }}"
           echo "shear  : ${{ needs.cargo_shear.result }}"
-          echo "matrix : ${{ needs.lint_build_test.result }}"
+          echo "lint   : ${{ needs.lint_build.result }}"
+          echo "tests  : ${{ needs.tests.result }}"
 
           # If nothing relevant changed (PR touching only root README, etc.),
           # declare success regardless of other jobs.
@@ -228,4 +479,10 @@ jobs:
           # Otherwise require the jobs to have succeeded
           [[ '${{ needs.general.result }}' == 'success' ]] || { echo 'general failed'; exit 1; }
           [[ '${{ needs.cargo_shear.result }}' == 'success' ]] || { echo 'cargo_shear failed'; exit 1; }
-          [[ '${{ needs.lint_build_test.result }}' == 'success' ]] || { echo 'matrix failed'; exit 1; }
+          [[ '${{ needs.lint_build.result }}' == 'success' ]] || { echo 'lint_build failed'; exit 1; }
+          [[ '${{ needs.tests.result }}' == 'success' ]] || { echo 'tests failed'; exit 1; }
+
+      - name: sccache summary note
+        if: always()
+        run: |
+          echo "Per-job sccache stats are attached to each matrix job's Step Summary."

.github/workflows/rust-release.yml

@@ -47,7 +47,7 @@ jobs:
 
   build:
     needs: tag-check
-    name: ${{ matrix.runner }} - ${{ matrix.target }}
+    name: Build - ${{ matrix.runner }} - ${{ matrix.target }}
     runs-on: ${{ matrix.runner }}
     timeout-minutes: 30
     defaults:
@@ -58,9 +58,9 @@ jobs:
       fail-fast: false
       matrix:
         include:
-          - runner: macos-14
+          - runner: macos-15-xlarge
             target: aarch64-apple-darwin
-          - runner: macos-14
+          - runner: macos-15-xlarge
             target: x86_64-apple-darwin
           - runner: ubuntu-24.04
             target: x86_64-unknown-linux-musl
@@ -94,10 +94,180 @@ jobs:
       - if: ${{ matrix.target == 'x86_64-unknown-linux-musl' || matrix.target == 'aarch64-unknown-linux-musl'}}
         name: Install musl build tools
         run: |
-          sudo apt install -y musl-tools pkg-config
+          sudo apt-get update
+          sudo apt-get install -y musl-tools pkg-config
 
       - name: Cargo build
-        run: cargo build --target ${{ matrix.target }} --release --bin codex
+        run: cargo build --target ${{ matrix.target }} --release --bin codex --bin codex-responses-api-proxy
+
+      - if: ${{ matrix.runner == 'macos-15-xlarge' }}
+        name: Configure Apple code signing
+        shell: bash
+        env:
+          KEYCHAIN_PASSWORD: actions
+          APPLE_CERTIFICATE: ${{ secrets.APPLE_CERTIFICATE_P12 }}
+          APPLE_CERTIFICATE_PASSWORD: ${{ secrets.APPLE_CERTIFICATE_PASSWORD }}
+        run: |
+          set -euo pipefail
+
+          if [[ -z "${APPLE_CERTIFICATE:-}" ]]; then
+            echo "APPLE_CERTIFICATE is required for macOS signing"
+            exit 1
+          fi
+
+          if [[ -z "${APPLE_CERTIFICATE_PASSWORD:-}" ]]; then
+            echo "APPLE_CERTIFICATE_PASSWORD is required for macOS signing"
+            exit 1
+          fi
+
+          cert_path="${RUNNER_TEMP}/apple_signing_certificate.p12"
+          echo "$APPLE_CERTIFICATE" | base64 -d > "$cert_path"
+
+          keychain_path="${RUNNER_TEMP}/codex-signing.keychain-db"
+          security create-keychain -p "$KEYCHAIN_PASSWORD" "$keychain_path"
+          security set-keychain-settings -lut 21600 "$keychain_path"
+          security unlock-keychain -p "$KEYCHAIN_PASSWORD" "$keychain_path"
+
+          keychain_args=()
+          cleanup_keychain() {
+            if ((${#keychain_args[@]} > 0)); then
+              security list-keychains -s "${keychain_args[@]}" || true
+              security default-keychain -s "${keychain_args[0]}" || true
+            else
+              security list-keychains -s || true
+            fi
+            if [[ -f "$keychain_path" ]]; then
+              security delete-keychain "$keychain_path" || true
+            fi
+          }
+
+          while IFS= read -r keychain; do
+            [[ -n "$keychain" ]] && keychain_args+=("$keychain")
+          done < <(security list-keychains | sed 's/^[[:space:]]*//;s/[[:space:]]*$//;s/"//g')
+
+          if ((${#keychain_args[@]} > 0)); then
+            security list-keychains -s "$keychain_path" "${keychain_args[@]}"
+          else
+            security list-keychains -s "$keychain_path"
+          fi
+
+          security default-keychain -s "$keychain_path"
+          security import "$cert_path" -k "$keychain_path" -P "$APPLE_CERTIFICATE_PASSWORD" -T /usr/bin/codesign -T /usr/bin/security
+          security set-key-partition-list -S apple-tool:,apple: -s -k "$KEYCHAIN_PASSWORD" "$keychain_path" > /dev/null
+
+          codesign_hashes=()
+          while IFS= read -r hash; do
+            [[ -n "$hash" ]] && codesign_hashes+=("$hash")
+          done < <(security find-identity -v -p codesigning "$keychain_path" \
+            | sed -n 's/.*\([0-9A-F]\{40\}\).*/\1/p' \
+            | sort -u)
+
+          if ((${#codesign_hashes[@]} == 0)); then
+            echo "No signing identities found in $keychain_path"
+            cleanup_keychain
+            rm -f "$cert_path"
+            exit 1
+          fi
+
+          if ((${#codesign_hashes[@]} > 1)); then
+            echo "Multiple signing identities found in $keychain_path:"
+            printf '  %s\n' "${codesign_hashes[@]}"
+            cleanup_keychain
+            rm -f "$cert_path"
+            exit 1
+          fi
+
+          APPLE_CODESIGN_IDENTITY="${codesign_hashes[0]}"
+
+          rm -f "$cert_path"
+
+          echo "APPLE_CODESIGN_IDENTITY=$APPLE_CODESIGN_IDENTITY" >> "$GITHUB_ENV"
+          echo "APPLE_CODESIGN_KEYCHAIN=$keychain_path" >> "$GITHUB_ENV"
+          echo "::add-mask::$APPLE_CODESIGN_IDENTITY"
+
+      - if: ${{ matrix.runner == 'macos-15-xlarge' }}
+        name: Sign macOS binaries
+        shell: bash
+        run: |
+          set -euo pipefail
+
+          if [[ -z "${APPLE_CODESIGN_IDENTITY:-}" ]]; then
+            echo "APPLE_CODESIGN_IDENTITY is required for macOS signing"
+            exit 1
+          fi
+
+          keychain_args=()
+          if [[ -n "${APPLE_CODESIGN_KEYCHAIN:-}" && -f "${APPLE_CODESIGN_KEYCHAIN}" ]]; then
+            keychain_args+=(--keychain "${APPLE_CODESIGN_KEYCHAIN}")
+          fi
+
+          for binary in codex codex-responses-api-proxy; do
+            path="target/${{ matrix.target }}/release/${binary}"
+            codesign --force --options runtime --timestamp --sign "$APPLE_CODESIGN_IDENTITY" "${keychain_args[@]}" "$path"
+          done
+
+      - if: ${{ matrix.runner == 'macos-15-xlarge' }}
+        name: Notarize macOS binaries
+        shell: bash
+        env:
+          APPLE_NOTARIZATION_KEY_P8: ${{ secrets.APPLE_NOTARIZATION_KEY_P8 }}
+          APPLE_NOTARIZATION_KEY_ID: ${{ secrets.APPLE_NOTARIZATION_KEY_ID }}
+          APPLE_NOTARIZATION_ISSUER_ID: ${{ secrets.APPLE_NOTARIZATION_ISSUER_ID }}
+        run: |
+          set -euo pipefail
+
+          for var in APPLE_NOTARIZATION_KEY_P8 APPLE_NOTARIZATION_KEY_ID APPLE_NOTARIZATION_ISSUER_ID; do
+            if [[ -z "${!var:-}" ]]; then
+              echo "$var is required for notarization"
+              exit 1
+            fi
+          done
+
+          notary_key_path="${RUNNER_TEMP}/notarytool.key.p8"
+          echo "$APPLE_NOTARIZATION_KEY_P8" | base64 -d > "$notary_key_path"
+          cleanup_notary() {
+            rm -f "$notary_key_path"
+          }
+          trap cleanup_notary EXIT
+
+          notarize_binary() {
+            local binary="$1"
+            local source_path="target/${{ matrix.target }}/release/${binary}"
+            local archive_path="${RUNNER_TEMP}/${binary}.zip"
+
+            if [[ ! -f "$source_path" ]]; then
+              echo "Binary $source_path not found"
+              exit 1
+            fi
+
+            rm -f "$archive_path"
+            ditto -c -k --keepParent "$source_path" "$archive_path"
+
+            submission_json=$(xcrun notarytool submit "$archive_path" \
+              --key "$notary_key_path" \
+              --key-id "$APPLE_NOTARIZATION_KEY_ID" \
+              --issuer "$APPLE_NOTARIZATION_ISSUER_ID" \
+              --output-format json \
+              --wait)
+
+            status=$(printf '%s\n' "$submission_json" | jq -r '.status // "Unknown"')
+            submission_id=$(printf '%s\n' "$submission_json" | jq -r '.id // ""')
+
+            if [[ -z "$submission_id" ]]; then
+              echo "Failed to retrieve submission ID for $binary"
+              exit 1
+            fi
+
+            echo "::notice title=Notarization::$binary submission ${submission_id} completed with status ${status}"
+
+            if [[ "$status" != "Accepted" ]]; then
+              echo "Notarization failed for ${binary} (submission ${submission_id}, status ${status})"
+              exit 1
+            fi
+          }
+
+          notarize_binary "codex"
+          notarize_binary "codex-responses-api-proxy"
 
       - name: Stage artifacts
         shell: bash
@@ -107,8 +277,10 @@ jobs:
 
           if [[ "${{ matrix.runner }}" == windows* ]]; then
             cp target/${{ matrix.target }}/release/codex.exe "$dest/codex-${{ matrix.target }}.exe"
+            cp target/${{ matrix.target }}/release/codex-responses-api-proxy.exe "$dest/codex-responses-api-proxy-${{ matrix.target }}.exe"
           else
             cp target/${{ matrix.target }}/release/codex "$dest/codex-${{ matrix.target }}"
+            cp target/${{ matrix.target }}/release/codex-responses-api-proxy "$dest/codex-responses-api-proxy-${{ matrix.target }}"
           fi
 
       - if: ${{ matrix.runner == 'windows-11-arm' }}
@@ -155,6 +327,29 @@ jobs:
             zstd -T0 -19 --rm "$dest/$base"
           done
 
+      - name: Remove signing keychain
+        if: ${{ always() && matrix.runner == 'macos-15-xlarge' }}
+        shell: bash
+        env:
+          APPLE_CODESIGN_KEYCHAIN: ${{ env.APPLE_CODESIGN_KEYCHAIN }}
+        run: |
+          set -euo pipefail
+          if [[ -n "${APPLE_CODESIGN_KEYCHAIN:-}" ]]; then
+            keychain_args=()
+            while IFS= read -r keychain; do
+              [[ "$keychain" == "$APPLE_CODESIGN_KEYCHAIN" ]] && continue
+              [[ -n "$keychain" ]] && keychain_args+=("$keychain")
+            done < <(security list-keychains | sed 's/^[[:space:]]*//;s/[[:space:]]*$//;s/"//g')
+            if ((${#keychain_args[@]} > 0)); then
+              security list-keychains -s "${keychain_args[@]}"
+              security default-keychain -s "${keychain_args[0]}"
+            fi
+
+            if [[ -f "$APPLE_CODESIGN_KEYCHAIN" ]]; then
+              security delete-keychain "$APPLE_CODESIGN_KEYCHAIN"
+            fi
+          fi
+
       - uses: actions/upload-artifact@v4
         with:
           name: ${{ matrix.target }}
@@ -214,18 +409,30 @@ jobs:
             echo "npm_tag=" >> "$GITHUB_OUTPUT"
           fi
 
-      # build_npm_package.py requires DotSlash when staging releases.
+      - name: Setup pnpm
+        uses: pnpm/action-setup@v4
+        with:
+          run_install: false
+
+      - name: Setup Node.js for npm packaging
+        uses: actions/setup-node@v5
+        with:
+          node-version: 22
+
+      - name: Install dependencies
+        run: pnpm install --frozen-lockfile
+
+      # stage_npm_packages.py requires DotSlash when staging releases.
       - uses: facebook/install-dotslash@v2
-      - name: Stage npm package
+      - name: Stage npm packages
         env:
           GH_TOKEN: ${{ github.token }}
         run: |
-          set -euo pipefail
-          TMP_DIR="${RUNNER_TEMP}/npm-stage"
-          ./codex-cli/scripts/build_npm_package.py \
+          ./scripts/stage_npm_packages.py \
             --release-version "${{ steps.release_name.outputs.name }}" \
-            --staging-dir "${TMP_DIR}" \
-            --pack-output "${GITHUB_WORKSPACE}/dist/npm/codex-npm-${{ steps.release_name.outputs.name }}.tgz"
+            --package codex \
+            --package codex-responses-api-proxy \
+            --package codex-sdk
 
       - name: Create GitHub Release
         uses: softprops/action-gh-release@v2
@@ -269,7 +476,7 @@ jobs:
       - name: Update npm
         run: npm install -g npm@latest
 
-      - name: Download npm tarball from release
+      - name: Download npm tarballs from release
         env:
           GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
         run: |
@@ -281,6 +488,14 @@ jobs:
             --repo "${GITHUB_REPOSITORY}" \
             --pattern "codex-npm-${version}.tgz" \
             --dir dist/npm
+          gh release download "$tag" \
+            --repo "${GITHUB_REPOSITORY}" \
+            --pattern "codex-responses-api-proxy-npm-${version}.tgz" \
+            --dir dist/npm
+          gh release download "$tag" \
+            --repo "${GITHUB_REPOSITORY}" \
+            --pattern "codex-sdk-npm-${version}.tgz" \
+            --dir dist/npm
 
       # No NODE_AUTH_TOKEN needed because we use OIDC.
       - name: Publish to npm
@@ -294,7 +509,15 @@ jobs:
             tag_args+=(--tag "${NPM_TAG}")
           fi
 
-          npm publish "${GITHUB_WORKSPACE}/dist/npm/codex-npm-${VERSION}.tgz" "${tag_args[@]}"
+          tarballs=(
+            "codex-npm-${VERSION}.tgz"
+            "codex-responses-api-proxy-npm-${VERSION}.tgz"
+            "codex-sdk-npm-${VERSION}.tgz"
+          )
+
+          for tarball in "${tarballs[@]}"; do
+            npm publish "${GITHUB_WORKSPACE}/dist/npm/${tarball}" "${tag_args[@]}"
+          done
 
   update-branch:
     name: Update latest-alpha-cli branch

.github/workflows/sdk.yml

@@ -0,0 +1,43 @@
+name: sdk
+
+on:
+  push:
+    branches: [main]
+  pull_request: {}
+
+jobs:
+  sdks:
+    runs-on: ubuntu-latest
+    timeout-minutes: 10
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v5
+
+      - name: Setup pnpm
+        uses: pnpm/action-setup@v4
+        with:
+          run_install: false
+
+      - name: Setup Node.js
+        uses: actions/setup-node@v5
+        with:
+          node-version: 22
+          cache: pnpm
+
+      - uses: dtolnay/rust-toolchain@1.90
+
+      - name: build codex
+        run: cargo build --bin codex
+        working-directory: codex-rs
+
+      - name: Install dependencies
+        run: pnpm install --frozen-lockfile
+
+      - name: Build SDK packages
+        run: pnpm -r --filter ./sdk/typescript run build
+
+      - name: Lint SDK packages
+        run: pnpm -r --filter ./sdk/typescript run lint
+
+      - name: Test SDK packages
+        run: pnpm -r --filter ./sdk/typescript run test

.gitignore

@@ -30,6 +30,7 @@ result
 # cli tools
 CLAUDE.md
 .claude/
+AGENTS.override.md
 
 # caches
 .cache/

.vscode/settings.json

@@ -3,6 +3,7 @@
     "rust-analyzer.check.command": "clippy",
     "rust-analyzer.check.extraArgs": ["--all-features", "--tests"],
     "rust-analyzer.rustfmt.extraArgs": ["--config", "imports_granularity=Item"],
+    "rust-analyzer.cargo.targetDir": "${workspaceFolder}/codex-rs/target/rust-analyzer",
     "[rust]": {
         "editor.defaultFormatter": "rust-lang.rust-analyzer",
         "editor.formatOnSave": true,

AGENTS.md

@@ -8,11 +8,18 @@ In the codex-rs folder where the rust code lives:
 - Never add or modify any code related to `CODEX_SANDBOX_NETWORK_DISABLED_ENV_VAR` or `CODEX_SANDBOX_ENV_VAR`.
   - You operate in a sandbox where `CODEX_SANDBOX_NETWORK_DISABLED=1` will be set whenever you use the `shell` tool. Any existing code that uses `CODEX_SANDBOX_NETWORK_DISABLED_ENV_VAR` was authored with this fact in mind. It is often used to early exit out of tests that the author knew you would not be able to run given your sandbox limitations.
   - Similarly, when you spawn a process using Seatbelt (`/usr/bin/sandbox-exec`), `CODEX_SANDBOX=seatbelt` will be set on the child process. Integration tests that want to run Seatbelt themselves cannot be run under Seatbelt, so checks for `CODEX_SANDBOX=seatbelt` are also often used to early exit out of tests, as appropriate.
+- Always collapse if statements per https://rust-lang.github.io/rust-clippy/master/index.html#collapsible_if
+- Always inline format! args when possible per https://rust-lang.github.io/rust-clippy/master/index.html#uninlined_format_args
+- Use method references over closures when possible per https://rust-lang.github.io/rust-clippy/master/index.html#redundant_closure_for_method_calls
+- Do not use unsigned integer even if the number cannot be negative.
+- When writing tests, prefer comparing the equality of entire objects over fields one by one.
+- When making a change that adds or changes an API, ensure that the documentation in the `docs/` folder is up to date if applicable.
 
 Run `just fmt` (in `codex-rs` directory) automatically after making Rust code changes; do not ask for approval to run it. Before finalizing a change to `codex-rs`, run `just fix -p <project>` (in `codex-rs` directory) to fix any linter issues in the code. Prefer scoping with `-p` to avoid slow workspace‑wide Clippy builds; only run `just fix` without `-p` if you changed shared crates. Additionally, run the tests:
+
 1. Run the test for the specific project that was changed. For example, if changes were made in `codex-rs/tui`, run `cargo test -p codex-tui`.
 2. Once those pass, if any changes were made in common, core, or protocol, run the complete test suite with `cargo test --all-features`.
-When running interactively, ask the user before running `just fix` to finalize. `just fmt` does not require approval. project-specific or individual tests can be run without asking the user, but do ask the user before running the complete test suite.
+   When running interactively, ask the user before running `just fix` to finalize. `just fmt` does not require approval. project-specific or individual tests can be run without asking the user, but do ask the user before running the complete test suite.
 
 ## TUI style conventions
 
@@ -28,6 +35,7 @@ See `codex-rs/tui/styles.md`.
     - Desired: vec!["  └ ".into(), "M".red(), " ".dim(), "tui/src/app.rs".dim()]
 
 ### TUI Styling (ratatui)
+
 - Prefer Stylize helpers: use "text".dim(), .bold(), .cyan(), .italic(), .underlined() instead of manual Style where possible.
 - Prefer simple conversions: use "text".into() for spans and vec![…].into() for lines; when inference is ambiguous (e.g., Paragraph::new/Cell::from), use Line::from(spans) or Span::from(text).
 - Computed styles: if the Style is computed at runtime, using `Span::styled` is OK (`Span::from(text).set_style(style)` is also acceptable).
@@ -39,6 +47,7 @@ See `codex-rs/tui/styles.md`.
 - Compactness: prefer the form that stays on one line after rustfmt; if only one of Line::from(vec![…]) or vec![…].into() avoids wrapping, choose that. If both wrap, pick the one with fewer wrapped lines.
 
 ### Text wrapping
+
 - Always use textwrap::wrap to wrap plain strings.
 - If you have a ratatui Line and you want to wrap it, use the helpers in tui/src/wrapping.rs, e.g. word_wrap_lines / word_wrap_line.
 - If you need to indent wrapped lines, use the initial_indent / subsequent_indent options from RtOptions if you can, rather than writing custom logic.
@@ -60,8 +69,34 @@ This repo uses snapshot tests (via `insta`), especially in `codex-rs/tui`, to va
   - `cargo insta accept -p codex-tui`
 
 If you don’t have the tool:
+
 - `cargo install cargo-insta`
 
 ### Test assertions
 
 - Tests should use pretty_assertions::assert_eq for clearer diffs. Import this at the top of the test module if it isn't already.
+
+### Integration tests (core)
+
+- Prefer the utilities in `core_test_support::responses` when writing end-to-end Codex tests.
+
+- All `mount_sse*` helpers return a `ResponseMock`; hold onto it so you can assert against outbound `/responses` POST bodies.
+- Use `ResponseMock::single_request()` when a test should only issue one POST, or `ResponseMock::requests()` to inspect every captured `ResponsesRequest`.
+- `ResponsesRequest` exposes helpers (`body_json`, `input`, `function_call_output`, `custom_tool_call_output`, `call_output`, `header`, `path`, `query_param`) so assertions can target structured payloads instead of manual JSON digging.
+- Build SSE payloads with the provided `ev_*` constructors and the `sse(...)`.
+
+- Typical pattern:
+
+  ```rust
+  let mock = responses::mount_sse_once(&server, responses::sse(vec![
+      responses::ev_response_created("resp-1"),
+      responses::ev_function_call(call_id, "shell", &serde_json::to_string(&args)?),
+      responses::ev_completed("resp-1"),
+  ])).await;
+
+  codex.submit(Op::UserTurn { ... }).await?;
+
+  // Assert request body if needed.
+  let request = mock.single_request();
+  // assert using request.function_call_output(call_id) or request.json_body() or other helpers.
+  ```

CHANGELOG.md

@@ -1 +1 @@
-The changelog can be found on the [releases page](https://github.com/openai/codex/releases)
+The changelog can be found on the [releases page](https://github.com/openai/codex/releases).

README.md

@@ -1,6 +1,4 @@
-<h1 align="center">OpenAI Codex CLI</h1>
-
-<p align="center"><code>npm i -g @openai/codex</code><br />or <code>brew install codex</code></p>
+<p align="center"><code>npm i -g @openai/codex</code><br />or <code>brew install --cask codex</code></p>
 
 <p align="center"><strong>Codex CLI</strong> is a coding agent from OpenAI that runs locally on your computer.
 </br>
@@ -26,7 +24,7 @@ npm install -g @openai/codex
 Alternatively, if you use Homebrew:
 
 ```shell
-brew install codex
+brew install --cask codex
 ```
 
 Then simply run `codex` to get started:
@@ -35,6 +33,8 @@ Then simply run `codex` to get started:
 codex
 ```
 
+If you're running into upgrade issues with Homebrew, see the [FAQ entry on brew upgrade codex](./docs/faq.md#brew-upgrade-codex-isnt-upgrading-me).
+
 <details>
 <summary>You can also go to the <a href="https://github.com/openai/codex/releases/latest">latest GitHub Release</a> and download the appropriate binary for your platform.</summary>
 
@@ -63,8 +63,7 @@ You can also use Codex with an API key, but this requires [additional setup](./d
 
 ### Model Context Protocol (MCP)
 
-Codex CLI supports [MCP servers](./docs/advanced.md#model-context-protocol-mcp). Enable by adding an `mcp_servers` section to your `~/.codex/config.toml`.
-
+Codex can access MCP servers. To configure them, refer to the [config docs](./docs/config.md#mcp_servers).
 
 ### Configuration
 
@@ -78,14 +77,18 @@ Codex CLI supports a rich set of configuration options, with preferences stored
   - [CLI usage](./docs/getting-started.md#cli-usage)
   - [Running with a prompt as input](./docs/getting-started.md#running-with-a-prompt-as-input)
   - [Example prompts](./docs/getting-started.md#example-prompts)
+  - [Custom prompts](./docs/prompts.md)
   - [Memory with AGENTS.md](./docs/getting-started.md#memory-with-agentsmd)
-  - [Configuration](./docs/config.md)
+- [**Configuration**](./docs/config.md)
 - [**Sandbox & approvals**](./docs/sandbox.md)
 - [**Authentication**](./docs/authentication.md)
   - [Auth methods](./docs/authentication.md#forcing-a-specific-auth-method-advanced)
   - [Login on a "Headless" machine](./docs/authentication.md#connecting-on-a-headless-machine)
+- **Automating Codex**
+  - [GitHub Action](https://github.com/openai/codex-action)
+  - [TypeScript SDK](./sdk/typescript/README.md)
+  - [Non-interactive mode (`codex exec`)](./docs/exec.md)
 - [**Advanced**](./docs/advanced.md)
-  - [Non-interactive / CI mode](./docs/advanced.md#non-interactive--ci-mode)
   - [Tracing / verbose logging](./docs/advanced.md#tracing--verbose-logging)
   - [Model Context Protocol (MCP)](./docs/advanced.md#model-context-protocol-mcp)
 - [**Zero data retention (ZDR)**](./docs/zdr.md)
@@ -102,4 +105,3 @@ Codex CLI supports a rich set of configuration options, with preferences stored
 ## License
 
 This repository is licensed under the [Apache-2.0 License](LICENSE).
-

cliff.toml

@@ -4,7 +4,7 @@
 header = """
 # Changelog
 
-You can install any of these versions: `npm install -g codex@version`
+You can install any of these versions: `npm install -g @openai/codex@<version>`
 """
 
 body = """

codex-cli/README.md

@@ -208,7 +208,7 @@ The hardening mechanism Codex uses depends on your OS:
 | Requirement                 | Details                                                         |
 | --------------------------- | --------------------------------------------------------------- |
 | Operating systems           | macOS 12+, Ubuntu 20.04+/Debian 10+, or Windows 11 **via WSL2** |
-| Node.js                     | **22 or newer** (LTS recommended)                               |
+| Node.js                     | **16 or newer** (Node 20 LTS recommended)                       |
 | Git (optional, recommended) | 2.23+ for built-in PR helpers                                   |
 | RAM                         | 4-GB minimum (8-GB recommended)                                 |
 
@@ -513,7 +513,7 @@ Codex runs model-generated commands in a sandbox. If a proposed command or file
 <details>
 <summary>Does it work on Windows?</summary>
 
-Not directly. It requires [Windows Subsystem for Linux (WSL2)](https://learn.microsoft.com/en-us/windows/wsl/install) - Codex has been tested on macOS and Linux with Node 22.
+Not directly. It requires [Windows Subsystem for Linux (WSL2)](https://learn.microsoft.com/en-us/windows/wsl/install) - Codex is regularly tested on macOS and Linux with Node 20+, and also supports Node 16.
 
 </details>
 

codex-cli/bin/codex.js

@@ -1,6 +1,7 @@
 #!/usr/bin/env node
 // Unified entry point for the Codex CLI.
 
+import { spawn } from "node:child_process";
 import { existsSync } from "fs";
 import path from "path";
 import { fileURLToPath } from "url";
@@ -68,7 +69,6 @@ const binaryPath = path.join(archRoot, "codex", codexBinaryName);
 // executing. This allows us to forward those signals to the child process
 // and guarantees that when either the child terminates or the parent
 // receives a fatal signal, both processes exit in a predictable manner.
-const { spawn } = await import("child_process");
 
 function getUpdatedPath(newDirs) {
   const pathSep = process.platform === "win32" ? ";" : ":";
@@ -80,6 +80,32 @@ function getUpdatedPath(newDirs) {
   return updatedPath;
 }
 
+/**
+ * Use heuristics to detect the package manager that was used to install Codex
+ * in order to give the user a hint about how to update it.
+ */
+function detectPackageManager() {
+  const userAgent = process.env.npm_config_user_agent || "";
+  if (/\bbun\//.test(userAgent)) {
+    return "bun";
+  }
+
+  const execPath = process.env.npm_execpath || "";
+  if (execPath.includes("bun")) {
+    return "bun";
+  }
+
+  if (
+    process.env.BUN_INSTALL ||
+    process.env.BUN_INSTALL_GLOBAL_DIR ||
+    process.env.BUN_INSTALL_BIN_DIR
+  ) {
+    return "bun";
+  }
+
+  return userAgent ? "npm" : null;
+}
+
 const additionalDirs = [];
 const pathDir = path.join(archRoot, "path");
 if (existsSync(pathDir)) {
@@ -87,9 +113,16 @@ if (existsSync(pathDir)) {
 }
 const updatedPath = getUpdatedPath(additionalDirs);
 
+const env = { ...process.env, PATH: updatedPath };
+const packageManagerEnvVar =
+  detectPackageManager() === "bun"
+    ? "CODEX_MANAGED_BY_BUN"
+    : "CODEX_MANAGED_BY_NPM";
+env[packageManagerEnvVar] = "1";
+
 const child = spawn(binaryPath, process.argv.slice(2), {
   stdio: "inherit",
-  env: { ...process.env, PATH: updatedPath, CODEX_MANAGED_BY_NPM: "1" },
+  env,
 });
 
 child.on("error", (err) => {

codex-cli/package-lock.json

@@ -11,7 +11,7 @@
         "codex": "bin/codex.js"
       },
       "engines": {
-        "node": ">=20"
+        "node": ">=16"
       }
     }
   }

codex-cli/package.json

@@ -7,7 +7,7 @@
   },
   "type": "module",
   "engines": {
-    "node": ">=20"
+    "node": ">=16"
   },
   "files": [
     "bin",

codex-cli/scripts/README.md

@@ -1,11 +1,19 @@
 # npm releases
 
-Run the following:
-
-To build the 0.2.x or later version of the npm module, which runs the Rust version of the CLI, build it as follows:
+Use the staging helper in the repo root to generate npm tarballs for a release. For
+example, to stage the CLI, responses proxy, and SDK packages for version `0.6.0`:
 
 ```bash
-./codex-cli/scripts/build_npm_package.py --release-version 0.6.0
+./scripts/stage_npm_packages.py \
+  --release-version 0.6.0 \
+  --package codex \
+  --package codex-responses-api-proxy \
+  --package codex-sdk
 ```
 
-Note this will create `./codex-cli/vendor/` as a side-effect.
+This downloads the native artifacts once, hydrates `vendor/` for each package, and writes
+tarballs to `dist/npm/`.
+
+If you need to invoke `build_npm_package.py` directly, run
+`codex-cli/scripts/install_native_deps.py` first and pass `--vendor-src` pointing to the
+directory that contains the populated `vendor/` tree.

codex-cli/scripts/build_npm_package.py

@@ -3,7 +3,6 @@
 
 import argparse
 import json
-import re
 import shutil
 import subprocess
 import sys
@@ -13,16 +12,29 @@ from pathlib import Path
 SCRIPT_DIR = Path(__file__).resolve().parent
 CODEX_CLI_ROOT = SCRIPT_DIR.parent
 REPO_ROOT = CODEX_CLI_ROOT.parent
-GITHUB_REPO = "openai/codex"
+RESPONSES_API_PROXY_NPM_ROOT = REPO_ROOT / "codex-rs" / "responses-api-proxy" / "npm"
+CODEX_SDK_ROOT = REPO_ROOT / "sdk" / "typescript"
 
-# The docs are not clear on what the expected value/format of
-# workflow/workflowName is:
-# https://cli.github.com/manual/gh_run_list
-WORKFLOW_NAME = ".github/workflows/rust-release.yml"
+PACKAGE_NATIVE_COMPONENTS: dict[str, list[str]] = {
+    "codex": ["codex", "rg"],
+    "codex-responses-api-proxy": ["codex-responses-api-proxy"],
+    "codex-sdk": ["codex"],
+}
+COMPONENT_DEST_DIR: dict[str, str] = {
+    "codex": "codex",
+    "codex-responses-api-proxy": "codex-responses-api-proxy",
+    "rg": "path",
+}
 
 
 def parse_args() -> argparse.Namespace:
     parser = argparse.ArgumentParser(description="Build or stage the Codex CLI npm package.")
+    parser.add_argument(
+        "--package",
+        choices=("codex", "codex-responses-api-proxy", "codex-sdk"),
+        default="codex",
+        help="Which npm package to stage (default: codex).",
+    )
     parser.add_argument(
         "--version",
         help="Version number to write to package.json inside the staged package.",
@@ -30,14 +42,9 @@ def parse_args() -> argparse.Namespace:
     parser.add_argument(
         "--release-version",
         help=(
-            "Version to stage for npm release. When provided, the script also resolves the "
-            "matching rust-release workflow unless --workflow-url is supplied."
+            "Version to stage for npm release."
         ),
     )
-    parser.add_argument(
-        "--workflow-url",
-        help="Optional GitHub Actions workflow run URL used to download native binaries.",
-    )
     parser.add_argument(
         "--staging-dir",
         type=Path,
@@ -57,12 +64,18 @@ def parse_args() -> argparse.Namespace:
         type=Path,
         help="Path where the generated npm tarball should be written.",
     )
+    parser.add_argument(
+        "--vendor-src",
+        type=Path,
+        help="Directory containing pre-installed native binaries to bundle (vendor root).",
+    )
     return parser.parse_args()
 
 
 def main() -> int:
     args = parse_args()
 
+    package = args.package
     version = args.version
     release_version = args.release_version
     if release_version:
@@ -76,40 +89,45 @@ def main() -> int:
     staging_dir, created_temp = prepare_staging_dir(args.staging_dir)
 
     try:
-        stage_sources(staging_dir, version)
+        stage_sources(staging_dir, version, package)
 
-        workflow_url = args.workflow_url
-        resolved_head_sha: str | None = None
-        if not workflow_url:
-            if release_version:
-                workflow = resolve_release_workflow(version)
-                workflow_url = workflow["url"]
-                resolved_head_sha = workflow.get("headSha")
-            else:
-                workflow_url = resolve_latest_alpha_workflow_url()
-        elif release_version:
-            try:
-                workflow = resolve_release_workflow(version)
-                resolved_head_sha = workflow.get("headSha")
-            except Exception:
-                resolved_head_sha = None
+        vendor_src = args.vendor_src.resolve() if args.vendor_src else None
+        native_components = PACKAGE_NATIVE_COMPONENTS.get(package, [])
 
-        if release_version and resolved_head_sha:
-            print(f"should `git checkout {resolved_head_sha}`")
+        if native_components:
+            if vendor_src is None:
+                components_str = ", ".join(native_components)
+                raise RuntimeError(
+                    "Native components "
+                    f"({components_str}) required for package '{package}'. Provide --vendor-src "
+                    "pointing to a directory containing pre-installed binaries."
+                )
 
-        if not workflow_url:
-            raise RuntimeError("Unable to determine workflow URL for native binaries.")
-
-        install_native_binaries(staging_dir, workflow_url)
+            copy_native_binaries(vendor_src, staging_dir, native_components)
 
         if release_version:
             staging_dir_str = str(staging_dir)
-            print(
-                f"Staged version {version} for release in {staging_dir_str}\n\n"
-                "Verify the CLI:\n"
-                f"    node {staging_dir_str}/bin/codex.js --version\n"
-                f"    node {staging_dir_str}/bin/codex.js --help\n\n"
-            )
+            if package == "codex":
+                print(
+                    f"Staged version {version} for release in {staging_dir_str}\n\n"
+                    "Verify the CLI:\n"
+                    f"    node {staging_dir_str}/bin/codex.js --version\n"
+                    f"    node {staging_dir_str}/bin/codex.js --help\n\n"
+                )
+            elif package == "codex-responses-api-proxy":
+                print(
+                    f"Staged version {version} for release in {staging_dir_str}\n\n"
+                    "Verify the responses API proxy:\n"
+                    f"    node {staging_dir_str}/bin/codex-responses-api-proxy.js --help\n\n"
+                )
+            else:
+                print(
+                    f"Staged version {version} for release in {staging_dir_str}\n\n"
+                    "Verify the SDK contents:\n"
+                    f"    ls {staging_dir_str}/dist\n"
+                    f"    ls {staging_dir_str}/vendor\n"
+                    "    node -e \"import('./dist/index.js').then(() => console.log('ok'))\"\n\n"
+                )
         else:
             print(f"Staged package in {staging_dir}")
 
@@ -136,99 +154,120 @@ def prepare_staging_dir(staging_dir: Path | None) -> tuple[Path, bool]:
     return temp_dir, True
 
 
-def stage_sources(staging_dir: Path, version: str) -> None:
-    bin_dir = staging_dir / "bin"
-    bin_dir.mkdir(parents=True, exist_ok=True)
+def stage_sources(staging_dir: Path, version: str, package: str) -> None:
+    if package == "codex":
+        bin_dir = staging_dir / "bin"
+        bin_dir.mkdir(parents=True, exist_ok=True)
+        shutil.copy2(CODEX_CLI_ROOT / "bin" / "codex.js", bin_dir / "codex.js")
+        rg_manifest = CODEX_CLI_ROOT / "bin" / "rg"
+        if rg_manifest.exists():
+            shutil.copy2(rg_manifest, bin_dir / "rg")
 
-    shutil.copy2(CODEX_CLI_ROOT / "bin" / "codex.js", bin_dir / "codex.js")
-    rg_manifest = CODEX_CLI_ROOT / "bin" / "rg"
-    if rg_manifest.exists():
-        shutil.copy2(rg_manifest, bin_dir / "rg")
+        readme_src = REPO_ROOT / "README.md"
+        if readme_src.exists():
+            shutil.copy2(readme_src, staging_dir / "README.md")
 
-    readme_src = REPO_ROOT / "README.md"
-    if readme_src.exists():
-        shutil.copy2(readme_src, staging_dir / "README.md")
+        package_json_path = CODEX_CLI_ROOT / "package.json"
+    elif package == "codex-responses-api-proxy":
+        bin_dir = staging_dir / "bin"
+        bin_dir.mkdir(parents=True, exist_ok=True)
+        launcher_src = RESPONSES_API_PROXY_NPM_ROOT / "bin" / "codex-responses-api-proxy.js"
+        shutil.copy2(launcher_src, bin_dir / "codex-responses-api-proxy.js")
 
-    with open(CODEX_CLI_ROOT / "package.json", "r", encoding="utf-8") as fh:
+        readme_src = RESPONSES_API_PROXY_NPM_ROOT / "README.md"
+        if readme_src.exists():
+            shutil.copy2(readme_src, staging_dir / "README.md")
+
+        package_json_path = RESPONSES_API_PROXY_NPM_ROOT / "package.json"
+    elif package == "codex-sdk":
+        package_json_path = CODEX_SDK_ROOT / "package.json"
+        stage_codex_sdk_sources(staging_dir)
+    else:
+        raise RuntimeError(f"Unknown package '{package}'.")
+
+    with open(package_json_path, "r", encoding="utf-8") as fh:
         package_json = json.load(fh)
     package_json["version"] = version
 
+    if package == "codex-sdk":
+        scripts = package_json.get("scripts")
+        if isinstance(scripts, dict):
+            scripts.pop("prepare", None)
+
+        files = package_json.get("files")
+        if isinstance(files, list):
+            if "vendor" not in files:
+                files.append("vendor")
+        else:
+            package_json["files"] = ["dist", "vendor"]
+
     with open(staging_dir / "package.json", "w", encoding="utf-8") as out:
         json.dump(package_json, out, indent=2)
         out.write("\n")
 
 
-def install_native_binaries(staging_dir: Path, workflow_url: str | None) -> None:
-    cmd = ["./scripts/install_native_deps.py"]
-    if workflow_url:
-        cmd.extend(["--workflow-url", workflow_url])
-    cmd.append(str(staging_dir))
-    subprocess.check_call(cmd, cwd=CODEX_CLI_ROOT)
+def run_command(cmd: list[str], cwd: Path | None = None) -> None:
+    print("+", " ".join(cmd))
+    subprocess.run(cmd, cwd=cwd, check=True)
 
 
-def resolve_latest_alpha_workflow_url() -> str:
-    version = determine_latest_alpha_version()
-    workflow = resolve_release_workflow(version)
-    return workflow["url"]
+def stage_codex_sdk_sources(staging_dir: Path) -> None:
+    package_root = CODEX_SDK_ROOT
+
+    run_command(["pnpm", "install", "--frozen-lockfile"], cwd=package_root)
+    run_command(["pnpm", "run", "build"], cwd=package_root)
+
+    dist_src = package_root / "dist"
+    if not dist_src.exists():
+        raise RuntimeError("codex-sdk build did not produce a dist directory.")
+
+    shutil.copytree(dist_src, staging_dir / "dist")
+
+    readme_src = package_root / "README.md"
+    if readme_src.exists():
+        shutil.copy2(readme_src, staging_dir / "README.md")
+
+    license_src = REPO_ROOT / "LICENSE"
+    if license_src.exists():
+        shutil.copy2(license_src, staging_dir / "LICENSE")
 
 
-def determine_latest_alpha_version() -> str:
-    releases = list_releases()
-    best_key: tuple[int, int, int, int] | None = None
-    best_version: str | None = None
-    pattern = re.compile(r"^rust-v(\d+)\.(\d+)\.(\d+)-alpha\.(\d+)$")
-    for release in releases:
-        tag = release.get("tag_name", "")
-        match = pattern.match(tag)
-        if not match:
+def copy_native_binaries(vendor_src: Path, staging_dir: Path, components: list[str]) -> None:
+    vendor_src = vendor_src.resolve()
+    if not vendor_src.exists():
+        raise RuntimeError(f"Vendor source directory not found: {vendor_src}")
+
+    components_set = {component for component in components if component in COMPONENT_DEST_DIR}
+    if not components_set:
+        return
+
+    vendor_dest = staging_dir / "vendor"
+    if vendor_dest.exists():
+        shutil.rmtree(vendor_dest)
+    vendor_dest.mkdir(parents=True, exist_ok=True)
+
+    for target_dir in vendor_src.iterdir():
+        if not target_dir.is_dir():
             continue
-        key = tuple(int(match.group(i)) for i in range(1, 5))
-        if best_key is None or key > best_key:
-            best_key = key
-            best_version = (
-                f"{match.group(1)}.{match.group(2)}.{match.group(3)}-alpha.{match.group(4)}"
-            )
 
-    if best_version is None:
-        raise RuntimeError("No alpha releases found when resolving workflow URL.")
-    return best_version
+        dest_target_dir = vendor_dest / target_dir.name
+        dest_target_dir.mkdir(parents=True, exist_ok=True)
 
+        for component in components_set:
+            dest_dir_name = COMPONENT_DEST_DIR.get(component)
+            if dest_dir_name is None:
+                continue
 
-def list_releases() -> list[dict]:
-    stdout = subprocess.check_output(
-        ["gh", "api", f"/repos/{GITHUB_REPO}/releases?per_page=100"],
-        text=True,
-    )
-    try:
-        releases = json.loads(stdout or "[]")
-    except json.JSONDecodeError as exc:
-        raise RuntimeError("Unable to parse releases JSON.") from exc
-    if not isinstance(releases, list):
-        raise RuntimeError("Unexpected response when listing releases.")
-    return releases
+            src_component_dir = target_dir / dest_dir_name
+            if not src_component_dir.exists():
+                raise RuntimeError(
+                    f"Missing native component '{component}' in vendor source: {src_component_dir}"
+                )
 
-
-def resolve_release_workflow(version: str) -> dict:
-    stdout = subprocess.check_output(
-        [
-            "gh",
-            "run",
-            "list",
-            "--branch",
-            f"rust-v{version}",
-            "--json",
-            "workflowName,url,headSha",
-            "--workflow",
-            WORKFLOW_NAME,
-            "--jq",
-            "first(.[])",
-        ],
-        text=True,
-    )
-    workflow = json.loads(stdout or "[]")
-    if not workflow:
-        raise RuntimeError(f"Unable to find rust-release workflow for version {version}.")
-    return workflow
+            dest_component_dir = dest_target_dir / dest_dir_name
+            if dest_component_dir.exists():
+                shutil.rmtree(dest_component_dir)
+            shutil.copytree(src_component_dir, dest_component_dir)
 
 
 def run_npm_pack(staging_dir: Path, output_path: Path) -> Path:

codex-cli/scripts/install_native_deps.py

@@ -9,6 +9,7 @@ import subprocess
 import tarfile
 import tempfile
 import zipfile
+from dataclasses import dataclass
 from concurrent.futures import ThreadPoolExecutor, as_completed
 from pathlib import Path
 from typing import Iterable, Sequence
@@ -20,7 +21,7 @@ CODEX_CLI_ROOT = SCRIPT_DIR.parent
 DEFAULT_WORKFLOW_URL = "https://github.com/openai/codex/actions/runs/17952349351"  # rust-v0.40.0
 VENDOR_DIR_NAME = "vendor"
 RG_MANIFEST = CODEX_CLI_ROOT / "bin" / "rg"
-CODEX_TARGETS = (
+BINARY_TARGETS = (
     "x86_64-unknown-linux-musl",
     "aarch64-unknown-linux-musl",
     "x86_64-apple-darwin",
@@ -29,6 +30,27 @@ CODEX_TARGETS = (
     "aarch64-pc-windows-msvc",
 )
 
+
+@dataclass(frozen=True)
+class BinaryComponent:
+    artifact_prefix: str  # matches the artifact filename prefix (e.g. codex-<target>.zst)
+    dest_dir: str  # directory under vendor/<target>/ where the binary is installed
+    binary_basename: str  # executable name inside dest_dir (before optional .exe)
+
+
+BINARY_COMPONENTS = {
+    "codex": BinaryComponent(
+        artifact_prefix="codex",
+        dest_dir="codex",
+        binary_basename="codex",
+    ),
+    "codex-responses-api-proxy": BinaryComponent(
+        artifact_prefix="codex-responses-api-proxy",
+        dest_dir="codex-responses-api-proxy",
+        binary_basename="codex-responses-api-proxy",
+    ),
+}
+
 RG_TARGET_PLATFORM_PAIRS: list[tuple[str, str]] = [
     ("x86_64-unknown-linux-musl", "linux-x86_64"),
     ("aarch64-unknown-linux-musl", "linux-aarch64"),
@@ -50,6 +72,16 @@ def parse_args() -> argparse.Namespace:
             "known good run when omitted."
         ),
     )
+    parser.add_argument(
+        "--component",
+        dest="components",
+        action="append",
+        choices=tuple(list(BINARY_COMPONENTS) + ["rg"]),
+        help=(
+            "Limit installation to the specified components."
+            " May be repeated. Defaults to 'codex' and 'rg'."
+        ),
+    )
     parser.add_argument(
         "root",
         nargs="?",
@@ -69,18 +101,28 @@ def main() -> int:
     vendor_dir = codex_cli_root / VENDOR_DIR_NAME
     vendor_dir.mkdir(parents=True, exist_ok=True)
 
+    components = args.components or ["codex", "rg"]
+
     workflow_url = (args.workflow_url or DEFAULT_WORKFLOW_URL).strip()
     if not workflow_url:
         workflow_url = DEFAULT_WORKFLOW_URL
 
     workflow_id = workflow_url.rstrip("/").split("/")[-1]
+    print(f"Downloading native artifacts from workflow {workflow_id}...")
 
     with tempfile.TemporaryDirectory(prefix="codex-native-artifacts-") as artifacts_dir_str:
         artifacts_dir = Path(artifacts_dir_str)
         _download_artifacts(workflow_id, artifacts_dir)
-        install_codex_binaries(artifacts_dir, vendor_dir, CODEX_TARGETS)
+        install_binary_components(
+            artifacts_dir,
+            vendor_dir,
+            BINARY_TARGETS,
+            [name for name in components if name in BINARY_COMPONENTS],
+        )
 
-    fetch_rg(vendor_dir, DEFAULT_RG_TARGETS, manifest_path=RG_MANIFEST)
+    if "rg" in components:
+        print("Fetching ripgrep binaries...")
+        fetch_rg(vendor_dir, DEFAULT_RG_TARGETS, manifest_path=RG_MANIFEST)
 
     print(f"Installed native dependencies into {vendor_dir}")
     return 0
@@ -124,6 +166,8 @@ def fetch_rg(
     results: dict[str, Path] = {}
     max_workers = min(len(task_configs), max(1, (os.cpu_count() or 1)))
 
+    print("Installing ripgrep binaries for targets: " + ", ".join(targets))
+
     with ThreadPoolExecutor(max_workers=max_workers) as executor:
         future_map = {
             executor.submit(
@@ -140,6 +184,7 @@ def fetch_rg(
         for future in as_completed(future_map):
             target = future_map[future]
             results[target] = future.result()
+            print(f"  installed ripgrep for {target}")
 
     return [results[target] for target in targets]
 
@@ -158,40 +203,60 @@ def _download_artifacts(workflow_id: str, dest_dir: Path) -> None:
     subprocess.check_call(cmd)
 
 
-def install_codex_binaries(
-    artifacts_dir: Path, vendor_dir: Path, targets: Iterable[str]
-) -> list[Path]:
+def install_binary_components(
+    artifacts_dir: Path,
+    vendor_dir: Path,
+    targets: Iterable[str],
+    component_names: Sequence[str],
+) -> None:
+    selected_components = [BINARY_COMPONENTS[name] for name in component_names if name in BINARY_COMPONENTS]
+    if not selected_components:
+        return
+
     targets = list(targets)
     if not targets:
-        return []
+        return
 
-    results: dict[str, Path] = {}
-    max_workers = min(len(targets), max(1, (os.cpu_count() or 1)))
-
-    with ThreadPoolExecutor(max_workers=max_workers) as executor:
-        future_map = {
-            executor.submit(_install_single_codex_binary, artifacts_dir, vendor_dir, target): target
-            for target in targets
-        }
-
-        for future in as_completed(future_map):
-            target = future_map[future]
-            results[target] = future.result()
-
-    return [results[target] for target in targets]
+    for component in selected_components:
+        print(
+            f"Installing {component.binary_basename} binaries for targets: "
+            + ", ".join(targets)
+        )
+        max_workers = min(len(targets), max(1, (os.cpu_count() or 1)))
+        with ThreadPoolExecutor(max_workers=max_workers) as executor:
+            futures = {
+                executor.submit(
+                    _install_single_binary,
+                    artifacts_dir,
+                    vendor_dir,
+                    target,
+                    component,
+                ): target
+                for target in targets
+            }
+            for future in as_completed(futures):
+                installed_path = future.result()
+                print(f"  installed {installed_path}")
 
 
-def _install_single_codex_binary(artifacts_dir: Path, vendor_dir: Path, target: str) -> Path:
+def _install_single_binary(
+    artifacts_dir: Path,
+    vendor_dir: Path,
+    target: str,
+    component: BinaryComponent,
+) -> Path:
     artifact_subdir = artifacts_dir / target
-    archive_name = _archive_name_for_target(target)
+    archive_name = _archive_name_for_target(component.artifact_prefix, target)
     archive_path = artifact_subdir / archive_name
     if not archive_path.exists():
         raise FileNotFoundError(f"Expected artifact not found: {archive_path}")
 
-    dest_dir = vendor_dir / target / "codex"
+    dest_dir = vendor_dir / target / component.dest_dir
     dest_dir.mkdir(parents=True, exist_ok=True)
 
-    binary_name = "codex.exe" if "windows" in target else "codex"
+    binary_name = (
+        f"{component.binary_basename}.exe" if "windows" in target else component.binary_basename
+    )
     dest = dest_dir / binary_name
     dest.unlink(missing_ok=True)
     extract_archive(archive_path, "zst", None, dest)
@@ -200,10 +265,10 @@ def _install_single_codex_binary(artifacts_dir: Path, vendor_dir: Path, target:
     return dest
 
 
-def _archive_name_for_target(target: str) -> str:
+def _archive_name_for_target(artifact_prefix: str, target: str) -> str:
     if "windows" in target:
-        return f"codex-{target}.exe.zst"
-    return f"codex-{target}.zst"
+        return f"{artifact_prefix}-{target}.exe.zst"
+    return f"{artifact_prefix}-{target}.zst"
 
 
 def _fetch_single_rg(

codex-rs/.cargo/config.toml

@@ -0,0 +1,5 @@
+[target.'cfg(all(windows, target_env = "msvc"))']
+rustflags = ["-C", "link-arg=/STACK:8388608"]
+
+[target.'cfg(all(windows, target_env = "gnu"))']
+rustflags = ["-C", "link-arg=-Wl,--stack,8388608"]

codex-rs/.gitignore

@@ -1,4 +1,5 @@
 /target/
+/target-*/
 
 # Recommended value of CARGO_TARGET_DIR when using Docker as explained in .devcontainer/README.md.
 /target-amd64/

codex-rs/Cargo.toml

@@ -1,27 +1,44 @@
 [workspace]
 members = [
+    "backend-client",
     "ansi-escape",
+    "async-utils",
+    "app-server",
+    "app-server-protocol",
     "apply-patch",
     "arg0",
+    "feedback",
+    "codex-backend-openapi-models",
+    "cloud-tasks",
+    "cloud-tasks-client",
     "cli",
     "common",
     "core",
     "exec",
     "execpolicy",
+    "keyring-store",
     "file-search",
-    "git-tooling",
     "linux-sandbox",
     "login",
-    "mcp-client",
     "mcp-server",
     "mcp-types",
     "ollama",
+    "process-hardening",
     "protocol",
     "protocol-ts",
     "rmcp-client",
     "responses-api-proxy",
+    "stdio-to-uds",
+    "otel",
     "tui",
+    "utils/git",
+    "utils/cache",
+    "utils/image",
+    "utils/json-to-toml",
+    "utils/pty",
     "utils/readiness",
+    "utils/string",
+    "utils/tokenizer",
 ]
 resolver = "2"
 
@@ -35,26 +52,42 @@ edition = "2024"
 
 [workspace.dependencies]
 # Internal
+app_test_support = { path = "app-server/tests/common" }
 codex-ansi-escape = { path = "ansi-escape" }
+codex-app-server = { path = "app-server" }
+codex-app-server-protocol = { path = "app-server-protocol" }
 codex-apply-patch = { path = "apply-patch" }
 codex-arg0 = { path = "arg0" }
+codex-async-utils = { path = "async-utils" }
+codex-backend-client = { path = "backend-client" }
 codex-chatgpt = { path = "chatgpt" }
 codex-common = { path = "common" }
 codex-core = { path = "core" }
 codex-exec = { path = "exec" }
+codex-feedback = { path = "feedback" }
 codex-file-search = { path = "file-search" }
-codex-git-tooling = { path = "git-tooling" }
+codex-git = { path = "utils/git" }
+codex-keyring-store = { path = "keyring-store" }
 codex-linux-sandbox = { path = "linux-sandbox" }
 codex-login = { path = "login" }
-codex-mcp-client = { path = "mcp-client" }
 codex-mcp-server = { path = "mcp-server" }
 codex-ollama = { path = "ollama" }
+codex-otel = { path = "otel" }
+codex-process-hardening = { path = "process-hardening" }
 codex-protocol = { path = "protocol" }
-codex-rmcp-client = { path = "rmcp-client" }
 codex-protocol-ts = { path = "protocol-ts" }
 codex-responses-api-proxy = { path = "responses-api-proxy" }
+codex-rmcp-client = { path = "rmcp-client" }
+codex-stdio-to-uds = { path = "stdio-to-uds" }
 codex-tui = { path = "tui" }
+codex-utils-cache = { path = "utils/cache" }
+codex-utils-image = { path = "utils/image" }
+codex-utils-json-to-toml = { path = "utils/json-to-toml" }
+codex-utils-pty = { path = "utils/pty" }
 codex-utils-readiness = { path = "utils/readiness" }
+codex-utils-string = { path = "utils/string" }
+codex-utils-tokenizer = { path = "utils/tokenizer" }
+codex-windows-sandbox = { path = "windows-sandbox" }
 core_test_support = { path = "core/tests/common" }
 mcp-types = { path = "mcp-types" }
 mcp_test_support = { path = "mcp-server/tests/common" }
@@ -66,9 +99,11 @@ anyhow = "1"
 arboard = "3"
 askama = "0.12"
 assert_cmd = "2"
+assert_matches = "1.5.0"
 async-channel = "2.3.1"
 async-stream = "0.3.6"
 async-trait = "0.1.89"
+axum = { version = "0.8", default-features = false }
 base64 = "0.22.1"
 bytes = "1.10.1"
 chrono = "0.4.42"
@@ -81,31 +116,42 @@ derive_more = "2"
 diffy = "0.4.2"
 dirs = "6"
 dotenvy = "0.15.7"
+dunce = "1.0.4"
 env-flags = "0.1.1"
 env_logger = "0.11.5"
-eventsource-stream = "0.2.3"
 escargot = "0.5"
-futures = "0.3"
-icu_decimal = "2.0.0"
-icu_locale_core = "2.0.0"
+eventsource-stream = "0.2.3"
+futures = { version = "0.3", default-features = false }
+http = "1.3.1"
+icu_decimal = "2.1"
+icu_provider = { version = "2.1", features = ["sync"] }
+icu_locale_core = "2.1"
 ignore = "0.4.23"
 image = { version = "^0.25.8", default-features = false }
-indexmap = "2.6.0"
+indexmap = "2.12.0"
 insta = "1.43.2"
 itertools = "0.14.0"
+keyring = "3.6"
 landlock = "0.4.1"
 lazy_static = "1"
 libc = "0.2.175"
 log = "0.4"
+lru = "0.12.5"
 maplit = "1.0.2"
 mime_guess = "2.0.5"
 multimap = "0.10.0"
+notify = "8.2.0"
 nucleo-matcher = "0.3.1"
 openssl-sys = "*"
+opentelemetry = "0.30.0"
+opentelemetry-appender-tracing = "0.30.0"
+opentelemetry-otlp = "0.30.0"
+opentelemetry-semantic-conventions = "0.30.0"
+opentelemetry_sdk = "0.30.0"
 os_info = "3.12.0"
 owo-colors = "4.2.0"
+paste = "1.0.15"
 path-absolutize = "3.1.1"
-path-clean = "1.0.1"
 pathdiff = "0.2"
 portable-pty = "0.9.0"
 predicates = "3"
@@ -113,13 +159,17 @@ pretty_assertions = "1.4.1"
 pulldown-cmark = "0.10"
 rand = "0.9"
 ratatui = "0.29.0"
+ratatui-macros = "0.6.0"
 regex-lite = "0.1.7"
 reqwest = "0.12"
+rmcp = { version = "0.8.3", default-features = false }
 schemars = "0.8.22"
 seccompiler = "0.5.0"
+sentry = "0.34.0"
 serde = "1"
 serde_json = "1"
 serde_with = "3.14"
+serial_test = "3.2.0"
 sha1 = "0.10.6"
 sha2 = "0.10"
 shlex = "1.3.0"
@@ -130,8 +180,9 @@ strum_macros = "0.27.2"
 supports-color = "3.0.2"
 sys-locale = "0.3.2"
 tempfile = "3.23.0"
+test-log = "0.2.18"
 textwrap = "0.16.2"
-thiserror = "2.0.16"
+thiserror = "2.0.17"
 time = "0.3"
 tiny_http = "0.12"
 tokio = "1"
@@ -140,12 +191,16 @@ tokio-test = "0.4"
 tokio-util = "0.7.16"
 toml = "0.9.5"
 toml_edit = "0.23.4"
+tonic = "0.13.1"
 tracing = "0.1.41"
 tracing-appender = "0.2.3"
 tracing-subscriber = "0.3.20"
-tree-sitter = "0.25.9"
-tree-sitter-bash = "0.25.0"
+tracing-test = "0.2.5"
+tree-sitter = "0.25.10"
+tree-sitter-bash = "0.25"
+tree-sitter-highlight = "0.25.10"
 ts-rs = "11"
+uds_windows = "1.1.0"
 unicode-segmentation = "1.12.0"
 unicode-width = "0.2"
 url = "2"
@@ -156,6 +211,7 @@ walkdir = "2.5.0"
 webbrowser = "1.0"
 which = "6"
 wildmatch = "2.5.0"
+
 wiremock = "0.6"
 zeroize = "1.8.1"
 
@@ -200,7 +256,7 @@ unwrap_used = "deny"
 # cargo-shear cannot see the platform-specific openssl-sys usage, so we
 # silence the false positive here instead of deleting a real dependency.
 [workspace.metadata.cargo-shear]
-ignored = ["openssl-sys", "codex-utils-readiness"]
+ignored = ["icu_provider", "openssl-sys", "codex-utils-readiness", "codex-utils-tokenizer"]
 
 [profile.release]
 lto = "fat"
@@ -211,6 +267,15 @@ strip = "symbols"
 # See https://github.com/openai/codex/issues/1411 for details.
 codegen-units = 1
 
+[profile.ci-test]
+debug = 1         # Reduce debug symbol size
+inherits = "test"
+opt-level = 0
+
 [patch.crates-io]
+# Uncomment to debug local changes.
 # ratatui = { path = "../../ratatui" }
 ratatui = { git = "https://github.com/nornagon/ratatui", branch = "nornagon-v0.29.0-patch" }
+
+# Uncomment to debug local changes.
+# rmcp = { path = "../../rust-sdk/crates/rmcp" }

codex-rs/README.md

@@ -4,18 +4,23 @@ We provide Codex CLI as a standalone, native executable to ensure a zero-depende
 
 ## Installing Codex
 
-Today, the easiest way to install Codex is via `npm`, though we plan to publish Codex to other package managers soon.
+Today, the easiest way to install Codex is via `npm`:
 
 ```shell
-npm i -g @openai/codex@native
+npm i -g @openai/codex
 codex
 ```
 
-You can also download a platform-specific release directly from our [GitHub Releases](https://github.com/openai/codex/releases).
+You can also install via Homebrew (`brew install --cask codex`) or download a platform-specific release directly from our [GitHub Releases](https://github.com/openai/codex/releases).
+
+## Documentation quickstart
+
+- First run with Codex? Follow the walkthrough in [`docs/getting-started.md`](../docs/getting-started.md) for prompts, keyboard shortcuts, and session management.
+- Already shipping with Codex and want deeper control? Jump to [`docs/advanced.md`](../docs/advanced.md) and the configuration reference at [`docs/config.md`](../docs/config.md).
 
 ## What's new in the Rust CLI
 
-While we are [working to close the gap between the TypeScript and Rust implementations of Codex CLI](https://github.com/openai/codex/issues/1262), note that the Rust CLI has a number of features that the TypeScript CLI does not!
+The Rust implementation is now the maintained Codex CLI and serves as the default experience. It includes a number of features that the legacy TypeScript CLI never supported.
 
 ### Config
 
@@ -23,14 +28,22 @@ Codex supports a rich set of configuration options. Note that the Rust CLI uses
 
 ### Model Context Protocol Support
 
-Codex CLI functions as an MCP client that can connect to MCP servers on startup. See the [`mcp_servers`](../docs/config.md#mcp_servers) section in the configuration documentation for details.
+#### MCP client
 
-It is still experimental, but you can also launch Codex as an MCP _server_ by running `codex mcp`. Use the [`@modelcontextprotocol/inspector`](https://github.com/modelcontextprotocol/inspector) to try it out:
+Codex CLI functions as an MCP client that allows the Codex CLI and IDE extension to connect to MCP servers on startup. See the [`configuration documentation`](../docs/config.md#mcp_servers) for details.
+
+#### MCP server (experimental)
+
+Codex can be launched as an MCP _server_ by running `codex mcp-server`. This allows _other_ MCP clients to use Codex as a tool for another agent.
+
+Use the [`@modelcontextprotocol/inspector`](https://github.com/modelcontextprotocol/inspector) to try it out:
 
 ```shell
-npx @modelcontextprotocol/inspector codex mcp
+npx @modelcontextprotocol/inspector codex mcp-server
 ```
 
+Use `codex mcp` to add/list/get/remove MCP server launchers defined in `config.toml`, and `codex mcp-server` to run the MCP server directly.
+
 ### Notifications
 
 You can enable notifications by configuring a script that is run whenever the agent finishes a turn. The [notify documentation](../docs/config.md#notify) includes a detailed example that explains how to get desktop notifications via [terminal-notifier](https://github.com/julienXX/terminal-notifier) on macOS.
@@ -39,39 +52,22 @@ You can enable notifications by configuring a script that is run whenever the ag
 
 To run Codex non-interactively, run `codex exec PROMPT` (you can also pass the prompt via `stdin`) and Codex will work on your task until it decides that it is done and exits. Output is printed to the terminal directly. You can set the `RUST_LOG` environment variable to see more about what's going on.
 
-### Use `@` for file search
-
-Typing `@` triggers a fuzzy-filename search over the workspace root. Use up/down to select among the results and Tab or Enter to replace the `@` with the selected path. You can use Esc to cancel the search.
-
-### Esc–Esc to edit a previous message
-
-When the chat composer is empty, press Esc to prime “backtrack” mode. Press Esc again to open a transcript preview highlighting the last user message; press Esc repeatedly to step to older user messages. Press Enter to confirm and Codex will fork the conversation from that point, trim the visible transcript accordingly, and pre‑fill the composer with the selected user message so you can edit and resubmit it.
-
-In the transcript preview, the footer shows an `Esc edit prev` hint while editing is active.
-
-### `--cd`/`-C` flag
-
-Sometimes it is not convenient to `cd` to the directory you want Codex to use as the "working root" before running Codex. Fortunately, `codex` supports a `--cd` option so you can specify whatever folder you want. You can confirm that Codex is honoring `--cd` by double-checking the **workdir** it reports in the TUI at the start of a new session.
-
-### Shell completions
-
-Generate shell completion scripts via:
-
-```shell
-codex completion bash
-codex completion zsh
-codex completion fish
-```
-
 ### Experimenting with the Codex Sandbox
 
 To test to see what happens when a command is run under the sandbox provided by Codex, we provide the following subcommands in Codex CLI:
 
 ```
 # macOS
-codex debug seatbelt [--full-auto] [COMMAND]...
+codex sandbox macos [--full-auto] [COMMAND]...
 
 # Linux
+codex sandbox linux [--full-auto] [COMMAND]...
+
+# Windows
+codex sandbox windows [--full-auto] [COMMAND]...
+
+# Legacy aliases
+codex debug seatbelt [--full-auto] [COMMAND]...
 codex debug landlock [--full-auto] [COMMAND]...
 ```
 

codex-rs/ansi-escape/src/lib.rs

@@ -3,11 +3,30 @@ use ansi_to_tui::IntoText;
 use ratatui::text::Line;
 use ratatui::text::Text;
 
+// Expand tabs in a best-effort way for transcript rendering.
+// Tabs can interact poorly with left-gutter prefixes in our TUI and CLI
+// transcript views (e.g., `nl` separates line numbers from content with a tab).
+// Replacing tabs with spaces avoids odd visual artifacts without changing
+// semantics for our use cases.
+fn expand_tabs(s: &str) -> std::borrow::Cow<'_, str> {
+    if s.contains('\t') {
+        // Keep it simple: replace each tab with 4 spaces.
+        // We do not try to align to tab stops since most usages (like `nl`)
+        // look acceptable with a fixed substitution and this avoids stateful math
+        // across spans.
+        std::borrow::Cow::Owned(s.replace('\t', "    "))
+    } else {
+        std::borrow::Cow::Borrowed(s)
+    }
+}
+
 /// This function should be used when the contents of `s` are expected to match
 /// a single line. If multiple lines are found, a warning is logged and only the
 /// first line is returned.
 pub fn ansi_escape_line(s: &str) -> Line<'static> {
-    let text = ansi_escape(s);
+    // Normalize tabs to spaces to avoid odd gutter collisions in transcript mode.
+    let s = expand_tabs(s);
+    let text = ansi_escape(&s);
     match text.lines.as_slice() {
         [] => "".into(),
         [only] => only.clone(),

codex-rs/app-server-protocol/Cargo.toml

@@ -0,0 +1,27 @@
+[package]
+edition = "2024"
+name = "codex-app-server-protocol"
+version = { workspace = true }
+
+[lib]
+name = "codex_app_server_protocol"
+path = "src/lib.rs"
+
+[lints]
+workspace = true
+
+[dependencies]
+anyhow = { workspace = true }
+clap = { workspace = true, features = ["derive"] }
+codex-protocol = { workspace = true }
+paste = { workspace = true }
+schemars = { workspace = true }
+serde = { workspace = true, features = ["derive"] }
+serde_json = { workspace = true }
+strum_macros = { workspace = true }
+ts-rs = { workspace = true }
+uuid = { workspace = true, features = ["serde", "v7"] }
+
+[dev-dependencies]
+anyhow = { workspace = true }
+pretty_assertions = { workspace = true }

codex-rs/app-server-protocol/src/bin/export.rs

@@ -0,0 +1,22 @@
+use anyhow::Result;
+use clap::Parser;
+use std::path::PathBuf;
+
+#[derive(Parser, Debug)]
+#[command(
+    about = "Generate TypeScript bindings and JSON Schemas for the Codex app-server protocol"
+)]
+struct Args {
+    /// Output directory where generated files will be written
+    #[arg(short = 'o', long = "out", value_name = "DIR")]
+    out_dir: PathBuf,
+
+    /// Optional Prettier executable path to format generated TypeScript files
+    #[arg(short = 'p', long = "prettier", value_name = "PRETTIER_BIN")]
+    prettier: Option<PathBuf>,
+}
+
+fn main() -> Result<()> {
+    let args = Args::parse();
+    codex_app_server_protocol::generate_types(&args.out_dir, args.prettier.as_deref())
+}

codex-rs/app-server-protocol/src/export.rs

@@ -0,0 +1,737 @@
+use crate::ClientNotification;
+use crate::ClientRequest;
+use crate::ServerNotification;
+use crate::ServerRequest;
+use crate::export_client_response_schemas;
+use crate::export_client_responses;
+use crate::export_server_response_schemas;
+use crate::export_server_responses;
+use anyhow::Context;
+use anyhow::Result;
+use anyhow::anyhow;
+use schemars::JsonSchema;
+use schemars::schema::RootSchema;
+use schemars::schema_for;
+use serde::Serialize;
+use serde_json::Map;
+use serde_json::Value;
+use std::collections::BTreeMap;
+use std::collections::HashSet;
+use std::ffi::OsStr;
+use std::fs;
+use std::io::Read;
+use std::io::Write;
+use std::path::Path;
+use std::path::PathBuf;
+use std::process::Command;
+use ts_rs::TS;
+
+const HEADER: &str = "// GENERATED CODE! DO NOT MODIFY BY HAND!\n\n";
+
+macro_rules! for_each_schema_type {
+    ($macro:ident) => {
+        $macro!(crate::RequestId);
+        $macro!(crate::JSONRPCMessage);
+        $macro!(crate::JSONRPCRequest);
+        $macro!(crate::JSONRPCNotification);
+        $macro!(crate::JSONRPCResponse);
+        $macro!(crate::JSONRPCError);
+        $macro!(crate::JSONRPCErrorError);
+        $macro!(crate::AddConversationListenerParams);
+        $macro!(crate::AddConversationSubscriptionResponse);
+        $macro!(crate::ApplyPatchApprovalParams);
+        $macro!(crate::ApplyPatchApprovalResponse);
+        $macro!(crate::ArchiveConversationParams);
+        $macro!(crate::ArchiveConversationResponse);
+        $macro!(crate::AuthMode);
+        $macro!(crate::AuthStatusChangeNotification);
+        $macro!(crate::CancelLoginChatGptParams);
+        $macro!(crate::CancelLoginChatGptResponse);
+        $macro!(crate::ClientInfo);
+        $macro!(crate::ClientNotification);
+        $macro!(crate::ClientRequest);
+        $macro!(crate::ConversationSummary);
+        $macro!(crate::ExecCommandApprovalParams);
+        $macro!(crate::ExecCommandApprovalResponse);
+        $macro!(crate::ExecOneOffCommandParams);
+        $macro!(crate::ExecOneOffCommandResponse);
+        $macro!(crate::FuzzyFileSearchParams);
+        $macro!(crate::FuzzyFileSearchResponse);
+        $macro!(crate::FuzzyFileSearchResult);
+        $macro!(crate::GetAuthStatusParams);
+        $macro!(crate::GetAuthStatusResponse);
+        $macro!(crate::GetUserAgentResponse);
+        $macro!(crate::GetUserSavedConfigResponse);
+        $macro!(crate::GitDiffToRemoteParams);
+        $macro!(crate::GitDiffToRemoteResponse);
+        $macro!(crate::GitSha);
+        $macro!(crate::InitializeParams);
+        $macro!(crate::InitializeResponse);
+        $macro!(crate::InputItem);
+        $macro!(crate::InterruptConversationParams);
+        $macro!(crate::InterruptConversationResponse);
+        $macro!(crate::ListConversationsParams);
+        $macro!(crate::ListConversationsResponse);
+        $macro!(crate::LoginApiKeyParams);
+        $macro!(crate::LoginApiKeyResponse);
+        $macro!(crate::LoginChatGptCompleteNotification);
+        $macro!(crate::LoginChatGptResponse);
+        $macro!(crate::LogoutChatGptParams);
+        $macro!(crate::LogoutChatGptResponse);
+        $macro!(crate::NewConversationParams);
+        $macro!(crate::NewConversationResponse);
+        $macro!(crate::Profile);
+        $macro!(crate::RemoveConversationListenerParams);
+        $macro!(crate::RemoveConversationSubscriptionResponse);
+        $macro!(crate::ResumeConversationParams);
+        $macro!(crate::ResumeConversationResponse);
+        $macro!(crate::SandboxSettings);
+        $macro!(crate::SendUserMessageParams);
+        $macro!(crate::SendUserMessageResponse);
+        $macro!(crate::SendUserTurnParams);
+        $macro!(crate::SendUserTurnResponse);
+        $macro!(crate::ServerNotification);
+        $macro!(crate::ServerRequest);
+        $macro!(crate::SessionConfiguredNotification);
+        $macro!(crate::SetDefaultModelParams);
+        $macro!(crate::SetDefaultModelResponse);
+        $macro!(crate::Tools);
+        $macro!(crate::UserInfoResponse);
+        $macro!(crate::UserSavedConfig);
+        $macro!(codex_protocol::protocol::EventMsg);
+        $macro!(codex_protocol::protocol::FileChange);
+        $macro!(codex_protocol::parse_command::ParsedCommand);
+        $macro!(codex_protocol::protocol::SandboxPolicy);
+    };
+}
+
+pub fn generate_types(out_dir: &Path, prettier: Option<&Path>) -> Result<()> {
+    generate_ts(out_dir, prettier)?;
+    generate_json(out_dir)?;
+    Ok(())
+}
+
+pub fn generate_ts(out_dir: &Path, prettier: Option<&Path>) -> Result<()> {
+    ensure_dir(out_dir)?;
+
+    ClientRequest::export_all_to(out_dir)?;
+    export_client_responses(out_dir)?;
+    ClientNotification::export_all_to(out_dir)?;
+
+    ServerRequest::export_all_to(out_dir)?;
+    export_server_responses(out_dir)?;
+    ServerNotification::export_all_to(out_dir)?;
+
+    generate_index_ts(out_dir)?;
+
+    let ts_files = ts_files_in(out_dir)?;
+    for file in &ts_files {
+        prepend_header_if_missing(file)?;
+    }
+
+    if let Some(prettier_bin) = prettier
+        && !ts_files.is_empty()
+    {
+        let status = Command::new(prettier_bin)
+            .arg("--write")
+            .args(ts_files.iter().map(|p| p.as_os_str()))
+            .status()
+            .with_context(|| format!("Failed to invoke Prettier at {}", prettier_bin.display()))?;
+        if !status.success() {
+            return Err(anyhow!("Prettier failed with status {status}"));
+        }
+    }
+
+    Ok(())
+}
+
+pub fn generate_json(out_dir: &Path) -> Result<()> {
+    ensure_dir(out_dir)?;
+    let mut bundle: BTreeMap<String, RootSchema> = BTreeMap::new();
+
+    macro_rules! add_schema {
+        ($ty:path) => {{
+            let name = type_basename(stringify!($ty));
+            let schema = write_json_schema_with_return::<$ty>(out_dir, &name)?;
+            bundle.insert(name, schema);
+        }};
+    }
+
+    for_each_schema_type!(add_schema);
+
+    export_client_response_schemas(out_dir)?;
+    export_server_response_schemas(out_dir)?;
+
+    let mut definitions = Map::new();
+
+    const SPECIAL_DEFINITIONS: &[&str] = &[
+        "ClientNotification",
+        "ClientRequest",
+        "EventMsg",
+        "FileChange",
+        "InputItem",
+        "ParsedCommand",
+        "SandboxPolicy",
+        "ServerNotification",
+        "ServerRequest",
+    ];
+
+    for (name, schema) in bundle {
+        let mut schema_value = serde_json::to_value(schema)?;
+        annotate_schema(&mut schema_value, Some(name.as_str()));
+
+        if let Value::Object(ref mut obj) = schema_value
+            && let Some(defs) = obj.remove("definitions")
+            && let Value::Object(defs_obj) = defs
+        {
+            for (def_name, mut def_schema) in defs_obj {
+                if !SPECIAL_DEFINITIONS.contains(&def_name.as_str()) {
+                    annotate_schema(&mut def_schema, Some(def_name.as_str()));
+                    definitions.insert(def_name, def_schema);
+                }
+            }
+        }
+        definitions.insert(name, schema_value);
+    }
+
+    let mut root = Map::new();
+    root.insert(
+        "$schema".to_string(),
+        Value::String("http://json-schema.org/draft-07/schema#".into()),
+    );
+    root.insert(
+        "title".to_string(),
+        Value::String("CodexAppServerProtocol".into()),
+    );
+    root.insert("type".to_string(), Value::String("object".into()));
+    root.insert("definitions".to_string(), Value::Object(definitions));
+
+    write_pretty_json(
+        out_dir.join("codex_app_server_protocol.schemas.json"),
+        &Value::Object(root),
+    )?;
+
+    Ok(())
+}
+
+fn write_json_schema_with_return<T>(out_dir: &Path, name: &str) -> Result<RootSchema>
+where
+    T: JsonSchema,
+{
+    let file_stem = name.trim();
+    let schema = schema_for!(T);
+    let mut schema_value = serde_json::to_value(schema)?;
+    annotate_schema(&mut schema_value, Some(file_stem));
+    write_pretty_json(out_dir.join(format!("{file_stem}.json")), &schema_value)
+        .with_context(|| format!("Failed to write JSON schema for {file_stem}"))?;
+    let annotated_schema = serde_json::from_value(schema_value)?;
+    Ok(annotated_schema)
+}
+
+pub(crate) fn write_json_schema<T>(out_dir: &Path, name: &str) -> Result<()>
+where
+    T: JsonSchema,
+{
+    write_json_schema_with_return::<T>(out_dir, name).map(|_| ())
+}
+
+fn write_pretty_json(path: PathBuf, value: &impl Serialize) -> Result<()> {
+    let json = serde_json::to_vec_pretty(value)
+        .with_context(|| format!("Failed to serialize JSON schema to {}", path.display()))?;
+    fs::write(&path, json).with_context(|| format!("Failed to write {}", path.display()))?;
+    Ok(())
+}
+fn type_basename(type_path: &str) -> String {
+    type_path
+        .rsplit_once("::")
+        .map(|(_, name)| name)
+        .unwrap_or(type_path)
+        .trim()
+        .to_string()
+}
+
+fn variant_definition_name(base: &str, variant: &Value) -> Option<String> {
+    if let Some(props) = variant.get("properties").and_then(Value::as_object) {
+        if let Some(method_literal) = literal_from_property(props, "method") {
+            let pascal = to_pascal_case(method_literal);
+            return Some(match base {
+                "ClientRequest" | "ServerRequest" => format!("{pascal}Request"),
+                "ClientNotification" | "ServerNotification" => format!("{pascal}Notification"),
+                _ => format!("{pascal}{base}"),
+            });
+        }
+
+        if let Some(type_literal) = literal_from_property(props, "type") {
+            let pascal = to_pascal_case(type_literal);
+            return Some(match base {
+                "EventMsg" => format!("{pascal}EventMsg"),
+                _ => format!("{pascal}{base}"),
+            });
+        }
+
+        if let Some(mode_literal) = literal_from_property(props, "mode") {
+            let pascal = to_pascal_case(mode_literal);
+            return Some(match base {
+                "SandboxPolicy" => format!("{pascal}SandboxPolicy"),
+                _ => format!("{pascal}{base}"),
+            });
+        }
+
+        if props.len() == 1
+            && let Some(key) = props.keys().next()
+        {
+            let pascal = to_pascal_case(key);
+            return Some(format!("{pascal}{base}"));
+        }
+    }
+
+    if let Some(required) = variant.get("required").and_then(Value::as_array)
+        && required.len() == 1
+        && let Some(key) = required[0].as_str()
+    {
+        let pascal = to_pascal_case(key);
+        return Some(format!("{pascal}{base}"));
+    }
+
+    None
+}
+
+fn literal_from_property<'a>(props: &'a Map<String, Value>, key: &str) -> Option<&'a str> {
+    props.get(key).and_then(string_literal)
+}
+
+fn string_literal(value: &Value) -> Option<&str> {
+    value.get("const").and_then(Value::as_str).or_else(|| {
+        value
+            .get("enum")
+            .and_then(Value::as_array)
+            .and_then(|arr| arr.first())
+            .and_then(Value::as_str)
+    })
+}
+
+fn annotate_schema(value: &mut Value, base: Option<&str>) {
+    match value {
+        Value::Object(map) => annotate_object(map, base),
+        Value::Array(items) => {
+            for item in items {
+                annotate_schema(item, base);
+            }
+        }
+        _ => {}
+    }
+}
+
+fn annotate_object(map: &mut Map<String, Value>, base: Option<&str>) {
+    let owner = map.get("title").and_then(Value::as_str).map(str::to_owned);
+    if let Some(owner) = owner.as_deref()
+        && let Some(Value::Object(props)) = map.get_mut("properties")
+    {
+        set_discriminator_titles(props, owner);
+    }
+
+    if let Some(Value::Array(variants)) = map.get_mut("oneOf") {
+        annotate_variant_list(variants, base);
+    }
+    if let Some(Value::Array(variants)) = map.get_mut("anyOf") {
+        annotate_variant_list(variants, base);
+    }
+
+    if let Some(Value::Object(defs)) = map.get_mut("definitions") {
+        for (name, schema) in defs.iter_mut() {
+            annotate_schema(schema, Some(name.as_str()));
+        }
+    }
+
+    if let Some(Value::Object(defs)) = map.get_mut("$defs") {
+        for (name, schema) in defs.iter_mut() {
+            annotate_schema(schema, Some(name.as_str()));
+        }
+    }
+
+    if let Some(Value::Object(props)) = map.get_mut("properties") {
+        for value in props.values_mut() {
+            annotate_schema(value, base);
+        }
+    }
+
+    if let Some(items) = map.get_mut("items") {
+        annotate_schema(items, base);
+    }
+
+    if let Some(additional) = map.get_mut("additionalProperties") {
+        annotate_schema(additional, base);
+    }
+
+    for (key, child) in map.iter_mut() {
+        match key.as_str() {
+            "oneOf"
+            | "anyOf"
+            | "definitions"
+            | "$defs"
+            | "properties"
+            | "items"
+            | "additionalProperties" => {}
+            _ => annotate_schema(child, base),
+        }
+    }
+}
+
+fn annotate_variant_list(variants: &mut [Value], base: Option<&str>) {
+    let mut seen = HashSet::new();
+
+    for variant in variants.iter() {
+        if let Some(name) = variant_title(variant) {
+            seen.insert(name.to_owned());
+        }
+    }
+
+    for variant in variants.iter_mut() {
+        let mut variant_name = variant_title(variant).map(str::to_owned);
+
+        if variant_name.is_none()
+            && let Some(base_name) = base
+            && let Some(name) = variant_definition_name(base_name, variant)
+        {
+            let mut candidate = name.clone();
+            let mut index = 2;
+            while seen.contains(&candidate) {
+                candidate = format!("{name}{index}");
+                index += 1;
+            }
+            if let Some(obj) = variant.as_object_mut() {
+                obj.insert("title".into(), Value::String(candidate.clone()));
+            }
+            seen.insert(candidate.clone());
+            variant_name = Some(candidate);
+        }
+
+        if let Some(name) = variant_name.as_deref()
+            && let Some(obj) = variant.as_object_mut()
+            && let Some(Value::Object(props)) = obj.get_mut("properties")
+        {
+            set_discriminator_titles(props, name);
+        }
+
+        annotate_schema(variant, base);
+    }
+}
+
+const DISCRIMINATOR_KEYS: &[&str] = &["type", "method", "mode", "status", "role", "reason"];
+
+fn set_discriminator_titles(props: &mut Map<String, Value>, owner: &str) {
+    for key in DISCRIMINATOR_KEYS {
+        if let Some(prop_schema) = props.get_mut(*key)
+            && string_literal(prop_schema).is_some()
+            && let Value::Object(prop_obj) = prop_schema
+        {
+            if prop_obj.contains_key("title") {
+                continue;
+            }
+            let suffix = to_pascal_case(key);
+            prop_obj.insert("title".into(), Value::String(format!("{owner}{suffix}")));
+        }
+    }
+}
+
+fn variant_title(value: &Value) -> Option<&str> {
+    value
+        .as_object()
+        .and_then(|obj| obj.get("title"))
+        .and_then(Value::as_str)
+}
+
+fn to_pascal_case(input: &str) -> String {
+    let mut result = String::new();
+    let mut capitalize_next = true;
+
+    for c in input.chars() {
+        if c == '_' || c == '-' {
+            capitalize_next = true;
+            continue;
+        }
+
+        if capitalize_next {
+            result.extend(c.to_uppercase());
+            capitalize_next = false;
+        } else {
+            result.push(c);
+        }
+    }
+
+    result
+}
+
+fn ensure_dir(dir: &Path) -> Result<()> {
+    fs::create_dir_all(dir)
+        .with_context(|| format!("Failed to create output directory {}", dir.display()))
+}
+
+fn prepend_header_if_missing(path: &Path) -> Result<()> {
+    let mut content = String::new();
+    {
+        let mut f = fs::File::open(path)
+            .with_context(|| format!("Failed to open {} for reading", path.display()))?;
+        f.read_to_string(&mut content)
+            .with_context(|| format!("Failed to read {}", path.display()))?;
+    }
+
+    if content.starts_with(HEADER) {
+        return Ok(());
+    }
+
+    let mut f = fs::File::create(path)
+        .with_context(|| format!("Failed to open {} for writing", path.display()))?;
+    f.write_all(HEADER.as_bytes())
+        .with_context(|| format!("Failed to write header to {}", path.display()))?;
+    f.write_all(content.as_bytes())
+        .with_context(|| format!("Failed to write content to {}", path.display()))?;
+    Ok(())
+}
+
+fn ts_files_in(dir: &Path) -> Result<Vec<PathBuf>> {
+    let mut files = Vec::new();
+    for entry in
+        fs::read_dir(dir).with_context(|| format!("Failed to read dir {}", dir.display()))?
+    {
+        let entry = entry?;
+        let path = entry.path();
+        if path.is_file() && path.extension() == Some(OsStr::new("ts")) {
+            files.push(path);
+        }
+    }
+    files.sort();
+    Ok(files)
+}
+
+fn generate_index_ts(out_dir: &Path) -> Result<PathBuf> {
+    let mut entries: Vec<String> = Vec::new();
+    let mut stems: Vec<String> = ts_files_in(out_dir)?
+        .into_iter()
+        .filter_map(|p| {
+            let stem = p.file_stem()?.to_string_lossy().into_owned();
+            if stem == "index" { None } else { Some(stem) }
+        })
+        .collect();
+    stems.sort();
+    stems.dedup();
+
+    for name in stems {
+        entries.push(format!("export type {{ {name} }} from \"./{name}\";\n"));
+    }
+
+    let mut content =
+        String::with_capacity(HEADER.len() + entries.iter().map(String::len).sum::<usize>());
+    content.push_str(HEADER);
+    for line in &entries {
+        content.push_str(line);
+    }
+
+    let index_path = out_dir.join("index.ts");
+    let mut f = fs::File::create(&index_path)
+        .with_context(|| format!("Failed to create {}", index_path.display()))?;
+    f.write_all(content.as_bytes())
+        .with_context(|| format!("Failed to write {}", index_path.display()))?;
+    Ok(index_path)
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use anyhow::Result;
+    use std::collections::BTreeSet;
+    use std::fs;
+    use std::path::PathBuf;
+    use uuid::Uuid;
+
+    #[test]
+    fn generated_ts_has_no_optional_nullable_fields() -> Result<()> {
+        let output_dir = std::env::temp_dir().join(format!("codex_ts_types_{}", Uuid::now_v7()));
+        fs::create_dir(&output_dir)?;
+
+        struct TempDirGuard(PathBuf);
+
+        impl Drop for TempDirGuard {
+            fn drop(&mut self) {
+                let _ = fs::remove_dir_all(&self.0);
+            }
+        }
+
+        let _guard = TempDirGuard(output_dir.clone());
+
+        generate_ts(&output_dir, None)?;
+
+        let mut undefined_offenders = Vec::new();
+        let mut optional_nullable_offenders = BTreeSet::new();
+        let mut stack = vec![output_dir];
+        while let Some(dir) = stack.pop() {
+            for entry in fs::read_dir(&dir)? {
+                let entry = entry?;
+                let path = entry.path();
+                if path.is_dir() {
+                    stack.push(path);
+                    continue;
+                }
+
+                if matches!(path.extension().and_then(|ext| ext.to_str()), Some("ts")) {
+                    let contents = fs::read_to_string(&path)?;
+                    if contents.contains("| undefined") {
+                        undefined_offenders.push(path.clone());
+                    }
+
+                    const SKIP_PREFIXES: &[&str] = &[
+                        "const ",
+                        "let ",
+                        "var ",
+                        "export const ",
+                        "export let ",
+                        "export var ",
+                    ];
+
+                    let mut search_start = 0;
+                    while let Some(idx) = contents[search_start..].find("| null") {
+                        let abs_idx = search_start + idx;
+                        // Find the property-colon for this field by scanning forward
+                        // from the start of the segment and ignoring nested braces,
+                        // brackets, and parens. This avoids colons inside nested
+                        // type literals like `{ [k in string]?: string }`.
+
+                        let line_start_idx =
+                            contents[..abs_idx].rfind('\n').map(|i| i + 1).unwrap_or(0);
+
+                        let mut segment_start_idx = line_start_idx;
+                        if let Some(rel_idx) = contents[line_start_idx..abs_idx].rfind(',') {
+                            segment_start_idx = segment_start_idx.max(line_start_idx + rel_idx + 1);
+                        }
+                        if let Some(rel_idx) = contents[line_start_idx..abs_idx].rfind('{') {
+                            segment_start_idx = segment_start_idx.max(line_start_idx + rel_idx + 1);
+                        }
+                        if let Some(rel_idx) = contents[line_start_idx..abs_idx].rfind('}') {
+                            segment_start_idx = segment_start_idx.max(line_start_idx + rel_idx + 1);
+                        }
+
+                        // Scan forward for the colon that separates the field name from its type.
+                        let mut level_brace = 0_i32;
+                        let mut level_brack = 0_i32;
+                        let mut level_paren = 0_i32;
+                        let mut in_single = false;
+                        let mut in_double = false;
+                        let mut escape = false;
+                        let mut prop_colon_idx = None;
+                        for (i, ch) in contents[segment_start_idx..abs_idx].char_indices() {
+                            let idx_abs = segment_start_idx + i;
+                            if escape {
+                                escape = false;
+                                continue;
+                            }
+                            match ch {
+                                '\\' => {
+                                    // Only treat as escape when inside a string.
+                                    if in_single || in_double {
+                                        escape = true;
+                                    }
+                                }
+                                '\'' => {
+                                    if !in_double {
+                                        in_single = !in_single;
+                                    }
+                                }
+                                '"' => {
+                                    if !in_single {
+                                        in_double = !in_double;
+                                    }
+                                }
+                                '{' if !in_single && !in_double => level_brace += 1,
+                                '}' if !in_single && !in_double => level_brace -= 1,
+                                '[' if !in_single && !in_double => level_brack += 1,
+                                ']' if !in_single && !in_double => level_brack -= 1,
+                                '(' if !in_single && !in_double => level_paren += 1,
+                                ')' if !in_single && !in_double => level_paren -= 1,
+                                ':' if !in_single
+                                    && !in_double
+                                    && level_brace == 0
+                                    && level_brack == 0
+                                    && level_paren == 0 =>
+                                {
+                                    prop_colon_idx = Some(idx_abs);
+                                    break;
+                                }
+                                _ => {}
+                            }
+                        }
+
+                        let Some(colon_idx) = prop_colon_idx else {
+                            search_start = abs_idx + 5;
+                            continue;
+                        };
+
+                        let mut field_prefix = contents[segment_start_idx..colon_idx].trim();
+                        if field_prefix.is_empty() {
+                            search_start = abs_idx + 5;
+                            continue;
+                        }
+
+                        if let Some(comment_idx) = field_prefix.rfind("*/") {
+                            field_prefix = field_prefix[comment_idx + 2..].trim_start();
+                        }
+
+                        if field_prefix.is_empty() {
+                            search_start = abs_idx + 5;
+                            continue;
+                        }
+
+                        if SKIP_PREFIXES
+                            .iter()
+                            .any(|prefix| field_prefix.starts_with(prefix))
+                        {
+                            search_start = abs_idx + 5;
+                            continue;
+                        }
+
+                        if field_prefix.contains('(') {
+                            search_start = abs_idx + 5;
+                            continue;
+                        }
+
+                        // If the last non-whitespace before ':' is '?', then this is an
+                        // optional field with a nullable type (i.e., "?: T | null"),
+                        // which we explicitly disallow.
+                        if field_prefix.chars().rev().find(|c| !c.is_whitespace()) == Some('?') {
+                            let line_number =
+                                contents[..abs_idx].chars().filter(|c| *c == '\n').count() + 1;
+                            let offending_line_end = contents[line_start_idx..]
+                                .find('\n')
+                                .map(|i| line_start_idx + i)
+                                .unwrap_or(contents.len());
+                            let offending_snippet =
+                                contents[line_start_idx..offending_line_end].trim();
+
+                            optional_nullable_offenders.insert(format!(
+                                "{}:{}: {offending_snippet}",
+                                path.display(),
+                                line_number
+                            ));
+                        }
+
+                        search_start = abs_idx + 5;
+                    }
+                }
+            }
+        }
+
+        assert!(
+            undefined_offenders.is_empty(),
+            "Generated TypeScript still includes unions with `undefined` in {undefined_offenders:?}"
+        );
+
+        // If this assertion fails, it means a field was generated as
+        // "?: T | null" — i.e., both optional (undefined) and nullable (null).
+        // We only want either "?: T" or ": T | null".
+        assert!(
+            optional_nullable_offenders.is_empty(),
+            "Generated TypeScript has optional fields with nullable types (disallowed '?: T | null'), add #[ts(optional)] to fix:\n{optional_nullable_offenders:?}"
+        );
+
+        Ok(())
+    }
+}

codex-rs/app-server-protocol/src/jsonrpc_lite.rs

@@ -0,0 +1,71 @@
+//! We do not do true JSON-RPC 2.0, as we neither send nor expect the
+//! "jsonrpc": "2.0" field.
+
+use schemars::JsonSchema;
+use serde::Deserialize;
+use serde::Serialize;
+use ts_rs::TS;
+
+pub const JSONRPC_VERSION: &str = "2.0";
+
+#[derive(Debug, Clone, PartialEq, Deserialize, Serialize, Hash, Eq, JsonSchema, TS)]
+#[serde(untagged)]
+pub enum RequestId {
+    String(String),
+    #[ts(type = "number")]
+    Integer(i64),
+}
+
+pub type Result = serde_json::Value;
+
+/// Refers to any valid JSON-RPC object that can be decoded off the wire, or encoded to be sent.
+#[derive(Debug, Clone, PartialEq, Deserialize, Serialize, JsonSchema, TS)]
+#[serde(untagged)]
+pub enum JSONRPCMessage {
+    Request(JSONRPCRequest),
+    Notification(JSONRPCNotification),
+    Response(JSONRPCResponse),
+    Error(JSONRPCError),
+}
+
+/// A request that expects a response.
+#[derive(Debug, Clone, PartialEq, Deserialize, Serialize, JsonSchema, TS)]
+pub struct JSONRPCRequest {
+    pub id: RequestId,
+    pub method: String,
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    #[ts(optional)]
+    pub params: Option<serde_json::Value>,
+}
+
+/// A notification which does not expect a response.
+#[derive(Debug, Clone, PartialEq, Deserialize, Serialize, JsonSchema, TS)]
+pub struct JSONRPCNotification {
+    pub method: String,
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    #[ts(optional)]
+    pub params: Option<serde_json::Value>,
+}
+
+/// A successful (non-error) response to a request.
+#[derive(Debug, Clone, PartialEq, Deserialize, Serialize, JsonSchema, TS)]
+pub struct JSONRPCResponse {
+    pub id: RequestId,
+    pub result: Result,
+}
+
+/// A response to a request that indicates an error occurred.
+#[derive(Debug, Clone, PartialEq, Deserialize, Serialize, JsonSchema, TS)]
+pub struct JSONRPCError {
+    pub error: JSONRPCErrorError,
+    pub id: RequestId,
+}
+
+#[derive(Debug, Clone, PartialEq, Deserialize, Serialize, JsonSchema, TS)]
+pub struct JSONRPCErrorError {
+    pub code: i64,
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    #[ts(optional)]
+    pub data: Option<serde_json::Value>,
+    pub message: String,
+}

codex-rs/app-server-protocol/src/lib.rs

@@ -0,0 +1,11 @@
+mod export;
+mod jsonrpc_lite;
+mod protocol;
+
+pub use export::generate_json;
+pub use export::generate_ts;
+pub use export::generate_types;
+pub use jsonrpc_lite::*;
+pub use protocol::common::*;
+pub use protocol::v1::*;
+pub use protocol::v2::*;

codex-rs/app-server-protocol/src/protocol/common.rs

@@ -0,0 +1,685 @@
+use std::collections::HashMap;
+use std::path::PathBuf;
+
+use crate::JSONRPCNotification;
+use crate::JSONRPCRequest;
+use crate::RequestId;
+use crate::protocol::v1;
+use crate::protocol::v2;
+use codex_protocol::ConversationId;
+use codex_protocol::parse_command::ParsedCommand;
+use codex_protocol::protocol::FileChange;
+use codex_protocol::protocol::RateLimitSnapshot;
+use codex_protocol::protocol::ReviewDecision;
+use codex_protocol::protocol::SandboxCommandAssessment;
+use paste::paste;
+use schemars::JsonSchema;
+use serde::Deserialize;
+use serde::Serialize;
+use strum_macros::Display;
+use ts_rs::TS;
+
+#[derive(Serialize, Deserialize, Clone, Debug, PartialEq, JsonSchema, TS)]
+#[ts(type = "string")]
+pub struct GitSha(pub String);
+
+impl GitSha {
+    pub fn new(sha: &str) -> Self {
+        Self(sha.to_string())
+    }
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, Copy, PartialEq, Eq, Display, JsonSchema, TS)]
+#[serde(rename_all = "lowercase")]
+pub enum AuthMode {
+    ApiKey,
+    ChatGPT,
+}
+
+/// Generates an `enum ClientRequest` where each variant is a request that the
+/// client can send to the server. Each variant has associated `params` and
+/// `response` types. Also generates a `export_client_responses()` function to
+/// export all response types to TypeScript.
+macro_rules! client_request_definitions {
+    (
+        $(
+            $(#[$variant_meta:meta])*
+            $variant:ident {
+                params: $(#[$params_meta:meta])* $params:ty,
+                response: $response:ty,
+            }
+        ),* $(,)?
+    ) => {
+        /// Request from the client to the server.
+        #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+        #[serde(tag = "method", rename_all = "camelCase")]
+        pub enum ClientRequest {
+            $(
+                $(#[$variant_meta])*
+                $variant {
+                    #[serde(rename = "id")]
+                    request_id: RequestId,
+                    $(#[$params_meta])*
+                    params: $params,
+                },
+            )*
+        }
+
+        pub fn export_client_responses(
+            out_dir: &::std::path::Path,
+        ) -> ::std::result::Result<(), ::ts_rs::ExportError> {
+            $(
+                <$response as ::ts_rs::TS>::export_all_to(out_dir)?;
+            )*
+            Ok(())
+        }
+
+        pub fn export_client_response_schemas(
+            out_dir: &::std::path::Path,
+        ) -> ::anyhow::Result<()> {
+            $(
+                crate::export::write_json_schema::<$response>(out_dir, stringify!($response))?;
+            )*
+            Ok(())
+        }
+    };
+}
+
+client_request_definitions! {
+    /// NEW APIs
+    #[serde(rename = "model/list")]
+    #[ts(rename = "model/list")]
+    ListModels {
+        params: v2::ListModelsParams,
+        response: v2::ListModelsResponse,
+    },
+
+    #[serde(rename = "account/login")]
+    #[ts(rename = "account/login")]
+    LoginAccount {
+        params: v2::LoginAccountParams,
+        response: v2::LoginAccountResponse,
+    },
+
+    #[serde(rename = "account/logout")]
+    #[ts(rename = "account/logout")]
+    LogoutAccount {
+        params: #[ts(type = "undefined")] #[serde(skip_serializing_if = "Option::is_none")] Option<()>,
+        response: v2::LogoutAccountResponse,
+    },
+
+    #[serde(rename = "account/rateLimits/read")]
+    #[ts(rename = "account/rateLimits/read")]
+    GetAccountRateLimits {
+        params: #[ts(type = "undefined")] #[serde(skip_serializing_if = "Option::is_none")] Option<()>,
+        response: v2::GetAccountRateLimitsResponse,
+    },
+
+    #[serde(rename = "feedback/upload")]
+    #[ts(rename = "feedback/upload")]
+    UploadFeedback {
+        params: v2::UploadFeedbackParams,
+        response: v2::UploadFeedbackResponse,
+    },
+
+    #[serde(rename = "account/read")]
+    #[ts(rename = "account/read")]
+    GetAccount {
+        params: #[ts(type = "undefined")] #[serde(skip_serializing_if = "Option::is_none")] Option<()>,
+        response: v2::GetAccountResponse,
+    },
+
+    /// DEPRECATED APIs below
+    Initialize {
+        params: v1::InitializeParams,
+        response: v1::InitializeResponse,
+    },
+    NewConversation {
+        params: v1::NewConversationParams,
+        response: v1::NewConversationResponse,
+    },
+    GetConversationSummary {
+        params: v1::GetConversationSummaryParams,
+        response: v1::GetConversationSummaryResponse,
+    },
+    /// List recorded Codex conversations (rollouts) with optional pagination and search.
+    ListConversations {
+        params: v1::ListConversationsParams,
+        response: v1::ListConversationsResponse,
+    },
+    /// Resume a recorded Codex conversation from a rollout file.
+    ResumeConversation {
+        params: v1::ResumeConversationParams,
+        response: v1::ResumeConversationResponse,
+    },
+    ArchiveConversation {
+        params: v1::ArchiveConversationParams,
+        response: v1::ArchiveConversationResponse,
+    },
+    SendUserMessage {
+        params: v1::SendUserMessageParams,
+        response: v1::SendUserMessageResponse,
+    },
+    SendUserTurn {
+        params: v1::SendUserTurnParams,
+        response: v1::SendUserTurnResponse,
+    },
+    InterruptConversation {
+        params: v1::InterruptConversationParams,
+        response: v1::InterruptConversationResponse,
+    },
+    AddConversationListener {
+        params: v1::AddConversationListenerParams,
+        response: v1::AddConversationSubscriptionResponse,
+    },
+    RemoveConversationListener {
+        params: v1::RemoveConversationListenerParams,
+        response: v1::RemoveConversationSubscriptionResponse,
+    },
+    GitDiffToRemote {
+        params: v1::GitDiffToRemoteParams,
+        response: v1::GitDiffToRemoteResponse,
+    },
+    LoginApiKey {
+        params: v1::LoginApiKeyParams,
+        response: v1::LoginApiKeyResponse,
+    },
+    LoginChatGpt {
+        params: #[ts(type = "undefined")] #[serde(skip_serializing_if = "Option::is_none")] Option<()>,
+        response: v1::LoginChatGptResponse,
+    },
+    CancelLoginChatGpt {
+        params: v1::CancelLoginChatGptParams,
+        response: v1::CancelLoginChatGptResponse,
+    },
+    LogoutChatGpt {
+        params: #[ts(type = "undefined")] #[serde(skip_serializing_if = "Option::is_none")] Option<()>,
+        response: v1::LogoutChatGptResponse,
+    },
+    GetAuthStatus {
+        params: v1::GetAuthStatusParams,
+        response: v1::GetAuthStatusResponse,
+    },
+    GetUserSavedConfig {
+        params: #[ts(type = "undefined")] #[serde(skip_serializing_if = "Option::is_none")] Option<()>,
+        response: v1::GetUserSavedConfigResponse,
+    },
+    SetDefaultModel {
+        params: v1::SetDefaultModelParams,
+        response: v1::SetDefaultModelResponse,
+    },
+    GetUserAgent {
+        params: #[ts(type = "undefined")] #[serde(skip_serializing_if = "Option::is_none")] Option<()>,
+        response: v1::GetUserAgentResponse,
+    },
+    UserInfo {
+        params: #[ts(type = "undefined")] #[serde(skip_serializing_if = "Option::is_none")] Option<()>,
+        response: v1::UserInfoResponse,
+    },
+    FuzzyFileSearch {
+        params: FuzzyFileSearchParams,
+        response: FuzzyFileSearchResponse,
+    },
+    /// Execute a command (argv vector) under the server's sandbox.
+    ExecOneOffCommand {
+        params: v1::ExecOneOffCommandParams,
+        response: v1::ExecOneOffCommandResponse,
+    },
+}
+
+/// Generates an `enum ServerRequest` where each variant is a request that the
+/// server can send to the client along with the corresponding params and
+/// response types. It also generates helper types used by the app/server
+/// infrastructure (payload enum, request constructor, and export helpers).
+macro_rules! server_request_definitions {
+    (
+        $(
+            $(#[$variant_meta:meta])*
+            $variant:ident
+        ),* $(,)?
+    ) => {
+        paste! {
+            /// Request initiated from the server and sent to the client.
+            #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+            #[serde(tag = "method", rename_all = "camelCase")]
+            pub enum ServerRequest {
+                $(
+                    $(#[$variant_meta])*
+                    $variant {
+                        #[serde(rename = "id")]
+                        request_id: RequestId,
+                        params: [<$variant Params>],
+                    },
+                )*
+            }
+
+            #[derive(Debug, Clone, PartialEq, JsonSchema)]
+            pub enum ServerRequestPayload {
+                $( $variant([<$variant Params>]), )*
+            }
+
+            impl ServerRequestPayload {
+                pub fn request_with_id(self, request_id: RequestId) -> ServerRequest {
+                    match self {
+                        $(Self::$variant(params) => ServerRequest::$variant { request_id, params },)*
+                    }
+                }
+            }
+        }
+
+        pub fn export_server_responses(
+            out_dir: &::std::path::Path,
+        ) -> ::std::result::Result<(), ::ts_rs::ExportError> {
+            paste! {
+                $(<[<$variant Response>] as ::ts_rs::TS>::export_all_to(out_dir)?;)*
+            }
+            Ok(())
+        }
+
+        pub fn export_server_response_schemas(
+            out_dir: &::std::path::Path,
+        ) -> ::anyhow::Result<()> {
+            paste! {
+                $(crate::export::write_json_schema::<[<$variant Response>]>(out_dir, stringify!([<$variant Response>]))?;)*
+            }
+            Ok(())
+        }
+    };
+}
+
+impl TryFrom<JSONRPCRequest> for ServerRequest {
+    type Error = serde_json::Error;
+
+    fn try_from(value: JSONRPCRequest) -> Result<Self, Self::Error> {
+        serde_json::from_value(serde_json::to_value(value)?)
+    }
+}
+
+server_request_definitions! {
+    /// Request to approve a patch.
+    ApplyPatchApproval,
+    /// Request to exec a command.
+    ExecCommandApproval,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct ApplyPatchApprovalParams {
+    pub conversation_id: ConversationId,
+    /// Use to correlate this with [codex_core::protocol::PatchApplyBeginEvent]
+    /// and [codex_core::protocol::PatchApplyEndEvent].
+    pub call_id: String,
+    pub file_changes: HashMap<PathBuf, FileChange>,
+    /// Optional explanatory reason (e.g. request for extra write access).
+    pub reason: Option<String>,
+    /// When set, the agent is asking the user to allow writes under this root
+    /// for the remainder of the session (unclear if this is honored today).
+    pub grant_root: Option<PathBuf>,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct ExecCommandApprovalParams {
+    pub conversation_id: ConversationId,
+    /// Use to correlate this with [codex_core::protocol::ExecCommandBeginEvent]
+    /// and [codex_core::protocol::ExecCommandEndEvent].
+    pub call_id: String,
+    pub command: Vec<String>,
+    pub cwd: PathBuf,
+    pub reason: Option<String>,
+    pub risk: Option<SandboxCommandAssessment>,
+    pub parsed_cmd: Vec<ParsedCommand>,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+pub struct ExecCommandApprovalResponse {
+    pub decision: ReviewDecision,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+pub struct ApplyPatchApprovalResponse {
+    pub decision: ReviewDecision,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+#[ts(rename_all = "camelCase")]
+pub struct FuzzyFileSearchParams {
+    pub query: String,
+    pub roots: Vec<String>,
+    // if provided, will cancel any previous request that used the same value
+    pub cancellation_token: Option<String>,
+}
+
+/// Superset of [`codex_file_search::FileMatch`]
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+pub struct FuzzyFileSearchResult {
+    pub root: String,
+    pub path: String,
+    pub file_name: String,
+    pub score: u32,
+    pub indices: Option<Vec<u32>>,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+pub struct FuzzyFileSearchResponse {
+    pub files: Vec<FuzzyFileSearchResult>,
+}
+
+/// Notification sent from the server to the client.
+#[derive(Serialize, Deserialize, Debug, Clone, JsonSchema, TS, Display)]
+#[serde(tag = "method", content = "params", rename_all = "camelCase")]
+#[strum(serialize_all = "camelCase")]
+pub enum ServerNotification {
+    /// NEW NOTIFICATIONS
+    #[serde(rename = "account/rateLimits/updated")]
+    #[ts(rename = "account/rateLimits/updated")]
+    #[strum(serialize = "account/rateLimits/updated")]
+    AccountRateLimitsUpdated(RateLimitSnapshot),
+
+    /// DEPRECATED NOTIFICATIONS below
+    /// Authentication status changed
+    AuthStatusChange(v1::AuthStatusChangeNotification),
+
+    /// ChatGPT login flow completed
+    LoginChatGptComplete(v1::LoginChatGptCompleteNotification),
+
+    /// The special session configured event for a new or resumed conversation.
+    SessionConfigured(v1::SessionConfiguredNotification),
+}
+
+impl ServerNotification {
+    pub fn to_params(self) -> Result<serde_json::Value, serde_json::Error> {
+        match self {
+            ServerNotification::AccountRateLimitsUpdated(params) => serde_json::to_value(params),
+            ServerNotification::AuthStatusChange(params) => serde_json::to_value(params),
+            ServerNotification::LoginChatGptComplete(params) => serde_json::to_value(params),
+            ServerNotification::SessionConfigured(params) => serde_json::to_value(params),
+        }
+    }
+}
+
+impl TryFrom<JSONRPCNotification> for ServerNotification {
+    type Error = serde_json::Error;
+
+    fn try_from(value: JSONRPCNotification) -> Result<Self, Self::Error> {
+        serde_json::from_value(serde_json::to_value(value)?)
+    }
+}
+
+/// Notification sent from the client to the server.
+#[derive(Serialize, Deserialize, Debug, Clone, JsonSchema, TS, Display)]
+#[serde(tag = "method", content = "params", rename_all = "camelCase")]
+#[strum(serialize_all = "camelCase")]
+pub enum ClientNotification {
+    Initialized,
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use anyhow::Result;
+    use codex_protocol::account::PlanType;
+    use codex_protocol::protocol::AskForApproval;
+    use pretty_assertions::assert_eq;
+    use serde_json::json;
+
+    #[test]
+    fn serialize_new_conversation() -> Result<()> {
+        let request = ClientRequest::NewConversation {
+            request_id: RequestId::Integer(42),
+            params: v1::NewConversationParams {
+                model: Some("gpt-5-codex".to_string()),
+                model_provider: None,
+                profile: None,
+                cwd: None,
+                approval_policy: Some(AskForApproval::OnRequest),
+                sandbox: None,
+                config: None,
+                base_instructions: None,
+                developer_instructions: None,
+                compact_prompt: None,
+                include_apply_patch_tool: None,
+            },
+        };
+        assert_eq!(
+            json!({
+                "method": "newConversation",
+                "id": 42,
+                "params": {
+                    "model": "gpt-5-codex",
+                    "modelProvider": null,
+                    "profile": null,
+                    "cwd": null,
+                    "approvalPolicy": "on-request",
+                    "sandbox": null,
+                    "config": null,
+                    "baseInstructions": null,
+                    "includeApplyPatchTool": null
+                }
+            }),
+            serde_json::to_value(&request)?,
+        );
+        Ok(())
+    }
+
+    #[test]
+    fn conversation_id_serializes_as_plain_string() -> Result<()> {
+        let id = ConversationId::from_string("67e55044-10b1-426f-9247-bb680e5fe0c8")?;
+
+        assert_eq!(
+            json!("67e55044-10b1-426f-9247-bb680e5fe0c8"),
+            serde_json::to_value(id)?
+        );
+        Ok(())
+    }
+
+    #[test]
+    fn conversation_id_deserializes_from_plain_string() -> Result<()> {
+        let id: ConversationId =
+            serde_json::from_value(json!("67e55044-10b1-426f-9247-bb680e5fe0c8"))?;
+
+        assert_eq!(
+            ConversationId::from_string("67e55044-10b1-426f-9247-bb680e5fe0c8")?,
+            id,
+        );
+        Ok(())
+    }
+
+    #[test]
+    fn serialize_client_notification() -> Result<()> {
+        let notification = ClientNotification::Initialized;
+        // Note there is no "params" field for this notification.
+        assert_eq!(
+            json!({
+                "method": "initialized",
+            }),
+            serde_json::to_value(&notification)?,
+        );
+        Ok(())
+    }
+
+    #[test]
+    fn serialize_server_request() -> Result<()> {
+        let conversation_id = ConversationId::from_string("67e55044-10b1-426f-9247-bb680e5fe0c8")?;
+        let params = ExecCommandApprovalParams {
+            conversation_id,
+            call_id: "call-42".to_string(),
+            command: vec!["echo".to_string(), "hello".to_string()],
+            cwd: PathBuf::from("/tmp"),
+            reason: Some("because tests".to_string()),
+            risk: None,
+            parsed_cmd: vec![ParsedCommand::Unknown {
+                cmd: "echo hello".to_string(),
+            }],
+        };
+        let request = ServerRequest::ExecCommandApproval {
+            request_id: RequestId::Integer(7),
+            params: params.clone(),
+        };
+
+        assert_eq!(
+            json!({
+                "method": "execCommandApproval",
+                "id": 7,
+                "params": {
+                    "conversationId": "67e55044-10b1-426f-9247-bb680e5fe0c8",
+                    "callId": "call-42",
+                    "command": ["echo", "hello"],
+                    "cwd": "/tmp",
+                    "reason": "because tests",
+                    "risk": null,
+                    "parsedCmd": [
+                        {
+                            "type": "unknown",
+                            "cmd": "echo hello"
+                        }
+                    ]
+                }
+            }),
+            serde_json::to_value(&request)?,
+        );
+
+        let payload = ServerRequestPayload::ExecCommandApproval(params);
+        assert_eq!(payload.request_with_id(RequestId::Integer(7)), request);
+        Ok(())
+    }
+
+    #[test]
+    fn serialize_get_account_rate_limits() -> Result<()> {
+        let request = ClientRequest::GetAccountRateLimits {
+            request_id: RequestId::Integer(1),
+            params: None,
+        };
+        assert_eq!(
+            json!({
+                "method": "account/rateLimits/read",
+                "id": 1,
+            }),
+            serde_json::to_value(&request)?,
+        );
+        Ok(())
+    }
+
+    #[test]
+    fn serialize_account_login_api_key() -> Result<()> {
+        let request = ClientRequest::LoginAccount {
+            request_id: RequestId::Integer(2),
+            params: v2::LoginAccountParams::ApiKey {
+                api_key: "secret".to_string(),
+            },
+        };
+        assert_eq!(
+            json!({
+                "method": "account/login",
+                "id": 2,
+                "params": {
+                    "type": "apiKey",
+                    "apiKey": "secret"
+                }
+            }),
+            serde_json::to_value(&request)?,
+        );
+        Ok(())
+    }
+
+    #[test]
+    fn serialize_account_login_chatgpt() -> Result<()> {
+        let request = ClientRequest::LoginAccount {
+            request_id: RequestId::Integer(3),
+            params: v2::LoginAccountParams::ChatGpt,
+        };
+        assert_eq!(
+            json!({
+                "method": "account/login",
+                "id": 3,
+                "params": {
+                    "type": "chatgpt"
+                }
+            }),
+            serde_json::to_value(&request)?,
+        );
+        Ok(())
+    }
+
+    #[test]
+    fn serialize_account_logout() -> Result<()> {
+        let request = ClientRequest::LogoutAccount {
+            request_id: RequestId::Integer(4),
+            params: None,
+        };
+        assert_eq!(
+            json!({
+                "method": "account/logout",
+                "id": 4,
+            }),
+            serde_json::to_value(&request)?,
+        );
+        Ok(())
+    }
+
+    #[test]
+    fn serialize_get_account() -> Result<()> {
+        let request = ClientRequest::GetAccount {
+            request_id: RequestId::Integer(5),
+            params: None,
+        };
+        assert_eq!(
+            json!({
+                "method": "account/read",
+                "id": 5,
+            }),
+            serde_json::to_value(&request)?,
+        );
+        Ok(())
+    }
+
+    #[test]
+    fn account_serializes_fields_in_camel_case() -> Result<()> {
+        let api_key = v2::Account::ApiKey {
+            api_key: "secret".to_string(),
+        };
+        assert_eq!(
+            json!({
+                "type": "apiKey",
+                "apiKey": "secret",
+            }),
+            serde_json::to_value(&api_key)?,
+        );
+
+        let chatgpt = v2::Account::ChatGpt {
+            email: Some("user@example.com".to_string()),
+            plan_type: PlanType::Plus,
+        };
+        assert_eq!(
+            json!({
+                "type": "chatgpt",
+                "email": "user@example.com",
+                "planType": "plus",
+            }),
+            serde_json::to_value(&chatgpt)?,
+        );
+
+        Ok(())
+    }
+
+    #[test]
+    fn serialize_list_models() -> Result<()> {
+        let request = ClientRequest::ListModels {
+            request_id: RequestId::Integer(6),
+            params: v2::ListModelsParams::default(),
+        };
+        assert_eq!(
+            json!({
+                "method": "model/list",
+                "id": 6,
+                "params": {
+                    "pageSize": null,
+                    "cursor": null
+                }
+            }),
+            serde_json::to_value(&request)?,
+        );
+        Ok(())
+    }
+}

codex-rs/app-server-protocol/src/protocol/mod.rs

@@ -0,0 +1,6 @@
+// Module declarations for the app-server protocol namespace.
+// Exposes protocol pieces used by `lib.rs` via `pub use protocol::common::*;`.
+
+pub mod common;
+pub mod v1;
+pub mod v2;

codex-rs/app-server-protocol/src/protocol/v1.rs

@@ -0,0 +1,405 @@
+use std::collections::HashMap;
+use std::path::PathBuf;
+
+use codex_protocol::ConversationId;
+use codex_protocol::config_types::ForcedLoginMethod;
+use codex_protocol::config_types::ReasoningEffort;
+use codex_protocol::config_types::ReasoningSummary;
+use codex_protocol::config_types::SandboxMode;
+use codex_protocol::config_types::Verbosity;
+use codex_protocol::models::ResponseItem;
+use codex_protocol::protocol::AskForApproval;
+use codex_protocol::protocol::EventMsg;
+use codex_protocol::protocol::SandboxPolicy;
+use codex_protocol::protocol::TurnAbortReason;
+use schemars::JsonSchema;
+use serde::Deserialize;
+use serde::Serialize;
+use ts_rs::TS;
+use uuid::Uuid;
+
+// Reuse shared types defined in `common.rs`.
+use crate::protocol::common::AuthMode;
+use crate::protocol::common::GitSha;
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Default, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct InitializeParams {
+    pub client_info: ClientInfo,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Default, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct ClientInfo {
+    pub name: String,
+    pub title: Option<String>,
+    pub version: String,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct InitializeResponse {
+    pub user_agent: String,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Default, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct NewConversationParams {
+    pub model: Option<String>,
+    pub model_provider: Option<String>,
+    pub profile: Option<String>,
+    pub cwd: Option<String>,
+    pub approval_policy: Option<AskForApproval>,
+    pub sandbox: Option<SandboxMode>,
+    pub config: Option<HashMap<String, serde_json::Value>>,
+    pub base_instructions: Option<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub developer_instructions: Option<String>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub compact_prompt: Option<String>,
+    pub include_apply_patch_tool: Option<bool>,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct NewConversationResponse {
+    pub conversation_id: ConversationId,
+    pub model: String,
+    pub reasoning_effort: Option<ReasoningEffort>,
+    pub rollout_path: PathBuf,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct ResumeConversationResponse {
+    pub conversation_id: ConversationId,
+    pub model: String,
+    pub initial_messages: Option<Vec<EventMsg>>,
+    pub rollout_path: PathBuf,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(untagged)]
+pub enum GetConversationSummaryParams {
+    RolloutPath {
+        #[serde(rename = "rolloutPath")]
+        rollout_path: PathBuf,
+    },
+    ConversationId {
+        #[serde(rename = "conversationId")]
+        conversation_id: ConversationId,
+    },
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct GetConversationSummaryResponse {
+    pub summary: ConversationSummary,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Default, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct ListConversationsParams {
+    pub page_size: Option<usize>,
+    pub cursor: Option<String>,
+    pub model_providers: Option<Vec<String>>,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct ConversationSummary {
+    pub conversation_id: ConversationId,
+    pub path: PathBuf,
+    pub preview: String,
+    pub timestamp: Option<String>,
+    pub model_provider: String,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct ListConversationsResponse {
+    pub items: Vec<ConversationSummary>,
+    pub next_cursor: Option<String>,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct ResumeConversationParams {
+    pub path: Option<PathBuf>,
+    pub conversation_id: Option<ConversationId>,
+    pub history: Option<Vec<ResponseItem>>,
+    pub overrides: Option<NewConversationParams>,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct AddConversationSubscriptionResponse {
+    #[schemars(with = "String")]
+    pub subscription_id: Uuid,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct ArchiveConversationParams {
+    pub conversation_id: ConversationId,
+    pub rollout_path: PathBuf,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct ArchiveConversationResponse {}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct RemoveConversationSubscriptionResponse {}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct LoginApiKeyParams {
+    pub api_key: String,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct LoginApiKeyResponse {}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct LoginChatGptResponse {
+    #[schemars(with = "String")]
+    pub login_id: Uuid,
+    pub auth_url: String,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct GitDiffToRemoteResponse {
+    pub sha: GitSha,
+    pub diff: String,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct CancelLoginChatGptParams {
+    #[schemars(with = "String")]
+    pub login_id: Uuid,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct GitDiffToRemoteParams {
+    pub cwd: PathBuf,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct CancelLoginChatGptResponse {}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct LogoutChatGptParams {}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct LogoutChatGptResponse {}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct GetAuthStatusParams {
+    pub include_token: Option<bool>,
+    pub refresh_token: Option<bool>,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct ExecOneOffCommandParams {
+    pub command: Vec<String>,
+    pub timeout_ms: Option<u64>,
+    pub cwd: Option<PathBuf>,
+    pub sandbox_policy: Option<SandboxPolicy>,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct ExecOneOffCommandResponse {
+    pub exit_code: i32,
+    pub stdout: String,
+    pub stderr: String,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct GetAuthStatusResponse {
+    pub auth_method: Option<AuthMode>,
+    pub auth_token: Option<String>,
+    pub requires_openai_auth: Option<bool>,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct GetUserAgentResponse {
+    pub user_agent: String,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct UserInfoResponse {
+    pub alleged_user_email: Option<String>,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct GetUserSavedConfigResponse {
+    pub config: UserSavedConfig,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct SetDefaultModelParams {
+    pub model: Option<String>,
+    pub reasoning_effort: Option<ReasoningEffort>,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct SetDefaultModelResponse {}
+
+#[derive(Deserialize, Debug, Clone, PartialEq, Serialize, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct UserSavedConfig {
+    pub approval_policy: Option<AskForApproval>,
+    pub sandbox_mode: Option<SandboxMode>,
+    pub sandbox_settings: Option<SandboxSettings>,
+    pub forced_chatgpt_workspace_id: Option<String>,
+    pub forced_login_method: Option<ForcedLoginMethod>,
+    pub model: Option<String>,
+    pub model_reasoning_effort: Option<ReasoningEffort>,
+    pub model_reasoning_summary: Option<ReasoningSummary>,
+    pub model_verbosity: Option<Verbosity>,
+    pub tools: Option<Tools>,
+    pub profile: Option<String>,
+    pub profiles: HashMap<String, Profile>,
+}
+
+#[derive(Deserialize, Debug, Clone, PartialEq, Serialize, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct Profile {
+    pub model: Option<String>,
+    pub model_provider: Option<String>,
+    pub approval_policy: Option<AskForApproval>,
+    pub model_reasoning_effort: Option<ReasoningEffort>,
+    pub model_reasoning_summary: Option<ReasoningSummary>,
+    pub model_verbosity: Option<Verbosity>,
+    pub chatgpt_base_url: Option<String>,
+}
+
+#[derive(Deserialize, Debug, Clone, PartialEq, Serialize, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct Tools {
+    pub web_search: Option<bool>,
+    pub view_image: Option<bool>,
+}
+
+#[derive(Deserialize, Debug, Clone, PartialEq, Serialize, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct SandboxSettings {
+    #[serde(default)]
+    pub writable_roots: Vec<PathBuf>,
+    pub network_access: Option<bool>,
+    pub exclude_tmpdir_env_var: Option<bool>,
+    pub exclude_slash_tmp: Option<bool>,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct SendUserMessageParams {
+    pub conversation_id: ConversationId,
+    pub items: Vec<InputItem>,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct SendUserTurnParams {
+    pub conversation_id: ConversationId,
+    pub items: Vec<InputItem>,
+    pub cwd: PathBuf,
+    pub approval_policy: AskForApproval,
+    pub sandbox_policy: SandboxPolicy,
+    pub model: String,
+    pub effort: Option<ReasoningEffort>,
+    pub summary: ReasoningSummary,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct SendUserTurnResponse {}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct InterruptConversationParams {
+    pub conversation_id: ConversationId,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct InterruptConversationResponse {
+    pub abort_reason: TurnAbortReason,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct SendUserMessageResponse {}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct AddConversationListenerParams {
+    pub conversation_id: ConversationId,
+    #[serde(default)]
+    pub experimental_raw_events: bool,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct RemoveConversationListenerParams {
+    #[schemars(with = "String")]
+    pub subscription_id: Uuid,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+#[serde(tag = "type", content = "data")]
+pub enum InputItem {
+    Text { text: String },
+    Image { image_url: String },
+    LocalImage { path: PathBuf },
+}
+
+// Deprecated notifications (v1)
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct LoginChatGptCompleteNotification {
+    #[schemars(with = "String")]
+    pub login_id: Uuid,
+    pub success: bool,
+    pub error: Option<String>,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct SessionConfiguredNotification {
+    pub session_id: ConversationId,
+    pub model: String,
+    pub reasoning_effort: Option<ReasoningEffort>,
+    pub history_log_id: u64,
+    #[ts(type = "number")]
+    pub history_entry_count: usize,
+    pub initial_messages: Option<Vec<EventMsg>>,
+    pub rollout_path: PathBuf,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct AuthStatusChangeNotification {
+    pub auth_method: Option<AuthMode>,
+}

codex-rs/app-server-protocol/src/protocol/v2.rs

@@ -0,0 +1,122 @@
+use codex_protocol::ConversationId;
+use codex_protocol::account::PlanType;
+use codex_protocol::config_types::ReasoningEffort;
+use codex_protocol::protocol::RateLimitSnapshot;
+use schemars::JsonSchema;
+use serde::Deserialize;
+use serde::Serialize;
+use ts_rs::TS;
+use uuid::Uuid;
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(tag = "type", rename_all = "camelCase")]
+#[ts(tag = "type")]
+pub enum Account {
+    #[serde(rename = "apiKey", rename_all = "camelCase")]
+    #[ts(rename = "apiKey", rename_all = "camelCase")]
+    ApiKey { api_key: String },
+
+    #[serde(rename = "chatgpt", rename_all = "camelCase")]
+    #[ts(rename = "chatgpt", rename_all = "camelCase")]
+    ChatGpt {
+        email: Option<String>,
+        plan_type: PlanType,
+    },
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(tag = "type")]
+#[ts(tag = "type")]
+pub enum LoginAccountParams {
+    #[serde(rename = "apiKey")]
+    #[ts(rename = "apiKey")]
+    ApiKey {
+        #[serde(rename = "apiKey")]
+        #[ts(rename = "apiKey")]
+        api_key: String,
+    },
+    #[serde(rename = "chatgpt")]
+    #[ts(rename = "chatgpt")]
+    ChatGpt,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct LoginAccountResponse {
+    /// Only set if the login method is ChatGPT.
+    #[schemars(with = "String")]
+    pub login_id: Option<Uuid>,
+
+    /// URL the client should open in a browser to initiate the OAuth flow.
+    /// Only set if the login method is ChatGPT.
+    pub auth_url: Option<String>,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct LogoutAccountResponse {}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct GetAccountRateLimitsResponse {
+    pub rate_limits: RateLimitSnapshot,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct GetAccountResponse {
+    pub account: Account,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Default, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct ListModelsParams {
+    /// Optional page size; defaults to a reasonable server-side value.
+    pub page_size: Option<usize>,
+    /// Opaque pagination cursor returned by a previous call.
+    pub cursor: Option<String>,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct Model {
+    pub id: String,
+    pub model: String,
+    pub display_name: String,
+    pub description: String,
+    pub supported_reasoning_efforts: Vec<ReasoningEffortOption>,
+    pub default_reasoning_effort: ReasoningEffort,
+    // Only one model should be marked as default.
+    pub is_default: bool,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct ReasoningEffortOption {
+    pub reasoning_effort: ReasoningEffort,
+    pub description: String,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct ListModelsResponse {
+    pub items: Vec<Model>,
+    /// Opaque cursor to pass to the next call to continue after the last item.
+    /// if None, there are no more items to return.
+    pub next_cursor: Option<String>,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct UploadFeedbackParams {
+    pub classification: String,
+    pub reason: Option<String>,
+    pub conversation_id: Option<ConversationId>,
+    pub include_logs: bool,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct UploadFeedbackResponse {
+    pub thread_id: String,
+}

codex-rs/app-server/Cargo.toml

@@ -0,0 +1,54 @@
+[package]
+edition = "2024"
+name = "codex-app-server"
+version = { workspace = true }
+
+[[bin]]
+name = "codex-app-server"
+path = "src/main.rs"
+
+[lib]
+name = "codex_app_server"
+path = "src/lib.rs"
+
+[lints]
+workspace = true
+
+[dependencies]
+anyhow = { workspace = true }
+codex-arg0 = { workspace = true }
+codex-common = { workspace = true, features = ["cli"] }
+codex-core = { workspace = true }
+codex-backend-client = { workspace = true }
+codex-file-search = { workspace = true }
+codex-login = { workspace = true }
+codex-protocol = { workspace = true }
+codex-app-server-protocol = { workspace = true }
+codex-feedback = { workspace = true }
+codex-utils-json-to-toml = { workspace = true }
+chrono = { workspace = true }
+serde = { workspace = true, features = ["derive"] }
+serde_json = { workspace = true }
+tokio = { workspace = true, features = [
+    "io-std",
+    "macros",
+    "process",
+    "rt-multi-thread",
+    "signal",
+] }
+tracing = { workspace = true, features = ["log"] }
+tracing-subscriber = { workspace = true, features = ["env-filter", "fmt"] }
+opentelemetry-appender-tracing = { workspace = true }
+uuid = { workspace = true, features = ["serde", "v7"] }
+
+[dev-dependencies]
+app_test_support = { workspace = true }
+assert_cmd = { workspace = true }
+base64 = { workspace = true }
+core_test_support = { workspace = true }
+os_info = { workspace = true }
+pretty_assertions = { workspace = true }
+serial_test = { workspace = true }
+tempfile = { workspace = true }
+toml = { workspace = true }
+wiremock = { workspace = true }

codex-rs/app-server/README.md

@@ -0,0 +1,15 @@
+# codex-app-server
+
+`codex app-server` is the harness Codex uses to power rich interfaces such as the [Codex VS Code extension](https://marketplace.visualstudio.com/items?itemName=openai.chatgpt). The message schema is currently unstable, but those who wish to build experimental UIs on top of Codex may find it valuable.
+
+## Protocol
+
+Similar to [MCP](https://modelcontextprotocol.io/), `codex app-server` supports bidirectional communication, streaming JSONL over stdio. The protocol is JSON-RPC 2.0, though the `"jsonrpc":"2.0"` header is omitted.
+
+## Message Schema
+
+Currently, you can dump a TypeScript version of the schema using `codex generate-ts`. It is specific to the version of Codex you used to run `generate-ts`, so the two are guaranteed to be compatible.
+
+```
+codex generate-ts --out DIR
+```

codex-rs/app-server/src/error_code.rs

@@ -0,0 +1,2 @@
+pub(crate) const INVALID_REQUEST_ERROR_CODE: i64 = -32600;
+pub(crate) const INTERNAL_ERROR_CODE: i64 = -32603;

codex-rs/app-server/src/fuzzy_file_search.rs

@@ -0,0 +1,93 @@
+use std::num::NonZero;
+use std::num::NonZeroUsize;
+use std::path::Path;
+use std::path::PathBuf;
+use std::sync::Arc;
+use std::sync::atomic::AtomicBool;
+
+use codex_app_server_protocol::FuzzyFileSearchResult;
+use codex_file_search as file_search;
+use tokio::task::JoinSet;
+use tracing::warn;
+
+const LIMIT_PER_ROOT: usize = 50;
+const MAX_THREADS: usize = 12;
+const COMPUTE_INDICES: bool = true;
+
+pub(crate) async fn run_fuzzy_file_search(
+    query: String,
+    roots: Vec<String>,
+    cancellation_flag: Arc<AtomicBool>,
+) -> Vec<FuzzyFileSearchResult> {
+    #[expect(clippy::expect_used)]
+    let limit_per_root =
+        NonZero::new(LIMIT_PER_ROOT).expect("LIMIT_PER_ROOT should be a valid non-zero usize");
+
+    let cores = std::thread::available_parallelism()
+        .map(std::num::NonZero::get)
+        .unwrap_or(1);
+    let threads = cores.min(MAX_THREADS);
+    let threads_per_root = (threads / roots.len()).max(1);
+    let threads = NonZero::new(threads_per_root).unwrap_or(NonZeroUsize::MIN);
+
+    let mut files: Vec<FuzzyFileSearchResult> = Vec::new();
+    let mut join_set = JoinSet::new();
+
+    for root in roots {
+        let search_dir = PathBuf::from(&root);
+        let query = query.clone();
+        let cancel_flag = cancellation_flag.clone();
+        join_set.spawn_blocking(move || {
+            match file_search::run(
+                query.as_str(),
+                limit_per_root,
+                &search_dir,
+                Vec::new(),
+                threads,
+                cancel_flag,
+                COMPUTE_INDICES,
+                true,
+            ) {
+                Ok(res) => Ok((root, res)),
+                Err(err) => Err((root, err)),
+            }
+        });
+    }
+
+    while let Some(res) = join_set.join_next().await {
+        match res {
+            Ok(Ok((root, res))) => {
+                for m in res.matches {
+                    let path = m.path;
+                    //TODO(shijie): Move file name generation to file_search lib.
+                    let file_name = Path::new(&path)
+                        .file_name()
+                        .map(|name| name.to_string_lossy().into_owned())
+                        .unwrap_or_else(|| path.clone());
+                    let result = FuzzyFileSearchResult {
+                        root: root.clone(),
+                        path,
+                        file_name,
+                        score: m.score,
+                        indices: m.indices,
+                    };
+                    files.push(result);
+                }
+            }
+            Ok(Err((root, err))) => {
+                warn!("fuzzy-file-search in dir '{root}' failed: {err}");
+            }
+            Err(err) => {
+                warn!("fuzzy-file-search join_next failed: {err}");
+            }
+        }
+    }
+
+    files.sort_by(file_search::cmp_by_score_desc_then_path_asc::<
+        FuzzyFileSearchResult,
+        _,
+        _,
+    >(|f| f.score, |f| f.path.as_str()));
+
+    files
+}

codex-rs/app-server/src/lib.rs

@@ -0,0 +1,172 @@
+#![deny(clippy::print_stdout, clippy::print_stderr)]
+
+use codex_common::CliConfigOverrides;
+use codex_core::config::Config;
+use codex_core::config::ConfigOverrides;
+use opentelemetry_appender_tracing::layer::OpenTelemetryTracingBridge;
+use std::io::ErrorKind;
+use std::io::Result as IoResult;
+use std::path::PathBuf;
+
+use crate::message_processor::MessageProcessor;
+use crate::outgoing_message::OutgoingMessage;
+use crate::outgoing_message::OutgoingMessageSender;
+use codex_app_server_protocol::JSONRPCMessage;
+use codex_feedback::CodexFeedback;
+use tokio::io::AsyncBufReadExt;
+use tokio::io::AsyncWriteExt;
+use tokio::io::BufReader;
+use tokio::io::{self};
+use tokio::sync::mpsc;
+use tracing::Level;
+use tracing::debug;
+use tracing::error;
+use tracing::info;
+use tracing_subscriber::EnvFilter;
+use tracing_subscriber::Layer;
+use tracing_subscriber::filter::Targets;
+use tracing_subscriber::layer::SubscriberExt;
+use tracing_subscriber::util::SubscriberInitExt;
+
+mod codex_message_processor;
+mod error_code;
+mod fuzzy_file_search;
+mod message_processor;
+mod models;
+mod outgoing_message;
+
+/// Size of the bounded channels used to communicate between tasks. The value
+/// is a balance between throughput and memory usage – 128 messages should be
+/// plenty for an interactive CLI.
+const CHANNEL_CAPACITY: usize = 128;
+
+pub async fn run_main(
+    codex_linux_sandbox_exe: Option<PathBuf>,
+    cli_config_overrides: CliConfigOverrides,
+) -> IoResult<()> {
+    // Set up channels.
+    let (incoming_tx, mut incoming_rx) = mpsc::channel::<JSONRPCMessage>(CHANNEL_CAPACITY);
+    let (outgoing_tx, mut outgoing_rx) = mpsc::unbounded_channel::<OutgoingMessage>();
+
+    // Task: read from stdin, push to `incoming_tx`.
+    let stdin_reader_handle = tokio::spawn({
+        async move {
+            let stdin = io::stdin();
+            let reader = BufReader::new(stdin);
+            let mut lines = reader.lines();
+
+            while let Some(line) = lines.next_line().await.unwrap_or_default() {
+                match serde_json::from_str::<JSONRPCMessage>(&line) {
+                    Ok(msg) => {
+                        if incoming_tx.send(msg).await.is_err() {
+                            // Receiver gone – nothing left to do.
+                            break;
+                        }
+                    }
+                    Err(e) => error!("Failed to deserialize JSONRPCMessage: {e}"),
+                }
+            }
+
+            debug!("stdin reader finished (EOF)");
+        }
+    });
+
+    // Parse CLI overrides once and derive the base Config eagerly so later
+    // components do not need to work with raw TOML values.
+    let cli_kv_overrides = cli_config_overrides.parse_overrides().map_err(|e| {
+        std::io::Error::new(
+            ErrorKind::InvalidInput,
+            format!("error parsing -c overrides: {e}"),
+        )
+    })?;
+    let config = Config::load_with_cli_overrides(cli_kv_overrides, ConfigOverrides::default())
+        .await
+        .map_err(|e| {
+            std::io::Error::new(ErrorKind::InvalidData, format!("error loading config: {e}"))
+        })?;
+
+    let feedback = CodexFeedback::new();
+
+    let otel =
+        codex_core::otel_init::build_provider(&config, env!("CARGO_PKG_VERSION")).map_err(|e| {
+            std::io::Error::new(
+                ErrorKind::InvalidData,
+                format!("error loading otel config: {e}"),
+            )
+        })?;
+
+    // Install a simple subscriber so `tracing` output is visible.  Users can
+    // control the log level with `RUST_LOG`.
+    let stderr_fmt = tracing_subscriber::fmt::layer()
+        .with_writer(std::io::stderr)
+        .with_filter(EnvFilter::from_default_env());
+
+    let feedback_layer = tracing_subscriber::fmt::layer()
+        .with_writer(feedback.make_writer())
+        .with_ansi(false)
+        .with_target(false)
+        .with_filter(Targets::new().with_default(Level::TRACE));
+
+    let _ = tracing_subscriber::registry()
+        .with(stderr_fmt)
+        .with(feedback_layer)
+        .with(otel.as_ref().map(|provider| {
+            OpenTelemetryTracingBridge::new(&provider.logger).with_filter(
+                tracing_subscriber::filter::filter_fn(codex_core::otel_init::codex_export_filter),
+            )
+        }))
+        .try_init();
+
+    // Task: process incoming messages.
+    let processor_handle = tokio::spawn({
+        let outgoing_message_sender = OutgoingMessageSender::new(outgoing_tx);
+        let mut processor = MessageProcessor::new(
+            outgoing_message_sender,
+            codex_linux_sandbox_exe,
+            std::sync::Arc::new(config),
+            feedback.clone(),
+        );
+        async move {
+            while let Some(msg) = incoming_rx.recv().await {
+                match msg {
+                    JSONRPCMessage::Request(r) => processor.process_request(r).await,
+                    JSONRPCMessage::Response(r) => processor.process_response(r).await,
+                    JSONRPCMessage::Notification(n) => processor.process_notification(n).await,
+                    JSONRPCMessage::Error(e) => processor.process_error(e),
+                }
+            }
+
+            info!("processor task exited (channel closed)");
+        }
+    });
+
+    // Task: write outgoing messages to stdout.
+    let stdout_writer_handle = tokio::spawn(async move {
+        let mut stdout = io::stdout();
+        while let Some(outgoing_message) = outgoing_rx.recv().await {
+            let Ok(value) = serde_json::to_value(outgoing_message) else {
+                error!("Failed to convert OutgoingMessage to JSON value");
+                continue;
+            };
+            match serde_json::to_string(&value) {
+                Ok(mut json) => {
+                    json.push('\n');
+                    if let Err(e) = stdout.write_all(json.as_bytes()).await {
+                        error!("Failed to write to stdout: {e}");
+                        break;
+                    }
+                }
+                Err(e) => error!("Failed to serialize JSONRPCMessage: {e}"),
+            }
+        }
+
+        info!("stdout writer exited (channel closed)");
+    });
+
+    // Wait for all tasks to finish.  The typical exit path is the stdin reader
+    // hitting EOF which, once it drops `incoming_tx`, propagates shutdown to
+    // the processor and then to the stdout task.
+    let _ = tokio::join!(stdin_reader_handle, processor_handle, stdout_writer_handle);
+
+    Ok(())
+}

codex-rs/app-server/src/main.rs

@@ -0,0 +1,10 @@
+use codex_app_server::run_main;
+use codex_arg0::arg0_dispatch_or_else;
+use codex_common::CliConfigOverrides;
+
+fn main() -> anyhow::Result<()> {
+    arg0_dispatch_or_else(|codex_linux_sandbox_exe| async move {
+        run_main(codex_linux_sandbox_exe, CliConfigOverrides::default()).await?;
+        Ok(())
+    })
+}

codex-rs/app-server/src/message_processor.rs

@@ -0,0 +1,144 @@
+use std::path::PathBuf;
+
+use crate::codex_message_processor::CodexMessageProcessor;
+use crate::error_code::INVALID_REQUEST_ERROR_CODE;
+use crate::outgoing_message::OutgoingMessageSender;
+use codex_app_server_protocol::ClientInfo;
+use codex_app_server_protocol::ClientRequest;
+use codex_app_server_protocol::InitializeResponse;
+
+use codex_app_server_protocol::JSONRPCError;
+use codex_app_server_protocol::JSONRPCErrorError;
+use codex_app_server_protocol::JSONRPCNotification;
+use codex_app_server_protocol::JSONRPCRequest;
+use codex_app_server_protocol::JSONRPCResponse;
+use codex_core::AuthManager;
+use codex_core::ConversationManager;
+use codex_core::config::Config;
+use codex_core::default_client::USER_AGENT_SUFFIX;
+use codex_core::default_client::get_codex_user_agent;
+use codex_feedback::CodexFeedback;
+use codex_protocol::protocol::SessionSource;
+use std::sync::Arc;
+
+pub(crate) struct MessageProcessor {
+    outgoing: Arc<OutgoingMessageSender>,
+    codex_message_processor: CodexMessageProcessor,
+    initialized: bool,
+}
+
+impl MessageProcessor {
+    /// Create a new `MessageProcessor`, retaining a handle to the outgoing
+    /// `Sender` so handlers can enqueue messages to be written to stdout.
+    pub(crate) fn new(
+        outgoing: OutgoingMessageSender,
+        codex_linux_sandbox_exe: Option<PathBuf>,
+        config: Arc<Config>,
+        feedback: CodexFeedback,
+    ) -> Self {
+        let outgoing = Arc::new(outgoing);
+        let auth_manager = AuthManager::shared(
+            config.codex_home.clone(),
+            false,
+            config.cli_auth_credentials_store_mode,
+        );
+        let conversation_manager = Arc::new(ConversationManager::new(
+            auth_manager.clone(),
+            SessionSource::VSCode,
+        ));
+        let codex_message_processor = CodexMessageProcessor::new(
+            auth_manager,
+            conversation_manager,
+            outgoing.clone(),
+            codex_linux_sandbox_exe,
+            config,
+            feedback,
+        );
+
+        Self {
+            outgoing,
+            codex_message_processor,
+            initialized: false,
+        }
+    }
+
+    pub(crate) async fn process_request(&mut self, request: JSONRPCRequest) {
+        let request_id = request.id.clone();
+        if let Ok(request_json) = serde_json::to_value(request)
+            && let Ok(codex_request) = serde_json::from_value::<ClientRequest>(request_json)
+        {
+            match codex_request {
+                // Handle Initialize internally so CodexMessageProcessor does not have to concern
+                // itself with the `initialized` bool.
+                ClientRequest::Initialize { request_id, params } => {
+                    if self.initialized {
+                        let error = JSONRPCErrorError {
+                            code: INVALID_REQUEST_ERROR_CODE,
+                            message: "Already initialized".to_string(),
+                            data: None,
+                        };
+                        self.outgoing.send_error(request_id, error).await;
+                        return;
+                    } else {
+                        let ClientInfo {
+                            name,
+                            title: _title,
+                            version,
+                        } = params.client_info;
+                        let user_agent_suffix = format!("{name}; {version}");
+                        if let Ok(mut suffix) = USER_AGENT_SUFFIX.lock() {
+                            *suffix = Some(user_agent_suffix);
+                        }
+
+                        let user_agent = get_codex_user_agent();
+                        let response = InitializeResponse { user_agent };
+                        self.outgoing.send_response(request_id, response).await;
+
+                        self.initialized = true;
+                        return;
+                    }
+                }
+                _ => {
+                    if !self.initialized {
+                        let error = JSONRPCErrorError {
+                            code: INVALID_REQUEST_ERROR_CODE,
+                            message: "Not initialized".to_string(),
+                            data: None,
+                        };
+                        self.outgoing.send_error(request_id, error).await;
+                        return;
+                    }
+                }
+            }
+
+            self.codex_message_processor
+                .process_request(codex_request)
+                .await;
+        } else {
+            let error = JSONRPCErrorError {
+                code: INVALID_REQUEST_ERROR_CODE,
+                message: "Invalid request".to_string(),
+                data: None,
+            };
+            self.outgoing.send_error(request_id, error).await;
+        }
+    }
+
+    pub(crate) async fn process_notification(&self, notification: JSONRPCNotification) {
+        // Currently, we do not expect to receive any notifications from the
+        // client, so we just log them.
+        tracing::info!("<- notification: {:?}", notification);
+    }
+
+    /// Handle a standalone JSON-RPC response originating from the peer.
+    pub(crate) async fn process_response(&mut self, response: JSONRPCResponse) {
+        tracing::info!("<- response: {:?}", response);
+        let JSONRPCResponse { id, result, .. } = response;
+        self.outgoing.notify_client_response(id, result).await
+    }
+
+    /// Handle an error object received from the peer.
+    pub(crate) fn process_error(&mut self, err: JSONRPCError) {
+        tracing::error!("<- error: {:?}", err);
+    }
+}

codex-rs/app-server/src/models.rs

@@ -0,0 +1,38 @@
+use codex_app_server_protocol::Model;
+use codex_app_server_protocol::ReasoningEffortOption;
+use codex_common::model_presets::ModelPreset;
+use codex_common::model_presets::ReasoningEffortPreset;
+use codex_common::model_presets::builtin_model_presets;
+
+pub fn supported_models() -> Vec<Model> {
+    builtin_model_presets(None)
+        .into_iter()
+        .map(model_from_preset)
+        .collect()
+}
+
+fn model_from_preset(preset: ModelPreset) -> Model {
+    Model {
+        id: preset.id.to_string(),
+        model: preset.model.to_string(),
+        display_name: preset.display_name.to_string(),
+        description: preset.description.to_string(),
+        supported_reasoning_efforts: reasoning_efforts_from_preset(
+            preset.supported_reasoning_efforts,
+        ),
+        default_reasoning_effort: preset.default_reasoning_effort,
+        is_default: preset.is_default,
+    }
+}
+
+fn reasoning_efforts_from_preset(
+    efforts: &'static [ReasoningEffortPreset],
+) -> Vec<ReasoningEffortOption> {
+    efforts
+        .iter()
+        .map(|preset| ReasoningEffortOption {
+            reasoning_effort: preset.effort,
+            description: preset.description.to_string(),
+        })
+        .collect()
+}

codex-rs/app-server/src/outgoing_message.rs

@@ -0,0 +1,207 @@
+use std::collections::HashMap;
+use std::sync::atomic::AtomicI64;
+use std::sync::atomic::Ordering;
+
+use codex_app_server_protocol::JSONRPCErrorError;
+use codex_app_server_protocol::RequestId;
+use codex_app_server_protocol::Result;
+use codex_app_server_protocol::ServerNotification;
+use codex_app_server_protocol::ServerRequest;
+use codex_app_server_protocol::ServerRequestPayload;
+use serde::Serialize;
+use tokio::sync::Mutex;
+use tokio::sync::mpsc;
+use tokio::sync::oneshot;
+use tracing::warn;
+
+use crate::error_code::INTERNAL_ERROR_CODE;
+
+/// Sends messages to the client and manages request callbacks.
+pub(crate) struct OutgoingMessageSender {
+    next_request_id: AtomicI64,
+    sender: mpsc::UnboundedSender<OutgoingMessage>,
+    request_id_to_callback: Mutex<HashMap<RequestId, oneshot::Sender<Result>>>,
+}
+
+impl OutgoingMessageSender {
+    pub(crate) fn new(sender: mpsc::UnboundedSender<OutgoingMessage>) -> Self {
+        Self {
+            next_request_id: AtomicI64::new(0),
+            sender,
+            request_id_to_callback: Mutex::new(HashMap::new()),
+        }
+    }
+
+    pub(crate) async fn send_request(
+        &self,
+        request: ServerRequestPayload,
+    ) -> oneshot::Receiver<Result> {
+        let id = RequestId::Integer(self.next_request_id.fetch_add(1, Ordering::Relaxed));
+        let outgoing_message_id = id.clone();
+        let (tx_approve, rx_approve) = oneshot::channel();
+        {
+            let mut request_id_to_callback = self.request_id_to_callback.lock().await;
+            request_id_to_callback.insert(id, tx_approve);
+        }
+
+        let outgoing_message =
+            OutgoingMessage::Request(request.request_with_id(outgoing_message_id));
+        let _ = self.sender.send(outgoing_message);
+        rx_approve
+    }
+
+    pub(crate) async fn notify_client_response(&self, id: RequestId, result: Result) {
+        let entry = {
+            let mut request_id_to_callback = self.request_id_to_callback.lock().await;
+            request_id_to_callback.remove_entry(&id)
+        };
+
+        match entry {
+            Some((id, sender)) => {
+                if let Err(err) = sender.send(result) {
+                    warn!("could not notify callback for {id:?} due to: {err:?}");
+                }
+            }
+            None => {
+                warn!("could not find callback for {id:?}");
+            }
+        }
+    }
+
+    pub(crate) async fn send_response<T: Serialize>(&self, id: RequestId, response: T) {
+        match serde_json::to_value(response) {
+            Ok(result) => {
+                let outgoing_message = OutgoingMessage::Response(OutgoingResponse { id, result });
+                let _ = self.sender.send(outgoing_message);
+            }
+            Err(err) => {
+                self.send_error(
+                    id,
+                    JSONRPCErrorError {
+                        code: INTERNAL_ERROR_CODE,
+                        message: format!("failed to serialize response: {err}"),
+                        data: None,
+                    },
+                )
+                .await;
+            }
+        }
+    }
+
+    pub(crate) async fn send_server_notification(&self, notification: ServerNotification) {
+        let _ = self
+            .sender
+            .send(OutgoingMessage::AppServerNotification(notification));
+    }
+
+    /// All notifications should be migrated to [`ServerNotification`] and
+    /// [`OutgoingMessage::Notification`] should be removed.
+    pub(crate) async fn send_notification(&self, notification: OutgoingNotification) {
+        let outgoing_message = OutgoingMessage::Notification(notification);
+        let _ = self.sender.send(outgoing_message);
+    }
+
+    pub(crate) async fn send_error(&self, id: RequestId, error: JSONRPCErrorError) {
+        let outgoing_message = OutgoingMessage::Error(OutgoingError { id, error });
+        let _ = self.sender.send(outgoing_message);
+    }
+}
+
+/// Outgoing message from the server to the client.
+#[derive(Debug, Clone, Serialize)]
+#[serde(untagged)]
+pub(crate) enum OutgoingMessage {
+    Request(ServerRequest),
+    Notification(OutgoingNotification),
+    /// AppServerNotification is specific to the case where this is run as an
+    /// "app server" as opposed to an MCP server.
+    AppServerNotification(ServerNotification),
+    Response(OutgoingResponse),
+    Error(OutgoingError),
+}
+
+#[derive(Debug, Clone, PartialEq, Serialize)]
+pub(crate) struct OutgoingNotification {
+    pub method: String,
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub params: Option<serde_json::Value>,
+}
+
+#[derive(Debug, Clone, PartialEq, Serialize)]
+pub(crate) struct OutgoingResponse {
+    pub id: RequestId,
+    pub result: Result,
+}
+
+#[derive(Debug, Clone, PartialEq, Serialize)]
+pub(crate) struct OutgoingError {
+    pub error: JSONRPCErrorError,
+    pub id: RequestId,
+}
+
+#[cfg(test)]
+mod tests {
+    use codex_app_server_protocol::LoginChatGptCompleteNotification;
+    use codex_protocol::protocol::RateLimitSnapshot;
+    use codex_protocol::protocol::RateLimitWindow;
+    use pretty_assertions::assert_eq;
+    use serde_json::json;
+    use uuid::Uuid;
+
+    use super::*;
+
+    #[test]
+    fn verify_server_notification_serialization() {
+        let notification =
+            ServerNotification::LoginChatGptComplete(LoginChatGptCompleteNotification {
+                login_id: Uuid::nil(),
+                success: true,
+                error: None,
+            });
+
+        let jsonrpc_notification = OutgoingMessage::AppServerNotification(notification);
+        assert_eq!(
+            json!({
+                "method": "loginChatGptComplete",
+                "params": {
+                    "loginId": Uuid::nil(),
+                    "success": true,
+                    "error": null,
+                },
+            }),
+            serde_json::to_value(jsonrpc_notification)
+                .expect("ensure the strum macros serialize the method field correctly"),
+            "ensure the strum macros serialize the method field correctly"
+        );
+    }
+
+    #[test]
+    fn verify_account_rate_limits_notification_serialization() {
+        let notification = ServerNotification::AccountRateLimitsUpdated(RateLimitSnapshot {
+            primary: Some(RateLimitWindow {
+                used_percent: 25.0,
+                window_minutes: Some(15),
+                resets_at: Some(123),
+            }),
+            secondary: None,
+        });
+
+        let jsonrpc_notification = OutgoingMessage::AppServerNotification(notification);
+        assert_eq!(
+            json!({
+                "method": "account/rateLimits/updated",
+                "params": {
+                    "primary": {
+                        "used_percent": 25.0,
+                        "window_minutes": 15,
+                        "resets_at": 123,
+                    },
+                    "secondary": null,
+                },
+            }),
+            serde_json::to_value(jsonrpc_notification)
+                .expect("ensure the notification serializes correctly"),
+            "ensure the notification serializes correctly"
+        );
+    }
+}

codex-rs/app-server/tests/all.rs

@@ -0,0 +1,3 @@
+// Single integration test binary that aggregates all test modules.
+// The submodules live in `tests/suite/`.
+mod suite;

codex-rs/app-server/tests/common/Cargo.toml

@@ -0,0 +1,24 @@
+[package]
+edition = "2024"
+name = "app_test_support"
+version = { workspace = true }
+
+[lib]
+path = "lib.rs"
+
+[dependencies]
+anyhow = { workspace = true }
+assert_cmd = { workspace = true }
+base64 = { workspace = true }
+chrono = { workspace = true }
+codex-app-server-protocol = { workspace = true }
+codex-core = { workspace = true }
+serde = { workspace = true }
+serde_json = { workspace = true }
+tokio = { workspace = true, features = [
+    "io-std",
+    "macros",
+    "process",
+    "rt-multi-thread",
+] }
+wiremock = { workspace = true }

codex-rs/app-server/tests/common/auth_fixtures.rs

@@ -0,0 +1,135 @@
+use std::path::Path;
+
+use anyhow::Context;
+use anyhow::Result;
+use base64::Engine;
+use base64::engine::general_purpose::URL_SAFE_NO_PAD;
+use chrono::DateTime;
+use chrono::Utc;
+use codex_core::auth::AuthCredentialsStoreMode;
+use codex_core::auth::AuthDotJson;
+use codex_core::auth::save_auth;
+use codex_core::token_data::TokenData;
+use codex_core::token_data::parse_id_token;
+use serde_json::json;
+
+/// Builder for writing a fake ChatGPT auth.json in tests.
+#[derive(Debug, Clone)]
+pub struct ChatGptAuthFixture {
+    access_token: String,
+    refresh_token: String,
+    account_id: Option<String>,
+    claims: ChatGptIdTokenClaims,
+    last_refresh: Option<Option<DateTime<Utc>>>,
+}
+
+impl ChatGptAuthFixture {
+    pub fn new(access_token: impl Into<String>) -> Self {
+        Self {
+            access_token: access_token.into(),
+            refresh_token: "refresh-token".to_string(),
+            account_id: None,
+            claims: ChatGptIdTokenClaims::default(),
+            last_refresh: None,
+        }
+    }
+
+    pub fn refresh_token(mut self, refresh_token: impl Into<String>) -> Self {
+        self.refresh_token = refresh_token.into();
+        self
+    }
+
+    pub fn account_id(mut self, account_id: impl Into<String>) -> Self {
+        self.account_id = Some(account_id.into());
+        self
+    }
+
+    pub fn plan_type(mut self, plan_type: impl Into<String>) -> Self {
+        self.claims.plan_type = Some(plan_type.into());
+        self
+    }
+
+    pub fn email(mut self, email: impl Into<String>) -> Self {
+        self.claims.email = Some(email.into());
+        self
+    }
+
+    pub fn last_refresh(mut self, last_refresh: Option<DateTime<Utc>>) -> Self {
+        self.last_refresh = Some(last_refresh);
+        self
+    }
+
+    pub fn claims(mut self, claims: ChatGptIdTokenClaims) -> Self {
+        self.claims = claims;
+        self
+    }
+}
+
+#[derive(Debug, Clone, Default)]
+pub struct ChatGptIdTokenClaims {
+    pub email: Option<String>,
+    pub plan_type: Option<String>,
+}
+
+impl ChatGptIdTokenClaims {
+    pub fn new() -> Self {
+        Self::default()
+    }
+
+    pub fn email(mut self, email: impl Into<String>) -> Self {
+        self.email = Some(email.into());
+        self
+    }
+
+    pub fn plan_type(mut self, plan_type: impl Into<String>) -> Self {
+        self.plan_type = Some(plan_type.into());
+        self
+    }
+}
+
+pub fn encode_id_token(claims: &ChatGptIdTokenClaims) -> Result<String> {
+    let header = json!({ "alg": "none", "typ": "JWT" });
+    let mut payload = serde_json::Map::new();
+    if let Some(email) = &claims.email {
+        payload.insert("email".to_string(), json!(email));
+    }
+    if let Some(plan_type) = &claims.plan_type {
+        payload.insert(
+            "https://api.openai.com/auth".to_string(),
+            json!({ "chatgpt_plan_type": plan_type }),
+        );
+    }
+    let payload = serde_json::Value::Object(payload);
+
+    let header_b64 =
+        URL_SAFE_NO_PAD.encode(serde_json::to_vec(&header).context("serialize jwt header")?);
+    let payload_b64 =
+        URL_SAFE_NO_PAD.encode(serde_json::to_vec(&payload).context("serialize jwt payload")?);
+    let signature_b64 = URL_SAFE_NO_PAD.encode(b"signature");
+    Ok(format!("{header_b64}.{payload_b64}.{signature_b64}"))
+}
+
+pub fn write_chatgpt_auth(
+    codex_home: &Path,
+    fixture: ChatGptAuthFixture,
+    cli_auth_credentials_store_mode: AuthCredentialsStoreMode,
+) -> Result<()> {
+    let id_token_raw = encode_id_token(&fixture.claims)?;
+    let id_token = parse_id_token(&id_token_raw).context("parse id token")?;
+    let tokens = TokenData {
+        id_token,
+        access_token: fixture.access_token,
+        refresh_token: fixture.refresh_token,
+        account_id: fixture.account_id,
+    };
+
+    let last_refresh = fixture.last_refresh.unwrap_or_else(|| Some(Utc::now()));
+
+    let auth = AuthDotJson {
+        openai_api_key: None,
+        tokens: Some(tokens),
+        last_refresh,
+    };
+
+    save_auth(codex_home, &auth, cli_auth_credentials_store_mode).context("write auth.json")
+}

codex-rs/app-server/tests/common/lib.rs

@@ -0,0 +1,22 @@
+mod auth_fixtures;
+mod mcp_process;
+mod mock_model_server;
+mod responses;
+
+pub use auth_fixtures::ChatGptAuthFixture;
+pub use auth_fixtures::ChatGptIdTokenClaims;
+pub use auth_fixtures::encode_id_token;
+pub use auth_fixtures::write_chatgpt_auth;
+use codex_app_server_protocol::JSONRPCResponse;
+pub use mcp_process::McpProcess;
+pub use mock_model_server::create_mock_chat_completions_server;
+pub use responses::create_apply_patch_sse_response;
+pub use responses::create_final_assistant_message_sse_response;
+pub use responses::create_shell_sse_response;
+use serde::de::DeserializeOwned;
+
+pub fn to_response<T: DeserializeOwned>(response: JSONRPCResponse) -> anyhow::Result<T> {
+    let value = serde_json::to_value(response.result)?;
+    let codex_response = serde_json::from_value(value)?;
+    Ok(codex_response)
+}

codex-rs/app-server/tests/common/mcp_process.rs

@@ -0,0 +1,527 @@
+use std::collections::VecDeque;
+use std::path::Path;
+use std::process::Stdio;
+use std::sync::atomic::AtomicI64;
+use std::sync::atomic::Ordering;
+use tokio::io::AsyncBufReadExt;
+use tokio::io::AsyncWriteExt;
+use tokio::io::BufReader;
+use tokio::process::Child;
+use tokio::process::ChildStdin;
+use tokio::process::ChildStdout;
+
+use anyhow::Context;
+use assert_cmd::prelude::*;
+use codex_app_server_protocol::AddConversationListenerParams;
+use codex_app_server_protocol::ArchiveConversationParams;
+use codex_app_server_protocol::CancelLoginChatGptParams;
+use codex_app_server_protocol::ClientInfo;
+use codex_app_server_protocol::ClientNotification;
+use codex_app_server_protocol::GetAuthStatusParams;
+use codex_app_server_protocol::InitializeParams;
+use codex_app_server_protocol::InterruptConversationParams;
+use codex_app_server_protocol::ListConversationsParams;
+use codex_app_server_protocol::ListModelsParams;
+use codex_app_server_protocol::LoginApiKeyParams;
+use codex_app_server_protocol::NewConversationParams;
+use codex_app_server_protocol::RemoveConversationListenerParams;
+use codex_app_server_protocol::ResumeConversationParams;
+use codex_app_server_protocol::SendUserMessageParams;
+use codex_app_server_protocol::SendUserTurnParams;
+use codex_app_server_protocol::ServerRequest;
+use codex_app_server_protocol::SetDefaultModelParams;
+use codex_app_server_protocol::UploadFeedbackParams;
+
+use codex_app_server_protocol::JSONRPCError;
+use codex_app_server_protocol::JSONRPCMessage;
+use codex_app_server_protocol::JSONRPCNotification;
+use codex_app_server_protocol::JSONRPCRequest;
+use codex_app_server_protocol::JSONRPCResponse;
+use codex_app_server_protocol::RequestId;
+use std::process::Command as StdCommand;
+use tokio::process::Command;
+
+pub struct McpProcess {
+    next_request_id: AtomicI64,
+    /// Retain this child process until the client is dropped. The Tokio runtime
+    /// will make a "best effort" to reap the process after it exits, but it is
+    /// not a guarantee. See the `kill_on_drop` documentation for details.
+    #[allow(dead_code)]
+    process: Child,
+    stdin: ChildStdin,
+    stdout: BufReader<ChildStdout>,
+    pending_user_messages: VecDeque<JSONRPCNotification>,
+}
+
+impl McpProcess {
+    pub async fn new(codex_home: &Path) -> anyhow::Result<Self> {
+        Self::new_with_env(codex_home, &[]).await
+    }
+
+    /// Creates a new MCP process, allowing tests to override or remove
+    /// specific environment variables for the child process only.
+    ///
+    /// Pass a tuple of (key, Some(value)) to set/override, or (key, None) to
+    /// remove a variable from the child's environment.
+    pub async fn new_with_env(
+        codex_home: &Path,
+        env_overrides: &[(&str, Option<&str>)],
+    ) -> anyhow::Result<Self> {
+        // Use assert_cmd to locate the binary path and then switch to tokio::process::Command
+        let std_cmd = StdCommand::cargo_bin("codex-app-server")
+            .context("should find binary for codex-mcp-server")?;
+
+        let program = std_cmd.get_program().to_owned();
+
+        let mut cmd = Command::new(program);
+
+        cmd.stdin(Stdio::piped());
+        cmd.stdout(Stdio::piped());
+        cmd.stderr(Stdio::piped());
+        cmd.env("CODEX_HOME", codex_home);
+        cmd.env("RUST_LOG", "debug");
+
+        for (k, v) in env_overrides {
+            match v {
+                Some(val) => {
+                    cmd.env(k, val);
+                }
+                None => {
+                    cmd.env_remove(k);
+                }
+            }
+        }
+
+        let mut process = cmd
+            .kill_on_drop(true)
+            .spawn()
+            .context("codex-mcp-server proc should start")?;
+        let stdin = process
+            .stdin
+            .take()
+            .ok_or_else(|| anyhow::format_err!("mcp should have stdin fd"))?;
+        let stdout = process
+            .stdout
+            .take()
+            .ok_or_else(|| anyhow::format_err!("mcp should have stdout fd"))?;
+        let stdout = BufReader::new(stdout);
+
+        // Forward child's stderr to our stderr so failures are visible even
+        // when stdout/stderr are captured by the test harness.
+        if let Some(stderr) = process.stderr.take() {
+            let mut stderr_reader = BufReader::new(stderr).lines();
+            tokio::spawn(async move {
+                while let Ok(Some(line)) = stderr_reader.next_line().await {
+                    eprintln!("[mcp stderr] {line}");
+                }
+            });
+        }
+        Ok(Self {
+            next_request_id: AtomicI64::new(0),
+            process,
+            stdin,
+            stdout,
+            pending_user_messages: VecDeque::new(),
+        })
+    }
+
+    /// Performs the initialization handshake with the MCP server.
+    pub async fn initialize(&mut self) -> anyhow::Result<()> {
+        let params = Some(serde_json::to_value(InitializeParams {
+            client_info: ClientInfo {
+                name: "codex-app-server-tests".to_string(),
+                title: None,
+                version: "0.1.0".to_string(),
+            },
+        })?);
+        let req_id = self.send_request("initialize", params).await?;
+        let initialized = self.read_jsonrpc_message().await?;
+        let JSONRPCMessage::Response(response) = initialized else {
+            unreachable!("expected JSONRPCMessage::Response for initialize, got {initialized:?}");
+        };
+        if response.id != RequestId::Integer(req_id) {
+            anyhow::bail!(
+                "initialize response id mismatch: expected {}, got {:?}",
+                req_id,
+                response.id
+            );
+        }
+
+        // Send notifications/initialized to ack the response.
+        self.send_notification(ClientNotification::Initialized)
+            .await?;
+
+        Ok(())
+    }
+
+    /// Send a `newConversation` JSON-RPC request.
+    pub async fn send_new_conversation_request(
+        &mut self,
+        params: NewConversationParams,
+    ) -> anyhow::Result<i64> {
+        let params = Some(serde_json::to_value(params)?);
+        self.send_request("newConversation", params).await
+    }
+
+    /// Send an `archiveConversation` JSON-RPC request.
+    pub async fn send_archive_conversation_request(
+        &mut self,
+        params: ArchiveConversationParams,
+    ) -> anyhow::Result<i64> {
+        let params = Some(serde_json::to_value(params)?);
+        self.send_request("archiveConversation", params).await
+    }
+
+    /// Send an `addConversationListener` JSON-RPC request.
+    pub async fn send_add_conversation_listener_request(
+        &mut self,
+        params: AddConversationListenerParams,
+    ) -> anyhow::Result<i64> {
+        let params = Some(serde_json::to_value(params)?);
+        self.send_request("addConversationListener", params).await
+    }
+
+    /// Send a `sendUserMessage` JSON-RPC request with a single text item.
+    pub async fn send_send_user_message_request(
+        &mut self,
+        params: SendUserMessageParams,
+    ) -> anyhow::Result<i64> {
+        // Wire format expects variants in camelCase; text item uses external tagging.
+        let params = Some(serde_json::to_value(params)?);
+        self.send_request("sendUserMessage", params).await
+    }
+
+    /// Send a `removeConversationListener` JSON-RPC request.
+    pub async fn send_remove_conversation_listener_request(
+        &mut self,
+        params: RemoveConversationListenerParams,
+    ) -> anyhow::Result<i64> {
+        let params = Some(serde_json::to_value(params)?);
+        self.send_request("removeConversationListener", params)
+            .await
+    }
+
+    /// Send a `sendUserTurn` JSON-RPC request.
+    pub async fn send_send_user_turn_request(
+        &mut self,
+        params: SendUserTurnParams,
+    ) -> anyhow::Result<i64> {
+        let params = Some(serde_json::to_value(params)?);
+        self.send_request("sendUserTurn", params).await
+    }
+
+    /// Send a `interruptConversation` JSON-RPC request.
+    pub async fn send_interrupt_conversation_request(
+        &mut self,
+        params: InterruptConversationParams,
+    ) -> anyhow::Result<i64> {
+        let params = Some(serde_json::to_value(params)?);
+        self.send_request("interruptConversation", params).await
+    }
+
+    /// Send a `getAuthStatus` JSON-RPC request.
+    pub async fn send_get_auth_status_request(
+        &mut self,
+        params: GetAuthStatusParams,
+    ) -> anyhow::Result<i64> {
+        let params = Some(serde_json::to_value(params)?);
+        self.send_request("getAuthStatus", params).await
+    }
+
+    /// Send a `getUserSavedConfig` JSON-RPC request.
+    pub async fn send_get_user_saved_config_request(&mut self) -> anyhow::Result<i64> {
+        self.send_request("getUserSavedConfig", None).await
+    }
+
+    /// Send a `getUserAgent` JSON-RPC request.
+    pub async fn send_get_user_agent_request(&mut self) -> anyhow::Result<i64> {
+        self.send_request("getUserAgent", None).await
+    }
+
+    /// Send an `account/rateLimits/read` JSON-RPC request.
+    pub async fn send_get_account_rate_limits_request(&mut self) -> anyhow::Result<i64> {
+        self.send_request("account/rateLimits/read", None).await
+    }
+
+    /// Send a `feedback/upload` JSON-RPC request.
+    pub async fn send_upload_feedback_request(
+        &mut self,
+        params: UploadFeedbackParams,
+    ) -> anyhow::Result<i64> {
+        let params = Some(serde_json::to_value(params)?);
+        self.send_request("feedback/upload", params).await
+    }
+
+    /// Send a `userInfo` JSON-RPC request.
+    pub async fn send_user_info_request(&mut self) -> anyhow::Result<i64> {
+        self.send_request("userInfo", None).await
+    }
+
+    /// Send a `setDefaultModel` JSON-RPC request.
+    pub async fn send_set_default_model_request(
+        &mut self,
+        params: SetDefaultModelParams,
+    ) -> anyhow::Result<i64> {
+        let params = Some(serde_json::to_value(params)?);
+        self.send_request("setDefaultModel", params).await
+    }
+
+    /// Send a `listConversations` JSON-RPC request.
+    pub async fn send_list_conversations_request(
+        &mut self,
+        params: ListConversationsParams,
+    ) -> anyhow::Result<i64> {
+        let params = Some(serde_json::to_value(params)?);
+        self.send_request("listConversations", params).await
+    }
+
+    /// Send a `model/list` JSON-RPC request.
+    pub async fn send_list_models_request(
+        &mut self,
+        params: ListModelsParams,
+    ) -> anyhow::Result<i64> {
+        let params = Some(serde_json::to_value(params)?);
+        self.send_request("model/list", params).await
+    }
+
+    /// Send a `resumeConversation` JSON-RPC request.
+    pub async fn send_resume_conversation_request(
+        &mut self,
+        params: ResumeConversationParams,
+    ) -> anyhow::Result<i64> {
+        let params = Some(serde_json::to_value(params)?);
+        self.send_request("resumeConversation", params).await
+    }
+
+    /// Send a `loginApiKey` JSON-RPC request.
+    pub async fn send_login_api_key_request(
+        &mut self,
+        params: LoginApiKeyParams,
+    ) -> anyhow::Result<i64> {
+        let params = Some(serde_json::to_value(params)?);
+        self.send_request("loginApiKey", params).await
+    }
+
+    /// Send a `loginChatGpt` JSON-RPC request.
+    pub async fn send_login_chat_gpt_request(&mut self) -> anyhow::Result<i64> {
+        self.send_request("loginChatGpt", None).await
+    }
+
+    /// Send a `cancelLoginChatGpt` JSON-RPC request.
+    pub async fn send_cancel_login_chat_gpt_request(
+        &mut self,
+        params: CancelLoginChatGptParams,
+    ) -> anyhow::Result<i64> {
+        let params = Some(serde_json::to_value(params)?);
+        self.send_request("cancelLoginChatGpt", params).await
+    }
+
+    /// Send a `logoutChatGpt` JSON-RPC request.
+    pub async fn send_logout_chat_gpt_request(&mut self) -> anyhow::Result<i64> {
+        self.send_request("logoutChatGpt", None).await
+    }
+
+    /// Send a `fuzzyFileSearch` JSON-RPC request.
+    pub async fn send_fuzzy_file_search_request(
+        &mut self,
+        query: &str,
+        roots: Vec<String>,
+        cancellation_token: Option<String>,
+    ) -> anyhow::Result<i64> {
+        let mut params = serde_json::json!({
+            "query": query,
+            "roots": roots,
+        });
+        if let Some(token) = cancellation_token {
+            params["cancellationToken"] = serde_json::json!(token);
+        }
+        self.send_request("fuzzyFileSearch", Some(params)).await
+    }
+
+    async fn send_request(
+        &mut self,
+        method: &str,
+        params: Option<serde_json::Value>,
+    ) -> anyhow::Result<i64> {
+        let request_id = self.next_request_id.fetch_add(1, Ordering::Relaxed);
+
+        let message = JSONRPCMessage::Request(JSONRPCRequest {
+            id: RequestId::Integer(request_id),
+            method: method.to_string(),
+            params,
+        });
+        self.send_jsonrpc_message(message).await?;
+        Ok(request_id)
+    }
+
+    pub async fn send_response(
+        &mut self,
+        id: RequestId,
+        result: serde_json::Value,
+    ) -> anyhow::Result<()> {
+        self.send_jsonrpc_message(JSONRPCMessage::Response(JSONRPCResponse { id, result }))
+            .await
+    }
+
+    pub async fn send_notification(
+        &mut self,
+        notification: ClientNotification,
+    ) -> anyhow::Result<()> {
+        let value = serde_json::to_value(notification)?;
+        self.send_jsonrpc_message(JSONRPCMessage::Notification(JSONRPCNotification {
+            method: value
+                .get("method")
+                .and_then(|m| m.as_str())
+                .ok_or_else(|| anyhow::format_err!("notification missing method field"))?
+                .to_string(),
+            params: value.get("params").cloned(),
+        }))
+        .await
+    }
+
+    async fn send_jsonrpc_message(&mut self, message: JSONRPCMessage) -> anyhow::Result<()> {
+        eprintln!("writing message to stdin: {message:?}");
+        let payload = serde_json::to_string(&message)?;
+        self.stdin.write_all(payload.as_bytes()).await?;
+        self.stdin.write_all(b"\n").await?;
+        self.stdin.flush().await?;
+        Ok(())
+    }
+
+    async fn read_jsonrpc_message(&mut self) -> anyhow::Result<JSONRPCMessage> {
+        let mut line = String::new();
+        self.stdout.read_line(&mut line).await?;
+        let message = serde_json::from_str::<JSONRPCMessage>(&line)?;
+        eprintln!("read message from stdout: {message:?}");
+        Ok(message)
+    }
+
+    pub async fn read_stream_until_request_message(&mut self) -> anyhow::Result<ServerRequest> {
+        eprintln!("in read_stream_until_request_message()");
+
+        loop {
+            let message = self.read_jsonrpc_message().await?;
+
+            match message {
+                JSONRPCMessage::Notification(notification) => {
+                    eprintln!("notification: {notification:?}");
+                    self.enqueue_user_message(notification);
+                }
+                JSONRPCMessage::Request(jsonrpc_request) => {
+                    return jsonrpc_request.try_into().with_context(
+                        || "failed to deserialize ServerRequest from JSONRPCRequest",
+                    );
+                }
+                JSONRPCMessage::Error(_) => {
+                    anyhow::bail!("unexpected JSONRPCMessage::Error: {message:?}");
+                }
+                JSONRPCMessage::Response(_) => {
+                    anyhow::bail!("unexpected JSONRPCMessage::Response: {message:?}");
+                }
+            }
+        }
+    }
+
+    pub async fn read_stream_until_response_message(
+        &mut self,
+        request_id: RequestId,
+    ) -> anyhow::Result<JSONRPCResponse> {
+        eprintln!("in read_stream_until_response_message({request_id:?})");
+
+        loop {
+            let message = self.read_jsonrpc_message().await?;
+            match message {
+                JSONRPCMessage::Notification(notification) => {
+                    eprintln!("notification: {notification:?}");
+                    self.enqueue_user_message(notification);
+                }
+                JSONRPCMessage::Request(_) => {
+                    anyhow::bail!("unexpected JSONRPCMessage::Request: {message:?}");
+                }
+                JSONRPCMessage::Error(_) => {
+                    anyhow::bail!("unexpected JSONRPCMessage::Error: {message:?}");
+                }
+                JSONRPCMessage::Response(jsonrpc_response) => {
+                    if jsonrpc_response.id == request_id {
+                        return Ok(jsonrpc_response);
+                    }
+                }
+            }
+        }
+    }
+
+    pub async fn read_stream_until_error_message(
+        &mut self,
+        request_id: RequestId,
+    ) -> anyhow::Result<JSONRPCError> {
+        loop {
+            let message = self.read_jsonrpc_message().await?;
+            match message {
+                JSONRPCMessage::Notification(notification) => {
+                    eprintln!("notification: {notification:?}");
+                    self.enqueue_user_message(notification);
+                }
+                JSONRPCMessage::Request(_) => {
+                    anyhow::bail!("unexpected JSONRPCMessage::Request: {message:?}");
+                }
+                JSONRPCMessage::Response(_) => {
+                    // Keep scanning; we're waiting for an error with matching id.
+                }
+                JSONRPCMessage::Error(err) => {
+                    if err.id == request_id {
+                        return Ok(err);
+                    }
+                }
+            }
+        }
+    }
+
+    pub async fn read_stream_until_notification_message(
+        &mut self,
+        method: &str,
+    ) -> anyhow::Result<JSONRPCNotification> {
+        eprintln!("in read_stream_until_notification_message({method})");
+
+        if let Some(notification) = self.take_pending_notification_by_method(method) {
+            return Ok(notification);
+        }
+
+        loop {
+            let message = self.read_jsonrpc_message().await?;
+            match message {
+                JSONRPCMessage::Notification(notification) => {
+                    if notification.method == method {
+                        return Ok(notification);
+                    }
+                    self.enqueue_user_message(notification);
+                }
+                JSONRPCMessage::Request(_) => {
+                    anyhow::bail!("unexpected JSONRPCMessage::Request: {message:?}");
+                }
+                JSONRPCMessage::Error(_) => {
+                    anyhow::bail!("unexpected JSONRPCMessage::Error: {message:?}");
+                }
+                JSONRPCMessage::Response(_) => {
+                    anyhow::bail!("unexpected JSONRPCMessage::Response: {message:?}");
+                }
+            }
+        }
+    }
+
+    fn take_pending_notification_by_method(&mut self, method: &str) -> Option<JSONRPCNotification> {
+        if let Some(pos) = self
+            .pending_user_messages
+            .iter()
+            .position(|notification| notification.method == method)
+        {
+            return self.pending_user_messages.remove(pos);
+        }
+        None
+    }
+
+    fn enqueue_user_message(&mut self, notification: JSONRPCNotification) {
+        if notification.method == "codex/event/user_message" {
+            self.pending_user_messages.push_back(notification);
+        }
+    }
+}

codex-rs/app-server/tests/common/mock_model_server.rs

@@ -0,0 +1,47 @@
+use std::sync::atomic::AtomicUsize;
+use std::sync::atomic::Ordering;
+
+use wiremock::Mock;
+use wiremock::MockServer;
+use wiremock::Respond;
+use wiremock::ResponseTemplate;
+use wiremock::matchers::method;
+use wiremock::matchers::path;
+
+/// Create a mock server that will provide the responses, in order, for
+/// requests to the `/v1/chat/completions` endpoint.
+pub async fn create_mock_chat_completions_server(responses: Vec<String>) -> MockServer {
+    let server = MockServer::start().await;
+
+    let num_calls = responses.len();
+    let seq_responder = SeqResponder {
+        num_calls: AtomicUsize::new(0),
+        responses,
+    };
+
+    Mock::given(method("POST"))
+        .and(path("/v1/chat/completions"))
+        .respond_with(seq_responder)
+        .expect(num_calls as u64)
+        .mount(&server)
+        .await;
+
+    server
+}
+
+struct SeqResponder {
+    num_calls: AtomicUsize,
+    responses: Vec<String>,
+}
+
+impl Respond for SeqResponder {
+    fn respond(&self, _: &wiremock::Request) -> ResponseTemplate {
+        let call_num = self.num_calls.fetch_add(1, Ordering::SeqCst);
+        match self.responses.get(call_num) {
+            Some(response) => ResponseTemplate::new(200)
+                .insert_header("content-type", "text/event-stream")
+                .set_body_raw(response.clone(), "text/event-stream"),
+            None => panic!("no response for {call_num}"),
+        }
+    }
+}

codex-rs/app-server/tests/common/responses.rs

@@ -0,0 +1,95 @@
+use serde_json::json;
+use std::path::Path;
+
+pub fn create_shell_sse_response(
+    command: Vec<String>,
+    workdir: Option<&Path>,
+    timeout_ms: Option<u64>,
+    call_id: &str,
+) -> anyhow::Result<String> {
+    // The `arguments`` for the `shell` tool is a serialized JSON object.
+    let tool_call_arguments = serde_json::to_string(&json!({
+        "command": command,
+        "workdir": workdir.map(|w| w.to_string_lossy()),
+        "timeout": timeout_ms
+    }))?;
+    let tool_call = json!({
+        "choices": [
+            {
+                "delta": {
+                    "tool_calls": [
+                        {
+                            "id": call_id,
+                            "function": {
+                                "name": "shell",
+                                "arguments": tool_call_arguments
+                            }
+                        }
+                    ]
+                },
+                "finish_reason": "tool_calls"
+            }
+        ]
+    });
+
+    let sse = format!(
+        "data: {}\n\ndata: DONE\n\n",
+        serde_json::to_string(&tool_call)?
+    );
+    Ok(sse)
+}
+
+pub fn create_final_assistant_message_sse_response(message: &str) -> anyhow::Result<String> {
+    let assistant_message = json!({
+        "choices": [
+            {
+                "delta": {
+                    "content": message
+                },
+                "finish_reason": "stop"
+            }
+        ]
+    });
+
+    let sse = format!(
+        "data: {}\n\ndata: DONE\n\n",
+        serde_json::to_string(&assistant_message)?
+    );
+    Ok(sse)
+}
+
+pub fn create_apply_patch_sse_response(
+    patch_content: &str,
+    call_id: &str,
+) -> anyhow::Result<String> {
+    // Use shell command to call apply_patch with heredoc format
+    let shell_command = format!("apply_patch <<'EOF'\n{patch_content}\nEOF");
+    let tool_call_arguments = serde_json::to_string(&json!({
+        "command": ["bash", "-lc", shell_command]
+    }))?;
+
+    let tool_call = json!({
+        "choices": [
+            {
+                "delta": {
+                    "tool_calls": [
+                        {
+                            "id": call_id,
+                            "function": {
+                                "name": "shell",
+                                "arguments": tool_call_arguments
+                            }
+                        }
+                    ]
+                },
+                "finish_reason": "tool_calls"
+            }
+        ]
+    });
+
+    let sse = format!(
+        "data: {}\n\ndata: DONE\n\n",
+        serde_json::to_string(&tool_call)?
+    );
+    Ok(sse)
+}

codex-rs/app-server/tests/suite/archive_conversation.rs

@@ -1,53 +1,44 @@
-use std::path::Path;
-
+use anyhow::Result;
+use app_test_support::McpProcess;
+use app_test_support::to_response;
+use codex_app_server_protocol::ArchiveConversationParams;
+use codex_app_server_protocol::ArchiveConversationResponse;
+use codex_app_server_protocol::JSONRPCResponse;
+use codex_app_server_protocol::NewConversationParams;
+use codex_app_server_protocol::NewConversationResponse;
+use codex_app_server_protocol::RequestId;
 use codex_core::ARCHIVED_SESSIONS_SUBDIR;
-use codex_protocol::mcp_protocol::ArchiveConversationParams;
-use codex_protocol::mcp_protocol::ArchiveConversationResponse;
-use codex_protocol::mcp_protocol::NewConversationParams;
-use codex_protocol::mcp_protocol::NewConversationResponse;
-use mcp_test_support::McpProcess;
-use mcp_test_support::to_response;
-use mcp_types::JSONRPCResponse;
-use mcp_types::RequestId;
+use std::path::Path;
 use tempfile::TempDir;
 use tokio::time::timeout;
 
 const DEFAULT_READ_TIMEOUT: std::time::Duration = std::time::Duration::from_secs(10);
 
 #[tokio::test(flavor = "multi_thread", worker_threads = 2)]
-async fn archive_conversation_moves_rollout_into_archived_directory() {
-    let codex_home = TempDir::new().expect("create temp dir");
-    create_config_toml(codex_home.path()).expect("write config.toml");
+async fn archive_conversation_moves_rollout_into_archived_directory() -> Result<()> {
+    let codex_home = TempDir::new()?;
+    create_config_toml(codex_home.path())?;
 
-    let mut mcp = McpProcess::new(codex_home.path())
-        .await
-        .expect("spawn mcp process");
-    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize())
-        .await
-        .expect("initialize timeout")
-        .expect("initialize request");
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
 
     let new_request_id = mcp
         .send_new_conversation_request(NewConversationParams {
             model: Some("mock-model".to_string()),
             ..Default::default()
         })
-        .await
-        .expect("send newConversation");
+        .await?;
     let new_response: JSONRPCResponse = timeout(
         DEFAULT_READ_TIMEOUT,
         mcp.read_stream_until_response_message(RequestId::Integer(new_request_id)),
     )
-    .await
-    .expect("newConversation timeout")
-    .expect("newConversation response");
+    .await??;
 
     let NewConversationResponse {
         conversation_id,
         rollout_path,
         ..
-    } = to_response::<NewConversationResponse>(new_response)
-        .expect("deserialize newConversation response");
+    } = to_response::<NewConversationResponse>(new_response)?;
 
     assert!(
         rollout_path.exists(),
@@ -60,19 +51,15 @@ async fn archive_conversation_moves_rollout_into_archived_directory() {
             conversation_id,
             rollout_path: rollout_path.clone(),
         })
-        .await
-        .expect("send archiveConversation");
+        .await?;
     let archive_response: JSONRPCResponse = timeout(
         DEFAULT_READ_TIMEOUT,
         mcp.read_stream_until_response_message(RequestId::Integer(archive_request_id)),
     )
-    .await
-    .expect("archiveConversation timeout")
-    .expect("archiveConversation response");
+    .await??;
 
     let _: ArchiveConversationResponse =
-        to_response::<ArchiveConversationResponse>(archive_response)
-            .expect("deserialize archiveConversation response");
+        to_response::<ArchiveConversationResponse>(archive_response)?;
 
     let archived_directory = codex_home.path().join(ARCHIVED_SESSIONS_SUBDIR);
     let archived_rollout_path =
@@ -90,6 +77,8 @@ async fn archive_conversation_moves_rollout_into_archived_directory() {
         "expected archived rollout path {} to exist",
         archived_rollout_path.display()
     );
+
+    Ok(())
 }
 
 fn create_config_toml(codex_home: &Path) -> std::io::Result<()> {

codex-rs/app-server/tests/suite/auth.rs

@@ -1,15 +1,16 @@
-use std::path::Path;
-
-use codex_protocol::mcp_protocol::AuthMode;
-use codex_protocol::mcp_protocol::GetAuthStatusParams;
-use codex_protocol::mcp_protocol::GetAuthStatusResponse;
-use codex_protocol::mcp_protocol::LoginApiKeyParams;
-use codex_protocol::mcp_protocol::LoginApiKeyResponse;
-use mcp_test_support::McpProcess;
-use mcp_test_support::to_response;
-use mcp_types::JSONRPCResponse;
-use mcp_types::RequestId;
+use anyhow::Result;
+use app_test_support::McpProcess;
+use app_test_support::to_response;
+use codex_app_server_protocol::AuthMode;
+use codex_app_server_protocol::GetAuthStatusParams;
+use codex_app_server_protocol::GetAuthStatusResponse;
+use codex_app_server_protocol::JSONRPCError;
+use codex_app_server_protocol::JSONRPCResponse;
+use codex_app_server_protocol::LoginApiKeyParams;
+use codex_app_server_protocol::LoginApiKeyResponse;
+use codex_app_server_protocol::RequestId;
 use pretty_assertions::assert_eq;
+use std::path::Path;
 use tempfile::TempDir;
 use tokio::time::timeout;
 
@@ -57,125 +58,112 @@ sandbox_mode = "danger-full-access"
     )
 }
 
-async fn login_with_api_key_via_request(mcp: &mut McpProcess, api_key: &str) {
+fn create_config_toml_forced_login(codex_home: &Path, forced_method: &str) -> std::io::Result<()> {
+    let config_toml = codex_home.join("config.toml");
+    let contents = format!(
+        r#"
+model = "mock-model"
+approval_policy = "never"
+sandbox_mode = "danger-full-access"
+forced_login_method = "{forced_method}"
+"#
+    );
+    std::fs::write(config_toml, contents)
+}
+
+async fn login_with_api_key_via_request(mcp: &mut McpProcess, api_key: &str) -> Result<()> {
     let request_id = mcp
         .send_login_api_key_request(LoginApiKeyParams {
             api_key: api_key.to_string(),
         })
-        .await
-        .unwrap_or_else(|e| panic!("send loginApiKey: {e}"));
+        .await?;
 
     let resp: JSONRPCResponse = timeout(
         DEFAULT_READ_TIMEOUT,
         mcp.read_stream_until_response_message(RequestId::Integer(request_id)),
     )
-    .await
-    .unwrap_or_else(|e| panic!("loginApiKey timeout: {e}"))
-    .unwrap_or_else(|e| panic!("loginApiKey response: {e}"));
-    let _: LoginApiKeyResponse =
-        to_response(resp).unwrap_or_else(|e| panic!("deserialize login response: {e}"));
+    .await??;
+    let _: LoginApiKeyResponse = to_response(resp)?;
+    Ok(())
 }
 
 #[tokio::test(flavor = "multi_thread", worker_threads = 2)]
-async fn get_auth_status_no_auth() {
-    let codex_home = TempDir::new().unwrap_or_else(|e| panic!("create tempdir: {e}"));
-    create_config_toml(codex_home.path()).unwrap_or_else(|err| panic!("write config.toml: {err}"));
+async fn get_auth_status_no_auth() -> Result<()> {
+    let codex_home = TempDir::new()?;
+    create_config_toml(codex_home.path())?;
 
-    let mut mcp = McpProcess::new_with_env(codex_home.path(), &[("OPENAI_API_KEY", None)])
-        .await
-        .expect("spawn mcp process");
-    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize())
-        .await
-        .expect("init timeout")
-        .expect("init failed");
+    let mut mcp = McpProcess::new_with_env(codex_home.path(), &[("OPENAI_API_KEY", None)]).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
 
     let request_id = mcp
         .send_get_auth_status_request(GetAuthStatusParams {
             include_token: Some(true),
             refresh_token: Some(false),
         })
-        .await
-        .expect("send getAuthStatus");
+        .await?;
 
     let resp: JSONRPCResponse = timeout(
         DEFAULT_READ_TIMEOUT,
         mcp.read_stream_until_response_message(RequestId::Integer(request_id)),
     )
-    .await
-    .expect("getAuthStatus timeout")
-    .expect("getAuthStatus response");
-    let status: GetAuthStatusResponse = to_response(resp).expect("deserialize status");
+    .await??;
+    let status: GetAuthStatusResponse = to_response(resp)?;
     assert_eq!(status.auth_method, None, "expected no auth method");
     assert_eq!(status.auth_token, None, "expected no token");
+    Ok(())
 }
 
 #[tokio::test(flavor = "multi_thread", worker_threads = 2)]
-async fn get_auth_status_with_api_key() {
-    let codex_home = TempDir::new().unwrap_or_else(|e| panic!("create tempdir: {e}"));
-    create_config_toml(codex_home.path()).unwrap_or_else(|err| panic!("write config.toml: {err}"));
+async fn get_auth_status_with_api_key() -> Result<()> {
+    let codex_home = TempDir::new()?;
+    create_config_toml(codex_home.path())?;
 
-    let mut mcp = McpProcess::new(codex_home.path())
-        .await
-        .expect("spawn mcp process");
-    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize())
-        .await
-        .expect("init timeout")
-        .expect("init failed");
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
 
-    login_with_api_key_via_request(&mut mcp, "sk-test-key").await;
+    login_with_api_key_via_request(&mut mcp, "sk-test-key").await?;
 
     let request_id = mcp
         .send_get_auth_status_request(GetAuthStatusParams {
             include_token: Some(true),
             refresh_token: Some(false),
         })
-        .await
-        .expect("send getAuthStatus");
+        .await?;
 
     let resp: JSONRPCResponse = timeout(
         DEFAULT_READ_TIMEOUT,
         mcp.read_stream_until_response_message(RequestId::Integer(request_id)),
     )
-    .await
-    .expect("getAuthStatus timeout")
-    .expect("getAuthStatus response");
-    let status: GetAuthStatusResponse = to_response(resp).expect("deserialize status");
+    .await??;
+    let status: GetAuthStatusResponse = to_response(resp)?;
     assert_eq!(status.auth_method, Some(AuthMode::ApiKey));
     assert_eq!(status.auth_token, Some("sk-test-key".to_string()));
+    Ok(())
 }
 
 #[tokio::test(flavor = "multi_thread", worker_threads = 2)]
-async fn get_auth_status_with_api_key_when_auth_not_required() {
-    let codex_home = TempDir::new().unwrap_or_else(|e| panic!("create tempdir: {e}"));
-    create_config_toml_custom_provider(codex_home.path(), false)
-        .unwrap_or_else(|err| panic!("write config.toml: {err}"));
+async fn get_auth_status_with_api_key_when_auth_not_required() -> Result<()> {
+    let codex_home = TempDir::new()?;
+    create_config_toml_custom_provider(codex_home.path(), false)?;
 
-    let mut mcp = McpProcess::new(codex_home.path())
-        .await
-        .expect("spawn mcp process");
-    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize())
-        .await
-        .expect("init timeout")
-        .expect("init failed");
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
 
-    login_with_api_key_via_request(&mut mcp, "sk-test-key").await;
+    login_with_api_key_via_request(&mut mcp, "sk-test-key").await?;
 
     let request_id = mcp
         .send_get_auth_status_request(GetAuthStatusParams {
             include_token: Some(true),
             refresh_token: Some(false),
         })
-        .await
-        .expect("send getAuthStatus");
+        .await?;
 
     let resp: JSONRPCResponse = timeout(
         DEFAULT_READ_TIMEOUT,
         mcp.read_stream_until_response_message(RequestId::Integer(request_id)),
     )
-    .await
-    .expect("getAuthStatus timeout")
-    .expect("getAuthStatus response");
-    let status: GetAuthStatusResponse = to_response(resp).expect("deserialize status");
+    .await??;
+    let status: GetAuthStatusResponse = to_response(resp)?;
     assert_eq!(status.auth_method, None, "expected no auth method");
     assert_eq!(status.auth_token, None, "expected no token");
     assert_eq!(
@@ -183,41 +171,60 @@ async fn get_auth_status_with_api_key_when_auth_not_required() {
         Some(false),
         "requires_openai_auth should be false",
     );
+    Ok(())
 }
 
 #[tokio::test(flavor = "multi_thread", worker_threads = 2)]
-async fn get_auth_status_with_api_key_no_include_token() {
-    let codex_home = TempDir::new().unwrap_or_else(|e| panic!("create tempdir: {e}"));
-    create_config_toml(codex_home.path()).unwrap_or_else(|err| panic!("write config.toml: {err}"));
+async fn get_auth_status_with_api_key_no_include_token() -> Result<()> {
+    let codex_home = TempDir::new()?;
+    create_config_toml(codex_home.path())?;
 
-    let mut mcp = McpProcess::new(codex_home.path())
-        .await
-        .expect("spawn mcp process");
-    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize())
-        .await
-        .expect("init timeout")
-        .expect("init failed");
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
 
-    login_with_api_key_via_request(&mut mcp, "sk-test-key").await;
+    login_with_api_key_via_request(&mut mcp, "sk-test-key").await?;
 
     // Build params via struct so None field is omitted in wire JSON.
     let params = GetAuthStatusParams {
         include_token: None,
         refresh_token: Some(false),
     };
-    let request_id = mcp
-        .send_get_auth_status_request(params)
-        .await
-        .expect("send getAuthStatus");
+    let request_id = mcp.send_get_auth_status_request(params).await?;
 
     let resp: JSONRPCResponse = timeout(
         DEFAULT_READ_TIMEOUT,
         mcp.read_stream_until_response_message(RequestId::Integer(request_id)),
     )
-    .await
-    .expect("getAuthStatus timeout")
-    .expect("getAuthStatus response");
-    let status: GetAuthStatusResponse = to_response(resp).expect("deserialize status");
+    .await??;
+    let status: GetAuthStatusResponse = to_response(resp)?;
     assert_eq!(status.auth_method, Some(AuthMode::ApiKey));
     assert!(status.auth_token.is_none(), "token must be omitted");
+    Ok(())
+}
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
+async fn login_api_key_rejected_when_forced_chatgpt() -> Result<()> {
+    let codex_home = TempDir::new()?;
+    create_config_toml_forced_login(codex_home.path(), "chatgpt")?;
+
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
+
+    let request_id = mcp
+        .send_login_api_key_request(LoginApiKeyParams {
+            api_key: "sk-test-key".to_string(),
+        })
+        .await?;
+
+    let err: JSONRPCError = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_error_message(RequestId::Integer(request_id)),
+    )
+    .await??;
+
+    assert_eq!(
+        err.error.message,
+        "API key login is disabled. Use ChatGPT login instead."
+    );
+    Ok(())
 }

codex-rs/app-server/tests/suite/codex_message_processor_flow.rs

@@ -0,0 +1,523 @@
+use anyhow::Result;
+use app_test_support::McpProcess;
+use app_test_support::create_final_assistant_message_sse_response;
+use app_test_support::create_mock_chat_completions_server;
+use app_test_support::create_shell_sse_response;
+use app_test_support::to_response;
+use codex_app_server_protocol::AddConversationListenerParams;
+use codex_app_server_protocol::AddConversationSubscriptionResponse;
+use codex_app_server_protocol::ExecCommandApprovalParams;
+use codex_app_server_protocol::InputItem;
+use codex_app_server_protocol::JSONRPCNotification;
+use codex_app_server_protocol::JSONRPCResponse;
+use codex_app_server_protocol::NewConversationParams;
+use codex_app_server_protocol::NewConversationResponse;
+use codex_app_server_protocol::RemoveConversationListenerParams;
+use codex_app_server_protocol::RemoveConversationSubscriptionResponse;
+use codex_app_server_protocol::RequestId;
+use codex_app_server_protocol::SendUserMessageParams;
+use codex_app_server_protocol::SendUserMessageResponse;
+use codex_app_server_protocol::SendUserTurnParams;
+use codex_app_server_protocol::SendUserTurnResponse;
+use codex_app_server_protocol::ServerRequest;
+use codex_core::protocol::AskForApproval;
+use codex_core::protocol::SandboxPolicy;
+use codex_core::protocol_config_types::ReasoningEffort;
+use codex_core::protocol_config_types::ReasoningSummary;
+use codex_core::spawn::CODEX_SANDBOX_NETWORK_DISABLED_ENV_VAR;
+use codex_protocol::config_types::SandboxMode;
+use codex_protocol::parse_command::ParsedCommand;
+use codex_protocol::protocol::Event;
+use codex_protocol::protocol::EventMsg;
+use pretty_assertions::assert_eq;
+use std::env;
+use std::path::Path;
+use tempfile::TempDir;
+use tokio::time::timeout;
+
+const DEFAULT_READ_TIMEOUT: std::time::Duration = std::time::Duration::from_secs(10);
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 4)]
+async fn test_codex_jsonrpc_conversation_flow() -> Result<()> {
+    if env::var(CODEX_SANDBOX_NETWORK_DISABLED_ENV_VAR).is_ok() {
+        println!(
+            "Skipping test because it cannot execute when network is disabled in a Codex sandbox."
+        );
+        return Ok(());
+    }
+
+    let tmp = TempDir::new()?;
+    // Temporary Codex home with config pointing at the mock server.
+    let codex_home = tmp.path().join("codex_home");
+    std::fs::create_dir(&codex_home)?;
+    let working_directory = tmp.path().join("workdir");
+    std::fs::create_dir(&working_directory)?;
+
+    // Create a mock model server that immediately ends each turn.
+    // Two turns are expected: initial session configure + one user message.
+    let responses = vec![
+        create_shell_sse_response(
+            vec!["ls".to_string()],
+            Some(&working_directory),
+            Some(5000),
+            "call1234",
+        )?,
+        create_final_assistant_message_sse_response("Enjoy your new git repo!")?,
+    ];
+    let server = create_mock_chat_completions_server(responses).await;
+    create_config_toml(&codex_home, &server.uri())?;
+
+    // Start MCP server and initialize.
+    let mut mcp = McpProcess::new(&codex_home).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
+
+    // 1) newConversation
+    let new_conv_id = mcp
+        .send_new_conversation_request(NewConversationParams {
+            cwd: Some(working_directory.to_string_lossy().into_owned()),
+            ..Default::default()
+        })
+        .await?;
+    let new_conv_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(new_conv_id)),
+    )
+    .await??;
+    let new_conv_resp = to_response::<NewConversationResponse>(new_conv_resp)?;
+    let NewConversationResponse {
+        conversation_id,
+        model,
+        reasoning_effort: _,
+        rollout_path: _,
+    } = new_conv_resp;
+    assert_eq!(model, "mock-model");
+
+    // 2) addConversationListener
+    let add_listener_id = mcp
+        .send_add_conversation_listener_request(AddConversationListenerParams {
+            conversation_id,
+            experimental_raw_events: false,
+        })
+        .await?;
+    let add_listener_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(add_listener_id)),
+    )
+    .await??;
+    let AddConversationSubscriptionResponse { subscription_id } =
+        to_response::<AddConversationSubscriptionResponse>(add_listener_resp)?;
+
+    // 3) sendUserMessage (should trigger notifications; we only validate an OK response)
+    let send_user_id = mcp
+        .send_send_user_message_request(SendUserMessageParams {
+            conversation_id,
+            items: vec![codex_app_server_protocol::InputItem::Text {
+                text: "text".to_string(),
+            }],
+        })
+        .await?;
+    let send_user_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(send_user_id)),
+    )
+    .await??;
+    let SendUserMessageResponse {} = to_response::<SendUserMessageResponse>(send_user_resp)?;
+
+    // Verify the task_finished notification is received.
+    // Note this also ensures that the final request to the server was made.
+    let task_finished_notification: JSONRPCNotification = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("codex/event/task_complete"),
+    )
+    .await??;
+    let serde_json::Value::Object(map) = task_finished_notification
+        .params
+        .expect("notification should have params")
+    else {
+        panic!("task_finished_notification should have params");
+    };
+    assert_eq!(
+        map.get("conversationId")
+            .expect("should have conversationId"),
+        &serde_json::Value::String(conversation_id.to_string())
+    );
+
+    // 4) removeConversationListener
+    let remove_listener_id = mcp
+        .send_remove_conversation_listener_request(RemoveConversationListenerParams {
+            subscription_id,
+        })
+        .await?;
+    let remove_listener_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(remove_listener_id)),
+    )
+    .await??;
+    let RemoveConversationSubscriptionResponse {} = to_response(remove_listener_resp)?;
+
+    Ok(())
+}
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 4)]
+async fn test_send_user_turn_changes_approval_policy_behavior() -> Result<()> {
+    if env::var(CODEX_SANDBOX_NETWORK_DISABLED_ENV_VAR).is_ok() {
+        println!(
+            "Skipping test because it cannot execute when network is disabled in a Codex sandbox."
+        );
+        return Ok(());
+    }
+
+    let tmp = TempDir::new()?;
+    let codex_home = tmp.path().join("codex_home");
+    std::fs::create_dir(&codex_home)?;
+    let working_directory = tmp.path().join("workdir");
+    std::fs::create_dir(&working_directory)?;
+
+    // Mock server will request a python shell call for the first and second turn, then finish.
+    let responses = vec![
+        create_shell_sse_response(
+            vec![
+                "python3".to_string(),
+                "-c".to_string(),
+                "print(42)".to_string(),
+            ],
+            Some(&working_directory),
+            Some(5000),
+            "call1",
+        )?,
+        create_final_assistant_message_sse_response("done 1")?,
+        create_shell_sse_response(
+            vec![
+                "python3".to_string(),
+                "-c".to_string(),
+                "print(42)".to_string(),
+            ],
+            Some(&working_directory),
+            Some(5000),
+            "call2",
+        )?,
+        create_final_assistant_message_sse_response("done 2")?,
+    ];
+    let server = create_mock_chat_completions_server(responses).await;
+    create_config_toml(&codex_home, &server.uri())?;
+
+    // Start MCP server and initialize.
+    let mut mcp = McpProcess::new(&codex_home).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
+
+    // 1) Start conversation with approval_policy=untrusted
+    let new_conv_id = mcp
+        .send_new_conversation_request(NewConversationParams {
+            cwd: Some(working_directory.to_string_lossy().into_owned()),
+            ..Default::default()
+        })
+        .await?;
+    let new_conv_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(new_conv_id)),
+    )
+    .await??;
+    let NewConversationResponse {
+        conversation_id, ..
+    } = to_response::<NewConversationResponse>(new_conv_resp)?;
+
+    // 2) addConversationListener
+    let add_listener_id = mcp
+        .send_add_conversation_listener_request(AddConversationListenerParams {
+            conversation_id,
+            experimental_raw_events: false,
+        })
+        .await?;
+    let _: AddConversationSubscriptionResponse = to_response::<AddConversationSubscriptionResponse>(
+        timeout(
+            DEFAULT_READ_TIMEOUT,
+            mcp.read_stream_until_response_message(RequestId::Integer(add_listener_id)),
+        )
+        .await??,
+    )?;
+
+    // 3) sendUserMessage triggers a shell call; approval policy is Untrusted so we should get an elicitation
+    let send_user_id = mcp
+        .send_send_user_message_request(SendUserMessageParams {
+            conversation_id,
+            items: vec![codex_app_server_protocol::InputItem::Text {
+                text: "run python".to_string(),
+            }],
+        })
+        .await?;
+    let _send_user_resp: SendUserMessageResponse = to_response::<SendUserMessageResponse>(
+        timeout(
+            DEFAULT_READ_TIMEOUT,
+            mcp.read_stream_until_response_message(RequestId::Integer(send_user_id)),
+        )
+        .await??,
+    )?;
+
+    // Expect an ExecCommandApproval request (elicitation)
+    let request = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_request_message(),
+    )
+    .await??;
+    let ServerRequest::ExecCommandApproval { request_id, params } = request else {
+        panic!("expected ExecCommandApproval request, got: {request:?}");
+    };
+
+    assert_eq!(
+        ExecCommandApprovalParams {
+            conversation_id,
+            call_id: "call1".to_string(),
+            command: vec![
+                "python3".to_string(),
+                "-c".to_string(),
+                "print(42)".to_string(),
+            ],
+            cwd: working_directory.clone(),
+            reason: None,
+            risk: None,
+            parsed_cmd: vec![ParsedCommand::Unknown {
+                cmd: "python3 -c 'print(42)'".to_string()
+            }],
+        },
+        params
+    );
+
+    // Approve so the first turn can complete
+    mcp.send_response(
+        request_id,
+        serde_json::json!({ "decision": codex_core::protocol::ReviewDecision::Approved }),
+    )
+    .await?;
+
+    // Wait for first TaskComplete
+    let _ = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("codex/event/task_complete"),
+    )
+    .await??;
+
+    // 4) sendUserTurn with approval_policy=never should run without elicitation
+    let send_turn_id = mcp
+        .send_send_user_turn_request(SendUserTurnParams {
+            conversation_id,
+            items: vec![codex_app_server_protocol::InputItem::Text {
+                text: "run python again".to_string(),
+            }],
+            cwd: working_directory.clone(),
+            approval_policy: AskForApproval::Never,
+            sandbox_policy: SandboxPolicy::new_read_only_policy(),
+            model: "mock-model".to_string(),
+            effort: Some(ReasoningEffort::Medium),
+            summary: ReasoningSummary::Auto,
+        })
+        .await?;
+    // Acknowledge sendUserTurn
+    let _send_turn_resp: SendUserTurnResponse = to_response::<SendUserTurnResponse>(
+        timeout(
+            DEFAULT_READ_TIMEOUT,
+            mcp.read_stream_until_response_message(RequestId::Integer(send_turn_id)),
+        )
+        .await??,
+    )?;
+
+    // Ensure we do NOT receive an ExecCommandApproval request before the task completes.
+    // If any Request is seen while waiting for task_complete, the helper will error and the test fails.
+    let _ = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("codex/event/task_complete"),
+    )
+    .await??;
+
+    Ok(())
+}
+
+// Helper: minimal config.toml pointing at mock provider.
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 4)]
+async fn test_send_user_turn_updates_sandbox_and_cwd_between_turns() -> Result<()> {
+    if env::var(CODEX_SANDBOX_NETWORK_DISABLED_ENV_VAR).is_ok() {
+        println!(
+            "Skipping test because it cannot execute when network is disabled in a Codex sandbox."
+        );
+        return Ok(());
+    }
+
+    let tmp = TempDir::new()?;
+    let codex_home = tmp.path().join("codex_home");
+    std::fs::create_dir(&codex_home)?;
+    let workspace_root = tmp.path().join("workspace");
+    std::fs::create_dir(&workspace_root)?;
+    let first_cwd = workspace_root.join("turn1");
+    let second_cwd = workspace_root.join("turn2");
+    std::fs::create_dir(&first_cwd)?;
+    std::fs::create_dir(&second_cwd)?;
+
+    let responses = vec![
+        create_shell_sse_response(
+            vec![
+                "bash".to_string(),
+                "-lc".to_string(),
+                "echo first turn".to_string(),
+            ],
+            None,
+            Some(5000),
+            "call-first",
+        )?,
+        create_final_assistant_message_sse_response("done first")?,
+        create_shell_sse_response(
+            vec![
+                "bash".to_string(),
+                "-lc".to_string(),
+                "echo second turn".to_string(),
+            ],
+            None,
+            Some(5000),
+            "call-second",
+        )?,
+        create_final_assistant_message_sse_response("done second")?,
+    ];
+    let server = create_mock_chat_completions_server(responses).await;
+    create_config_toml(&codex_home, &server.uri())?;
+
+    let mut mcp = McpProcess::new(&codex_home).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
+
+    let new_conv_id = mcp
+        .send_new_conversation_request(NewConversationParams {
+            cwd: Some(first_cwd.to_string_lossy().into_owned()),
+            approval_policy: Some(AskForApproval::Never),
+            sandbox: Some(SandboxMode::WorkspaceWrite),
+            ..Default::default()
+        })
+        .await?;
+    let new_conv_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(new_conv_id)),
+    )
+    .await??;
+    let NewConversationResponse {
+        conversation_id,
+        model,
+        ..
+    } = to_response::<NewConversationResponse>(new_conv_resp)?;
+
+    let add_listener_id = mcp
+        .send_add_conversation_listener_request(AddConversationListenerParams {
+            conversation_id,
+            experimental_raw_events: false,
+        })
+        .await?;
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(add_listener_id)),
+    )
+    .await??;
+
+    let first_turn_id = mcp
+        .send_send_user_turn_request(SendUserTurnParams {
+            conversation_id,
+            items: vec![InputItem::Text {
+                text: "first turn".to_string(),
+            }],
+            cwd: first_cwd.clone(),
+            approval_policy: AskForApproval::Never,
+            sandbox_policy: SandboxPolicy::WorkspaceWrite {
+                writable_roots: vec![first_cwd.clone()],
+                network_access: false,
+                exclude_tmpdir_env_var: false,
+                exclude_slash_tmp: false,
+            },
+            model: model.clone(),
+            effort: Some(ReasoningEffort::Medium),
+            summary: ReasoningSummary::Auto,
+        })
+        .await?;
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(first_turn_id)),
+    )
+    .await??;
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("codex/event/task_complete"),
+    )
+    .await??;
+
+    let second_turn_id = mcp
+        .send_send_user_turn_request(SendUserTurnParams {
+            conversation_id,
+            items: vec![InputItem::Text {
+                text: "second turn".to_string(),
+            }],
+            cwd: second_cwd.clone(),
+            approval_policy: AskForApproval::Never,
+            sandbox_policy: SandboxPolicy::DangerFullAccess,
+            model: model.clone(),
+            effort: Some(ReasoningEffort::Medium),
+            summary: ReasoningSummary::Auto,
+        })
+        .await?;
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(second_turn_id)),
+    )
+    .await??;
+
+    let exec_begin_notification = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("codex/event/exec_command_begin"),
+    )
+    .await??;
+    let params = exec_begin_notification
+        .params
+        .clone()
+        .expect("exec_command_begin params");
+    let event: Event = serde_json::from_value(params).expect("deserialize exec begin event");
+    let exec_begin = match event.msg {
+        EventMsg::ExecCommandBegin(exec_begin) => exec_begin,
+        other => panic!("expected ExecCommandBegin event, got {other:?}"),
+    };
+    assert_eq!(
+        exec_begin.cwd, second_cwd,
+        "exec turn should run from updated cwd"
+    );
+    assert_eq!(
+        exec_begin.command,
+        vec![
+            "bash".to_string(),
+            "-lc".to_string(),
+            "echo second turn".to_string()
+        ],
+        "exec turn should run expected command"
+    );
+
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("codex/event/task_complete"),
+    )
+    .await??;
+
+    Ok(())
+}
+
+fn create_config_toml(codex_home: &Path, server_uri: &str) -> std::io::Result<()> {
+    let config_toml = codex_home.join("config.toml");
+    std::fs::write(
+        config_toml,
+        format!(
+            r#"
+model = "mock-model"
+approval_policy = "untrusted"
+
+model_provider = "mock_provider"
+
+[model_providers.mock_provider]
+name = "Mock provider for test"
+base_url = "{server_uri}/v1"
+wire_api = "chat"
+request_max_retries = 0
+stream_max_retries = 0
+"#
+        ),
+    )
+}

codex-rs/app-server/tests/suite/config.rs

@@ -1,21 +1,22 @@
-use std::collections::HashMap;
-use std::path::Path;
-
+use anyhow::Result;
+use app_test_support::McpProcess;
+use app_test_support::to_response;
+use codex_app_server_protocol::GetUserSavedConfigResponse;
+use codex_app_server_protocol::JSONRPCResponse;
+use codex_app_server_protocol::Profile;
+use codex_app_server_protocol::RequestId;
+use codex_app_server_protocol::SandboxSettings;
+use codex_app_server_protocol::Tools;
+use codex_app_server_protocol::UserSavedConfig;
 use codex_core::protocol::AskForApproval;
+use codex_protocol::config_types::ForcedLoginMethod;
 use codex_protocol::config_types::ReasoningEffort;
 use codex_protocol::config_types::ReasoningSummary;
 use codex_protocol::config_types::SandboxMode;
 use codex_protocol::config_types::Verbosity;
-use codex_protocol::mcp_protocol::GetUserSavedConfigResponse;
-use codex_protocol::mcp_protocol::Profile;
-use codex_protocol::mcp_protocol::SandboxSettings;
-use codex_protocol::mcp_protocol::Tools;
-use codex_protocol::mcp_protocol::UserSavedConfig;
-use mcp_test_support::McpProcess;
-use mcp_test_support::to_response;
-use mcp_types::JSONRPCResponse;
-use mcp_types::RequestId;
 use pretty_assertions::assert_eq;
+use std::collections::HashMap;
+use std::path::Path;
 use tempfile::TempDir;
 use tokio::time::timeout;
 
@@ -33,6 +34,8 @@ model_reasoning_summary = "detailed"
 model_reasoning_effort = "high"
 model_verbosity = "medium"
 profile = "test"
+forced_chatgpt_workspace_id = "12345678-0000-0000-0000-000000000000"
+forced_login_method = "chatgpt"
 
 [sandbox_workspace_write]
 writable_roots = ["/tmp"]
@@ -57,31 +60,21 @@ chatgpt_base_url = "https://api.chatgpt.com"
 }
 
 #[tokio::test(flavor = "multi_thread", worker_threads = 4)]
-async fn get_config_toml_parses_all_fields() {
-    let codex_home = TempDir::new().unwrap_or_else(|e| panic!("create tempdir: {e}"));
-    create_config_toml(codex_home.path()).expect("write config.toml");
+async fn get_config_toml_parses_all_fields() -> Result<()> {
+    let codex_home = TempDir::new()?;
+    create_config_toml(codex_home.path())?;
 
-    let mut mcp = McpProcess::new(codex_home.path())
-        .await
-        .expect("spawn mcp process");
-    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize())
-        .await
-        .expect("init timeout")
-        .expect("init failed");
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
 
-    let request_id = mcp
-        .send_get_user_saved_config_request()
-        .await
-        .expect("send getUserSavedConfig");
+    let request_id = mcp.send_get_user_saved_config_request().await?;
     let resp: JSONRPCResponse = timeout(
         DEFAULT_READ_TIMEOUT,
         mcp.read_stream_until_response_message(RequestId::Integer(request_id)),
     )
-    .await
-    .expect("getUserSavedConfig timeout")
-    .expect("getUserSavedConfig response");
+    .await??;
 
-    let config: GetUserSavedConfigResponse = to_response(resp).expect("deserialize config");
+    let config: GetUserSavedConfigResponse = to_response(resp)?;
     let expected = GetUserSavedConfigResponse {
         config: UserSavedConfig {
             approval_policy: Some(AskForApproval::OnRequest),
@@ -92,6 +85,8 @@ async fn get_config_toml_parses_all_fields() {
                 exclude_tmpdir_env_var: Some(true),
                 exclude_slash_tmp: Some(true),
             }),
+            forced_chatgpt_workspace_id: Some("12345678-0000-0000-0000-000000000000".into()),
+            forced_login_method: Some(ForcedLoginMethod::Chatgpt),
             model: Some("gpt-5-codex".into()),
             model_reasoning_effort: Some(ReasoningEffort::High),
             model_reasoning_summary: Some(ReasoningSummary::Detailed),
@@ -117,38 +112,31 @@ async fn get_config_toml_parses_all_fields() {
     };
 
     assert_eq!(config, expected);
+    Ok(())
 }
 
 #[tokio::test(flavor = "multi_thread", worker_threads = 2)]
-async fn get_config_toml_empty() {
-    let codex_home = TempDir::new().unwrap_or_else(|e| panic!("create tempdir: {e}"));
+async fn get_config_toml_empty() -> Result<()> {
+    let codex_home = TempDir::new()?;
 
-    let mut mcp = McpProcess::new(codex_home.path())
-        .await
-        .expect("spawn mcp process");
-    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize())
-        .await
-        .expect("init timeout")
-        .expect("init failed");
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
 
-    let request_id = mcp
-        .send_get_user_saved_config_request()
-        .await
-        .expect("send getUserSavedConfig");
+    let request_id = mcp.send_get_user_saved_config_request().await?;
     let resp: JSONRPCResponse = timeout(
         DEFAULT_READ_TIMEOUT,
         mcp.read_stream_until_response_message(RequestId::Integer(request_id)),
     )
-    .await
-    .expect("getUserSavedConfig timeout")
-    .expect("getUserSavedConfig response");
+    .await??;
 
-    let config: GetUserSavedConfigResponse = to_response(resp).expect("deserialize config");
+    let config: GetUserSavedConfigResponse = to_response(resp)?;
     let expected = GetUserSavedConfigResponse {
         config: UserSavedConfig {
             approval_policy: None,
             sandbox_mode: None,
             sandbox_settings: None,
+            forced_chatgpt_workspace_id: None,
+            forced_login_method: None,
             model: None,
             model_reasoning_effort: None,
             model_reasoning_summary: None,
@@ -160,4 +148,5 @@ async fn get_config_toml_empty() {
     };
 
     assert_eq!(config, expected);
+    Ok(())
 }

codex-rs/app-server/tests/suite/create_conversation.rs

@@ -1,45 +1,38 @@
-use std::path::Path;
-
-use codex_protocol::mcp_protocol::AddConversationListenerParams;
-use codex_protocol::mcp_protocol::AddConversationSubscriptionResponse;
-use codex_protocol::mcp_protocol::InputItem;
-use codex_protocol::mcp_protocol::NewConversationParams;
-use codex_protocol::mcp_protocol::NewConversationResponse;
-use codex_protocol::mcp_protocol::SendUserMessageParams;
-use codex_protocol::mcp_protocol::SendUserMessageResponse;
-use mcp_test_support::McpProcess;
-use mcp_test_support::create_final_assistant_message_sse_response;
-use mcp_test_support::create_mock_chat_completions_server;
-use mcp_test_support::to_response;
-use mcp_types::JSONRPCResponse;
-use mcp_types::RequestId;
+use anyhow::Result;
+use app_test_support::McpProcess;
+use app_test_support::create_final_assistant_message_sse_response;
+use app_test_support::create_mock_chat_completions_server;
+use app_test_support::to_response;
+use codex_app_server_protocol::AddConversationListenerParams;
+use codex_app_server_protocol::AddConversationSubscriptionResponse;
+use codex_app_server_protocol::InputItem;
+use codex_app_server_protocol::JSONRPCResponse;
+use codex_app_server_protocol::NewConversationParams;
+use codex_app_server_protocol::NewConversationResponse;
+use codex_app_server_protocol::RequestId;
+use codex_app_server_protocol::SendUserMessageParams;
+use codex_app_server_protocol::SendUserMessageResponse;
 use pretty_assertions::assert_eq;
 use serde_json::json;
+use std::path::Path;
 use tempfile::TempDir;
 use tokio::time::timeout;
 
 const DEFAULT_READ_TIMEOUT: std::time::Duration = std::time::Duration::from_secs(10);
 
 #[tokio::test(flavor = "multi_thread", worker_threads = 2)]
-async fn test_conversation_create_and_send_message_ok() {
+async fn test_conversation_create_and_send_message_ok() -> Result<()> {
     // Mock server – we won't strictly rely on it, but provide one to satisfy any model wiring.
-    let responses = vec![
-        create_final_assistant_message_sse_response("Done").expect("build mock assistant message"),
-    ];
+    let responses = vec![create_final_assistant_message_sse_response("Done")?];
     let server = create_mock_chat_completions_server(responses).await;
 
     // Temporary Codex home with config pointing at the mock server.
-    let codex_home = TempDir::new().expect("create temp dir");
-    create_config_toml(codex_home.path(), &server.uri()).expect("write config.toml");
+    let codex_home = TempDir::new()?;
+    create_config_toml(codex_home.path(), &server.uri())?;
 
     // Start MCP server process and initialize.
-    let mut mcp = McpProcess::new(codex_home.path())
-        .await
-        .expect("spawn mcp process");
-    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize())
-        .await
-        .expect("init timeout")
-        .expect("init failed");
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
 
     // Create a conversation via the new JSON-RPC API.
     let new_conv_id = mcp
@@ -47,40 +40,35 @@ async fn test_conversation_create_and_send_message_ok() {
             model: Some("o3".to_string()),
             ..Default::default()
         })
-        .await
-        .expect("send newConversation");
+        .await?;
     let new_conv_resp: JSONRPCResponse = timeout(
         DEFAULT_READ_TIMEOUT,
         mcp.read_stream_until_response_message(RequestId::Integer(new_conv_id)),
     )
-    .await
-    .expect("newConversation timeout")
-    .expect("newConversation resp");
+    .await??;
     let NewConversationResponse {
         conversation_id,
         model,
         reasoning_effort: _,
         rollout_path: _,
-    } = to_response::<NewConversationResponse>(new_conv_resp)
-        .expect("deserialize newConversation response");
+    } = to_response::<NewConversationResponse>(new_conv_resp)?;
     assert_eq!(model, "o3");
 
     // Add a listener so we receive notifications for this conversation (not strictly required for this test).
     let add_listener_id = mcp
-        .send_add_conversation_listener_request(AddConversationListenerParams { conversation_id })
-        .await
-        .expect("send addConversationListener");
+        .send_add_conversation_listener_request(AddConversationListenerParams {
+            conversation_id,
+            experimental_raw_events: false,
+        })
+        .await?;
     let _sub: AddConversationSubscriptionResponse =
         to_response::<AddConversationSubscriptionResponse>(
             timeout(
                 DEFAULT_READ_TIMEOUT,
                 mcp.read_stream_until_response_message(RequestId::Integer(add_listener_id)),
             )
-            .await
-            .expect("addConversationListener timeout")
-            .expect("addConversationListener resp"),
-        )
-        .expect("deserialize addConversationListener response");
+            .await??,
+        )?;
 
     // Now send a user message via the wire API and expect an OK (empty object) result.
     let send_id = mcp
@@ -90,36 +78,32 @@ async fn test_conversation_create_and_send_message_ok() {
                 text: "Hello".to_string(),
             }],
         })
-        .await
-        .expect("send sendUserMessage");
+        .await?;
     let send_resp: JSONRPCResponse = timeout(
         DEFAULT_READ_TIMEOUT,
         mcp.read_stream_until_response_message(RequestId::Integer(send_id)),
     )
-    .await
-    .expect("sendUserMessage timeout")
-    .expect("sendUserMessage resp");
-    let _ok: SendUserMessageResponse = to_response::<SendUserMessageResponse>(send_resp)
-        .expect("deserialize sendUserMessage response");
+    .await??;
+    let _ok: SendUserMessageResponse = to_response::<SendUserMessageResponse>(send_resp)?;
 
     // avoid race condition by waiting for the mock server to receive the chat.completions request
     let deadline = std::time::Instant::now() + DEFAULT_READ_TIMEOUT;
-    loop {
+    let requests = loop {
         let requests = server.received_requests().await.unwrap_or_default();
         if !requests.is_empty() {
-            break;
+            break requests;
         }
         if std::time::Instant::now() >= deadline {
             panic!("mock server did not receive the chat.completions request in time");
         }
         tokio::time::sleep(std::time::Duration::from_millis(10)).await;
-    }
+    };
 
     // Verify the outbound request body matches expectations for Chat Completions.
-    let request = &server.received_requests().await.unwrap()[0];
-    let body = request
-        .body_json::<serde_json::Value>()
-        .expect("parse request body as JSON");
+    let request = requests
+        .first()
+        .expect("mock server should have received at least one request");
+    let body = request.body_json::<serde_json::Value>()?;
     assert_eq!(body["model"], json!("o3"));
     assert!(body["stream"].as_bool().unwrap_or(false));
     let messages = body["messages"]
@@ -130,6 +114,7 @@ async fn test_conversation_create_and_send_message_ok() {
     assert_eq!(last["content"], json!("Hello"));
 
     drop(server);
+    Ok(())
 }
 
 // Helper to create a config.toml pointing at the mock model server.

codex-rs/app-server/tests/suite/fuzzy_file_search.rs

@@ -0,0 +1,128 @@
+use anyhow::Result;
+use anyhow::anyhow;
+use app_test_support::McpProcess;
+use codex_app_server_protocol::JSONRPCResponse;
+use codex_app_server_protocol::RequestId;
+use pretty_assertions::assert_eq;
+use serde_json::json;
+use tempfile::TempDir;
+use tokio::time::timeout;
+
+const DEFAULT_READ_TIMEOUT: std::time::Duration = std::time::Duration::from_secs(10);
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
+async fn test_fuzzy_file_search_sorts_and_includes_indices() -> Result<()> {
+    // Prepare a temporary Codex home and a separate root with test files.
+    let codex_home = TempDir::new()?;
+    let root = TempDir::new()?;
+
+    // Create files designed to have deterministic ordering for query "abe".
+    std::fs::write(root.path().join("abc"), "x")?;
+    std::fs::write(root.path().join("abcde"), "x")?;
+    std::fs::write(root.path().join("abexy"), "x")?;
+    std::fs::write(root.path().join("zzz.txt"), "x")?;
+    let sub_dir = root.path().join("sub");
+    std::fs::create_dir_all(&sub_dir)?;
+    let sub_abce_path = sub_dir.join("abce");
+    std::fs::write(&sub_abce_path, "x")?;
+    let sub_abce_rel = sub_abce_path
+        .strip_prefix(root.path())?
+        .to_string_lossy()
+        .to_string();
+
+    // Start MCP server and initialize.
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
+
+    let root_path = root.path().to_string_lossy().to_string();
+    // Send fuzzyFileSearch request.
+    let request_id = mcp
+        .send_fuzzy_file_search_request("abe", vec![root_path.clone()], None)
+        .await?;
+
+    // Read response and verify shape and ordering.
+    let resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(request_id)),
+    )
+    .await??;
+
+    let value = resp.result;
+    // The path separator on Windows affects the score.
+    let expected_score = if cfg!(windows) { 69 } else { 72 };
+
+    assert_eq!(
+        value,
+        json!({
+            "files": [
+                {
+                    "root": root_path.clone(),
+                    "path": "abexy",
+                    "file_name": "abexy",
+                    "score": 88,
+                    "indices": [0, 1, 2],
+                },
+                {
+                    "root": root_path.clone(),
+                    "path": "abcde",
+                    "file_name": "abcde",
+                    "score": 74,
+                    "indices": [0, 1, 4],
+                },
+                {
+                    "root": root_path.clone(),
+                    "path": sub_abce_rel,
+                    "file_name": "abce",
+                    "score": expected_score,
+                    "indices": [4, 5, 7],
+                },
+            ]
+        })
+    );
+
+    Ok(())
+}
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
+async fn test_fuzzy_file_search_accepts_cancellation_token() -> Result<()> {
+    let codex_home = TempDir::new()?;
+    let root = TempDir::new()?;
+
+    std::fs::write(root.path().join("alpha.txt"), "contents")?;
+
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
+
+    let root_path = root.path().to_string_lossy().to_string();
+    let request_id = mcp
+        .send_fuzzy_file_search_request("alp", vec![root_path.clone()], None)
+        .await?;
+
+    let request_id_2 = mcp
+        .send_fuzzy_file_search_request(
+            "alp",
+            vec![root_path.clone()],
+            Some(request_id.to_string()),
+        )
+        .await?;
+
+    let resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(request_id_2)),
+    )
+    .await??;
+
+    let files = resp
+        .result
+        .get("files")
+        .ok_or_else(|| anyhow!("files key missing"))?
+        .as_array()
+        .ok_or_else(|| anyhow!("files not array"))?
+        .clone();
+
+    assert_eq!(files.len(), 1);
+    assert_eq!(files[0]["root"], root_path);
+    assert_eq!(files[0]["path"], "alpha.txt");
+
+    Ok(())
+}

codex-rs/app-server/tests/suite/interrupt.rs

@@ -1,26 +1,26 @@
 #![cfg(unix)]
-// Support code lives in the `mcp_test_support` crate under tests/common.
+// Support code lives in the `app_test_support` crate under tests/common.
 
 use std::path::Path;
 
+use codex_app_server_protocol::AddConversationListenerParams;
+use codex_app_server_protocol::InterruptConversationParams;
+use codex_app_server_protocol::InterruptConversationResponse;
+use codex_app_server_protocol::JSONRPCResponse;
+use codex_app_server_protocol::NewConversationParams;
+use codex_app_server_protocol::NewConversationResponse;
+use codex_app_server_protocol::RequestId;
+use codex_app_server_protocol::SendUserMessageParams;
+use codex_app_server_protocol::SendUserMessageResponse;
 use codex_core::protocol::TurnAbortReason;
-use codex_protocol::mcp_protocol::AddConversationListenerParams;
-use codex_protocol::mcp_protocol::InterruptConversationParams;
-use codex_protocol::mcp_protocol::InterruptConversationResponse;
-use codex_protocol::mcp_protocol::NewConversationParams;
-use codex_protocol::mcp_protocol::NewConversationResponse;
-use codex_protocol::mcp_protocol::SendUserMessageParams;
-use codex_protocol::mcp_protocol::SendUserMessageResponse;
 use core_test_support::skip_if_no_network;
-use mcp_types::JSONRPCResponse;
-use mcp_types::RequestId;
 use tempfile::TempDir;
 use tokio::time::timeout;
 
-use mcp_test_support::McpProcess;
-use mcp_test_support::create_mock_chat_completions_server;
-use mcp_test_support::create_shell_sse_response;
-use mcp_test_support::to_response;
+use app_test_support::McpProcess;
+use app_test_support::create_mock_chat_completions_server;
+use app_test_support::create_shell_sse_response;
+use app_test_support::to_response;
 
 const DEFAULT_READ_TIMEOUT: std::time::Duration = std::time::Duration::from_secs(10);
 
@@ -88,7 +88,10 @@ async fn shell_command_interruption() -> anyhow::Result<()> {
 
     // 2) addConversationListener
     let add_listener_id = mcp
-        .send_add_conversation_listener_request(AddConversationListenerParams { conversation_id })
+        .send_add_conversation_listener_request(AddConversationListenerParams {
+            conversation_id,
+            experimental_raw_events: false,
+        })
         .await?;
     let _add_listener_resp: JSONRPCResponse = timeout(
         DEFAULT_READ_TIMEOUT,
@@ -100,7 +103,7 @@ async fn shell_command_interruption() -> anyhow::Result<()> {
     let send_user_id = mcp
         .send_send_user_message_request(SendUserMessageParams {
             conversation_id,
-            items: vec![codex_protocol::mcp_protocol::InputItem::Text {
+            items: vec![codex_app_server_protocol::InputItem::Text {
                 text: "run first sleep command".to_string(),
             }],
         })

codex-rs/app-server/tests/suite/list_resume.rs

@@ -0,0 +1,426 @@
+use anyhow::Result;
+use app_test_support::McpProcess;
+use app_test_support::to_response;
+use codex_app_server_protocol::JSONRPCNotification;
+use codex_app_server_protocol::JSONRPCResponse;
+use codex_app_server_protocol::ListConversationsParams;
+use codex_app_server_protocol::ListConversationsResponse;
+use codex_app_server_protocol::NewConversationParams; // reused for overrides shape
+use codex_app_server_protocol::RequestId;
+use codex_app_server_protocol::ResumeConversationParams;
+use codex_app_server_protocol::ResumeConversationResponse;
+use codex_app_server_protocol::ServerNotification;
+use codex_app_server_protocol::SessionConfiguredNotification;
+use codex_core::protocol::EventMsg;
+use codex_protocol::models::ContentItem;
+use codex_protocol::models::ResponseItem;
+use pretty_assertions::assert_eq;
+use serde_json::json;
+use std::fs;
+use std::path::Path;
+use tempfile::TempDir;
+use tokio::time::timeout;
+use uuid::Uuid;
+
+const DEFAULT_READ_TIMEOUT: std::time::Duration = std::time::Duration::from_secs(10);
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
+async fn test_list_and_resume_conversations() -> Result<()> {
+    // Prepare a temporary CODEX_HOME with a few fake rollout files.
+    let codex_home = TempDir::new()?;
+    create_fake_rollout(
+        codex_home.path(),
+        "2025-01-02T12-00-00",
+        "2025-01-02T12:00:00Z",
+        "Hello A",
+        Some("openai"),
+    )?;
+    create_fake_rollout(
+        codex_home.path(),
+        "2025-01-01T13-00-00",
+        "2025-01-01T13:00:00Z",
+        "Hello B",
+        Some("openai"),
+    )?;
+    create_fake_rollout(
+        codex_home.path(),
+        "2025-01-01T12-00-00",
+        "2025-01-01T12:00:00Z",
+        "Hello C",
+        None,
+    )?;
+
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
+
+    // Request first page with size 2
+    let req_id = mcp
+        .send_list_conversations_request(ListConversationsParams {
+            page_size: Some(2),
+            cursor: None,
+            model_providers: None,
+        })
+        .await?;
+    let resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(req_id)),
+    )
+    .await??;
+    let ListConversationsResponse { items, next_cursor } =
+        to_response::<ListConversationsResponse>(resp)?;
+
+    assert_eq!(items.len(), 2);
+    // Newest first; preview text should match
+    assert_eq!(items[0].preview, "Hello A");
+    assert_eq!(items[1].preview, "Hello B");
+    assert_eq!(items[0].model_provider, "openai");
+    assert_eq!(items[1].model_provider, "openai");
+    assert!(items[0].path.is_absolute());
+    assert!(next_cursor.is_some());
+
+    // Request the next page using the cursor
+    let req_id2 = mcp
+        .send_list_conversations_request(ListConversationsParams {
+            page_size: Some(2),
+            cursor: next_cursor,
+            model_providers: None,
+        })
+        .await?;
+    let resp2: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(req_id2)),
+    )
+    .await??;
+    let ListConversationsResponse {
+        items: items2,
+        next_cursor: next2,
+        ..
+    } = to_response::<ListConversationsResponse>(resp2)?;
+    assert_eq!(items2.len(), 1);
+    assert_eq!(items2[0].preview, "Hello C");
+    assert_eq!(items2[0].model_provider, "openai");
+    assert_eq!(next2, None);
+
+    // Add a conversation with an explicit non-OpenAI provider for filter tests.
+    create_fake_rollout(
+        codex_home.path(),
+        "2025-01-01T11-30-00",
+        "2025-01-01T11:30:00Z",
+        "Hello TP",
+        Some("test-provider"),
+    )?;
+
+    // Filtering by model provider should return only matching sessions.
+    let filter_req_id = mcp
+        .send_list_conversations_request(ListConversationsParams {
+            page_size: Some(10),
+            cursor: None,
+            model_providers: Some(vec!["test-provider".to_string()]),
+        })
+        .await?;
+    let filter_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(filter_req_id)),
+    )
+    .await??;
+    let ListConversationsResponse {
+        items: filtered_items,
+        next_cursor: filtered_next,
+    } = to_response::<ListConversationsResponse>(filter_resp)?;
+    assert_eq!(filtered_items.len(), 1);
+    assert_eq!(filtered_next, None);
+    assert_eq!(filtered_items[0].preview, "Hello TP");
+    assert_eq!(filtered_items[0].model_provider, "test-provider");
+
+    // Empty filter should include every session regardless of provider metadata.
+    let unfiltered_req_id = mcp
+        .send_list_conversations_request(ListConversationsParams {
+            page_size: Some(10),
+            cursor: None,
+            model_providers: Some(Vec::new()),
+        })
+        .await?;
+    let unfiltered_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(unfiltered_req_id)),
+    )
+    .await??;
+    let ListConversationsResponse {
+        items: unfiltered_items,
+        next_cursor: unfiltered_next,
+    } = to_response::<ListConversationsResponse>(unfiltered_resp)?;
+    assert_eq!(unfiltered_items.len(), 4);
+    assert!(unfiltered_next.is_none());
+
+    let empty_req_id = mcp
+        .send_list_conversations_request(ListConversationsParams {
+            page_size: Some(10),
+            cursor: None,
+            model_providers: Some(vec!["other".to_string()]),
+        })
+        .await?;
+    let empty_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(empty_req_id)),
+    )
+    .await??;
+    let ListConversationsResponse {
+        items: empty_items,
+        next_cursor: empty_next,
+    } = to_response::<ListConversationsResponse>(empty_resp)?;
+    assert!(empty_items.is_empty());
+    assert!(empty_next.is_none());
+
+    let first_item = &items[0];
+
+    // Now resume one of the sessions from an explicit rollout path.
+    let resume_req_id = mcp
+        .send_resume_conversation_request(ResumeConversationParams {
+            path: Some(first_item.path.clone()),
+            conversation_id: None,
+            history: None,
+            overrides: Some(NewConversationParams {
+                model: Some("o3".to_string()),
+                ..Default::default()
+            }),
+        })
+        .await?;
+
+    // Expect a codex/event notification with msg.type == sessionConfigured
+    let notification: JSONRPCNotification = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("sessionConfigured"),
+    )
+    .await??;
+    let session_configured: ServerNotification = notification.try_into()?;
+    let ServerNotification::SessionConfigured(SessionConfiguredNotification {
+        model,
+        rollout_path,
+        initial_messages: session_initial_messages,
+        ..
+    }) = session_configured
+    else {
+        unreachable!("expected sessionConfigured notification");
+    };
+    assert_eq!(model, "o3");
+    assert_eq!(rollout_path, first_item.path.clone());
+    let session_initial_messages = session_initial_messages
+        .expect("expected initial messages when resuming from rollout path");
+    match session_initial_messages.as_slice() {
+        [EventMsg::UserMessage(message)] => {
+            assert_eq!(message.message, first_item.preview.clone());
+        }
+        other => panic!("unexpected initial messages from rollout resume: {other:#?}"),
+    }
+
+    // Then the response for resumeConversation
+    let resume_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(resume_req_id)),
+    )
+    .await??;
+    let ResumeConversationResponse {
+        conversation_id,
+        model: resume_model,
+        initial_messages: response_initial_messages,
+        ..
+    } = to_response::<ResumeConversationResponse>(resume_resp)?;
+    // conversation id should be a valid UUID
+    assert!(!conversation_id.to_string().is_empty());
+    assert_eq!(resume_model, "o3");
+    let response_initial_messages =
+        response_initial_messages.expect("expected initial messages in resume response");
+    match response_initial_messages.as_slice() {
+        [EventMsg::UserMessage(message)] => {
+            assert_eq!(message.message, first_item.preview.clone());
+        }
+        other => panic!("unexpected initial messages in resume response: {other:#?}"),
+    }
+
+    // Resuming with only a conversation id should locate the rollout automatically.
+    let resume_by_id_req_id = mcp
+        .send_resume_conversation_request(ResumeConversationParams {
+            path: None,
+            conversation_id: Some(first_item.conversation_id),
+            history: None,
+            overrides: Some(NewConversationParams {
+                model: Some("o3".to_string()),
+                ..Default::default()
+            }),
+        })
+        .await?;
+    let notification: JSONRPCNotification = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("sessionConfigured"),
+    )
+    .await??;
+    let session_configured: ServerNotification = notification.try_into()?;
+    let ServerNotification::SessionConfigured(SessionConfiguredNotification {
+        model,
+        rollout_path,
+        initial_messages: session_initial_messages,
+        ..
+    }) = session_configured
+    else {
+        unreachable!("expected sessionConfigured notification");
+    };
+    assert_eq!(model, "o3");
+    assert_eq!(rollout_path, first_item.path.clone());
+    let session_initial_messages = session_initial_messages
+        .expect("expected initial messages when resuming from conversation id");
+    match session_initial_messages.as_slice() {
+        [EventMsg::UserMessage(message)] => {
+            assert_eq!(message.message, first_item.preview.clone());
+        }
+        other => panic!("unexpected initial messages from conversation id resume: {other:#?}"),
+    }
+    let resume_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(resume_by_id_req_id)),
+    )
+    .await??;
+    let ResumeConversationResponse {
+        conversation_id: by_id_conversation_id,
+        model: by_id_model,
+        initial_messages: by_id_initial_messages,
+        ..
+    } = to_response::<ResumeConversationResponse>(resume_resp)?;
+    assert!(!by_id_conversation_id.to_string().is_empty());
+    assert_eq!(by_id_model, "o3");
+    let by_id_initial_messages = by_id_initial_messages
+        .expect("expected initial messages when resuming from conversation id response");
+    match by_id_initial_messages.as_slice() {
+        [EventMsg::UserMessage(message)] => {
+            assert_eq!(message.message, first_item.preview.clone());
+        }
+        other => {
+            panic!("unexpected initial messages in conversation id resume response: {other:#?}")
+        }
+    }
+
+    // Resuming with explicit history should succeed even without a stored rollout.
+    let fork_history_text = "Hello from history";
+    let history = vec![ResponseItem::Message {
+        id: None,
+        role: "user".to_string(),
+        content: vec![ContentItem::InputText {
+            text: fork_history_text.to_string(),
+        }],
+    }];
+    let resume_with_history_req_id = mcp
+        .send_resume_conversation_request(ResumeConversationParams {
+            path: None,
+            conversation_id: None,
+            history: Some(history),
+            overrides: Some(NewConversationParams {
+                model: Some("o3".to_string()),
+                ..Default::default()
+            }),
+        })
+        .await?;
+    let notification: JSONRPCNotification = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("sessionConfigured"),
+    )
+    .await??;
+    let session_configured: ServerNotification = notification.try_into()?;
+    let ServerNotification::SessionConfigured(SessionConfiguredNotification {
+        model,
+        initial_messages: session_initial_messages,
+        ..
+    }) = session_configured
+    else {
+        unreachable!("expected sessionConfigured notification");
+    };
+    assert_eq!(model, "o3");
+    assert!(
+        session_initial_messages.as_ref().is_none_or(Vec::is_empty),
+        "expected no initial messages when resuming from explicit history but got {session_initial_messages:#?}"
+    );
+    let resume_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(resume_with_history_req_id)),
+    )
+    .await??;
+    let ResumeConversationResponse {
+        conversation_id: history_conversation_id,
+        model: history_model,
+        initial_messages: history_initial_messages,
+        ..
+    } = to_response::<ResumeConversationResponse>(resume_resp)?;
+    assert!(!history_conversation_id.to_string().is_empty());
+    assert_eq!(history_model, "o3");
+    assert!(
+        history_initial_messages.as_ref().is_none_or(Vec::is_empty),
+        "expected no initial messages in resume response when history is provided but got {history_initial_messages:#?}"
+    );
+
+    Ok(())
+}
+
+fn create_fake_rollout(
+    codex_home: &Path,
+    filename_ts: &str,
+    meta_rfc3339: &str,
+    preview: &str,
+    model_provider: Option<&str>,
+) -> Result<()> {
+    let uuid = Uuid::new_v4();
+    // sessions/YYYY/MM/DD/ derived from filename_ts (YYYY-MM-DDThh-mm-ss)
+    let year = &filename_ts[0..4];
+    let month = &filename_ts[5..7];
+    let day = &filename_ts[8..10];
+    let dir = codex_home.join("sessions").join(year).join(month).join(day);
+    fs::create_dir_all(&dir)?;
+
+    let file_path = dir.join(format!("rollout-{filename_ts}-{uuid}.jsonl"));
+    let mut lines = Vec::new();
+    // Meta line with timestamp (flattened meta in payload for new schema)
+    let mut payload = json!({
+        "id": uuid,
+        "timestamp": meta_rfc3339,
+        "cwd": "/",
+        "originator": "codex",
+        "cli_version": "0.0.0",
+        "instructions": null,
+    });
+    if let Some(provider) = model_provider {
+        payload["model_provider"] = json!(provider);
+    }
+    lines.push(
+        json!({
+            "timestamp": meta_rfc3339,
+            "type": "session_meta",
+            "payload": payload
+        })
+        .to_string(),
+    );
+    // Minimal user message entry as a persisted response item (with envelope timestamp)
+    lines.push(
+        json!({
+            "timestamp": meta_rfc3339,
+            "type":"response_item",
+            "payload": {
+                "type":"message",
+                "role":"user",
+                "content":[{"type":"input_text","text": preview}]
+            }
+        })
+        .to_string(),
+    );
+    // Add a matching user message event line to satisfy filters
+    lines.push(
+        json!({
+            "timestamp": meta_rfc3339,
+            "type":"event_msg",
+            "payload": {
+                "type":"user_message",
+                "message": preview,
+                "kind": "plain"
+            }
+        })
+        .to_string(),
+    );
+    fs::write(file_path, lines.join("\n") + "\n")?;
+    Ok(())
+}

codex-rs/app-server/tests/suite/login.rs

@@ -0,0 +1,206 @@
+use anyhow::Result;
+use app_test_support::McpProcess;
+use app_test_support::to_response;
+use codex_app_server_protocol::CancelLoginChatGptParams;
+use codex_app_server_protocol::CancelLoginChatGptResponse;
+use codex_app_server_protocol::GetAuthStatusParams;
+use codex_app_server_protocol::GetAuthStatusResponse;
+use codex_app_server_protocol::JSONRPCError;
+use codex_app_server_protocol::JSONRPCResponse;
+use codex_app_server_protocol::LoginChatGptResponse;
+use codex_app_server_protocol::LogoutChatGptResponse;
+use codex_app_server_protocol::RequestId;
+use codex_core::auth::AuthCredentialsStoreMode;
+use codex_login::login_with_api_key;
+use serial_test::serial;
+use std::path::Path;
+use std::time::Duration;
+use tempfile::TempDir;
+use tokio::time::timeout;
+
+const DEFAULT_READ_TIMEOUT: std::time::Duration = std::time::Duration::from_secs(10);
+
+// Helper to create a config.toml; mirrors create_conversation.rs
+fn create_config_toml(codex_home: &Path) -> std::io::Result<()> {
+    let config_toml = codex_home.join("config.toml");
+    std::fs::write(
+        config_toml,
+        r#"
+model = "mock-model"
+approval_policy = "never"
+sandbox_mode = "danger-full-access"
+
+model_provider = "mock_provider"
+
+[model_providers.mock_provider]
+name = "Mock provider for test"
+base_url = "http://127.0.0.1:0/v1"
+wire_api = "chat"
+request_max_retries = 0
+stream_max_retries = 0
+"#,
+    )
+}
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
+async fn logout_chatgpt_removes_auth() -> Result<()> {
+    let codex_home = TempDir::new()?;
+    create_config_toml(codex_home.path())?;
+    login_with_api_key(
+        codex_home.path(),
+        "sk-test-key",
+        AuthCredentialsStoreMode::File,
+    )?;
+    assert!(codex_home.path().join("auth.json").exists());
+
+    let mut mcp = McpProcess::new_with_env(codex_home.path(), &[("OPENAI_API_KEY", None)]).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
+
+    let id = mcp.send_logout_chat_gpt_request().await?;
+    let resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(id)),
+    )
+    .await??;
+    let _ok: LogoutChatGptResponse = to_response(resp)?;
+
+    assert!(
+        !codex_home.path().join("auth.json").exists(),
+        "auth.json should be deleted"
+    );
+
+    // Verify status reflects signed-out state.
+    let status_id = mcp
+        .send_get_auth_status_request(GetAuthStatusParams {
+            include_token: Some(true),
+            refresh_token: Some(false),
+        })
+        .await?;
+    let status_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(status_id)),
+    )
+    .await??;
+    let status: GetAuthStatusResponse = to_response(status_resp)?;
+    assert_eq!(status.auth_method, None);
+    assert_eq!(status.auth_token, None);
+    Ok(())
+}
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
+// Serialize tests that launch the login server since it binds to a fixed port.
+#[serial(login_port)]
+async fn login_and_cancel_chatgpt() -> Result<()> {
+    let codex_home = TempDir::new()?;
+    create_config_toml(codex_home.path())?;
+
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
+
+    let login_id = mcp.send_login_chat_gpt_request().await?;
+    let login_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(login_id)),
+    )
+    .await??;
+    let login: LoginChatGptResponse = to_response(login_resp)?;
+
+    let cancel_id = mcp
+        .send_cancel_login_chat_gpt_request(CancelLoginChatGptParams {
+            login_id: login.login_id,
+        })
+        .await?;
+    let cancel_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(cancel_id)),
+    )
+    .await??;
+    let _ok: CancelLoginChatGptResponse = to_response(cancel_resp)?;
+
+    // Optionally observe the completion notification; do not fail if it races.
+    let maybe_note = timeout(
+        Duration::from_secs(2),
+        mcp.read_stream_until_notification_message("codex/event/login_chat_gpt_complete"),
+    )
+    .await;
+    if maybe_note.is_err() {
+        eprintln!("warning: did not observe login_chat_gpt_complete notification after cancel");
+    }
+    Ok(())
+}
+
+fn create_config_toml_forced_login(codex_home: &Path, forced_method: &str) -> std::io::Result<()> {
+    let config_toml = codex_home.join("config.toml");
+    let contents = format!(
+        r#"
+model = "mock-model"
+approval_policy = "never"
+sandbox_mode = "danger-full-access"
+forced_login_method = "{forced_method}"
+"#
+    );
+    std::fs::write(config_toml, contents)
+}
+
+fn create_config_toml_forced_workspace(
+    codex_home: &Path,
+    workspace_id: &str,
+) -> std::io::Result<()> {
+    let config_toml = codex_home.join("config.toml");
+    let contents = format!(
+        r#"
+model = "mock-model"
+approval_policy = "never"
+sandbox_mode = "danger-full-access"
+forced_chatgpt_workspace_id = "{workspace_id}"
+"#
+    );
+    std::fs::write(config_toml, contents)
+}
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
+async fn login_chatgpt_rejected_when_forced_api() -> Result<()> {
+    let codex_home = TempDir::new()?;
+    create_config_toml_forced_login(codex_home.path(), "api")?;
+
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
+
+    let request_id = mcp.send_login_chat_gpt_request().await?;
+    let err: JSONRPCError = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_error_message(RequestId::Integer(request_id)),
+    )
+    .await??;
+
+    assert_eq!(
+        err.error.message,
+        "ChatGPT login is disabled. Use API key login instead."
+    );
+    Ok(())
+}
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
+// Serialize tests that launch the login server since it binds to a fixed port.
+#[serial(login_port)]
+async fn login_chatgpt_includes_forced_workspace_query_param() -> Result<()> {
+    let codex_home = TempDir::new()?;
+    create_config_toml_forced_workspace(codex_home.path(), "ws-forced")?;
+
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
+
+    let request_id = mcp.send_login_chat_gpt_request().await?;
+    let resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(request_id)),
+    )
+    .await??;
+
+    let login: LoginChatGptResponse = to_response(resp)?;
+    assert!(
+        login.auth_url.contains("allowed_workspace_id=ws-forced"),
+        "auth URL should include forced workspace"
+    );
+    Ok(())
+}

codex-rs/app-server/tests/suite/mod.rs

@@ -0,0 +1,15 @@
+mod archive_conversation;
+mod auth;
+mod codex_message_processor_flow;
+mod config;
+mod create_conversation;
+mod fuzzy_file_search;
+mod interrupt;
+mod list_resume;
+mod login;
+mod model_list;
+mod rate_limits;
+mod send_message;
+mod set_default_model;
+mod user_agent;
+mod user_info;

codex-rs/app-server/tests/suite/model_list.rs

@@ -0,0 +1,183 @@
+use std::time::Duration;
+
+use anyhow::Result;
+use anyhow::anyhow;
+use app_test_support::McpProcess;
+use app_test_support::to_response;
+use codex_app_server_protocol::JSONRPCError;
+use codex_app_server_protocol::JSONRPCResponse;
+use codex_app_server_protocol::ListModelsParams;
+use codex_app_server_protocol::ListModelsResponse;
+use codex_app_server_protocol::Model;
+use codex_app_server_protocol::ReasoningEffortOption;
+use codex_app_server_protocol::RequestId;
+use codex_protocol::config_types::ReasoningEffort;
+use pretty_assertions::assert_eq;
+use tempfile::TempDir;
+use tokio::time::timeout;
+
+const DEFAULT_TIMEOUT: Duration = Duration::from_secs(10);
+const INVALID_REQUEST_ERROR_CODE: i64 = -32600;
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
+async fn list_models_returns_all_models_with_large_limit() -> Result<()> {
+    let codex_home = TempDir::new()?;
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+
+    timeout(DEFAULT_TIMEOUT, mcp.initialize()).await??;
+
+    let request_id = mcp
+        .send_list_models_request(ListModelsParams {
+            page_size: Some(100),
+            cursor: None,
+        })
+        .await?;
+
+    let response: JSONRPCResponse = timeout(
+        DEFAULT_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(request_id)),
+    )
+    .await??;
+
+    let ListModelsResponse { items, next_cursor } = to_response::<ListModelsResponse>(response)?;
+
+    let expected_models = vec![
+        Model {
+            id: "gpt-5-codex".to_string(),
+            model: "gpt-5-codex".to_string(),
+            display_name: "gpt-5-codex".to_string(),
+            description: "Optimized for coding tasks with many tools.".to_string(),
+            supported_reasoning_efforts: vec![
+                ReasoningEffortOption {
+                    reasoning_effort: ReasoningEffort::Low,
+                    description: "Fastest responses with limited reasoning".to_string(),
+                },
+                ReasoningEffortOption {
+                    reasoning_effort: ReasoningEffort::Medium,
+                    description: "Dynamically adjusts reasoning based on the task".to_string(),
+                },
+                ReasoningEffortOption {
+                    reasoning_effort: ReasoningEffort::High,
+                    description: "Maximizes reasoning depth for complex or ambiguous problems"
+                        .to_string(),
+                },
+            ],
+            default_reasoning_effort: ReasoningEffort::Medium,
+            is_default: true,
+        },
+        Model {
+            id: "gpt-5".to_string(),
+            model: "gpt-5".to_string(),
+            display_name: "gpt-5".to_string(),
+            description: "Broad world knowledge with strong general reasoning.".to_string(),
+            supported_reasoning_efforts: vec![
+                ReasoningEffortOption {
+                    reasoning_effort: ReasoningEffort::Minimal,
+                    description: "Fastest responses with little reasoning".to_string(),
+                },
+                ReasoningEffortOption {
+                    reasoning_effort: ReasoningEffort::Low,
+                    description: "Balances speed with some reasoning; useful for straightforward \
+                                   queries and short explanations"
+                        .to_string(),
+                },
+                ReasoningEffortOption {
+                    reasoning_effort: ReasoningEffort::Medium,
+                    description: "Provides a solid balance of reasoning depth and latency for \
+                         general-purpose tasks"
+                        .to_string(),
+                },
+                ReasoningEffortOption {
+                    reasoning_effort: ReasoningEffort::High,
+                    description: "Maximizes reasoning depth for complex or ambiguous problems"
+                        .to_string(),
+                },
+            ],
+            default_reasoning_effort: ReasoningEffort::Medium,
+            is_default: false,
+        },
+    ];
+
+    assert_eq!(items, expected_models);
+    assert!(next_cursor.is_none());
+    Ok(())
+}
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
+async fn list_models_pagination_works() -> Result<()> {
+    let codex_home = TempDir::new()?;
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+
+    timeout(DEFAULT_TIMEOUT, mcp.initialize()).await??;
+
+    let first_request = mcp
+        .send_list_models_request(ListModelsParams {
+            page_size: Some(1),
+            cursor: None,
+        })
+        .await?;
+
+    let first_response: JSONRPCResponse = timeout(
+        DEFAULT_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(first_request)),
+    )
+    .await??;
+
+    let ListModelsResponse {
+        items: first_items,
+        next_cursor: first_cursor,
+    } = to_response::<ListModelsResponse>(first_response)?;
+
+    assert_eq!(first_items.len(), 1);
+    assert_eq!(first_items[0].id, "gpt-5-codex");
+    let next_cursor = first_cursor.ok_or_else(|| anyhow!("cursor for second page"))?;
+
+    let second_request = mcp
+        .send_list_models_request(ListModelsParams {
+            page_size: Some(1),
+            cursor: Some(next_cursor.clone()),
+        })
+        .await?;
+
+    let second_response: JSONRPCResponse = timeout(
+        DEFAULT_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(second_request)),
+    )
+    .await??;
+
+    let ListModelsResponse {
+        items: second_items,
+        next_cursor: second_cursor,
+    } = to_response::<ListModelsResponse>(second_response)?;
+
+    assert_eq!(second_items.len(), 1);
+    assert_eq!(second_items[0].id, "gpt-5");
+    assert!(second_cursor.is_none());
+    Ok(())
+}
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
+async fn list_models_rejects_invalid_cursor() -> Result<()> {
+    let codex_home = TempDir::new()?;
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+
+    timeout(DEFAULT_TIMEOUT, mcp.initialize()).await??;
+
+    let request_id = mcp
+        .send_list_models_request(ListModelsParams {
+            page_size: None,
+            cursor: Some("invalid".to_string()),
+        })
+        .await?;
+
+    let error: JSONRPCError = timeout(
+        DEFAULT_TIMEOUT,
+        mcp.read_stream_until_error_message(RequestId::Integer(request_id)),
+    )
+    .await??;
+
+    assert_eq!(error.id, RequestId::Integer(request_id));
+    assert_eq!(error.error.code, INVALID_REQUEST_ERROR_CODE);
+    assert_eq!(error.error.message, "invalid cursor: invalid");
+    Ok(())
+}

codex-rs/app-server/tests/suite/rate_limits.rs

@@ -0,0 +1,181 @@
+use anyhow::Result;
+use app_test_support::ChatGptAuthFixture;
+use app_test_support::McpProcess;
+use app_test_support::to_response;
+use app_test_support::write_chatgpt_auth;
+use codex_app_server_protocol::GetAccountRateLimitsResponse;
+use codex_app_server_protocol::JSONRPCError;
+use codex_app_server_protocol::JSONRPCResponse;
+use codex_app_server_protocol::LoginApiKeyParams;
+use codex_app_server_protocol::RequestId;
+use codex_core::auth::AuthCredentialsStoreMode;
+use codex_protocol::protocol::RateLimitSnapshot;
+use codex_protocol::protocol::RateLimitWindow;
+use pretty_assertions::assert_eq;
+use serde_json::json;
+use std::path::Path;
+use tempfile::TempDir;
+use tokio::time::timeout;
+use wiremock::Mock;
+use wiremock::MockServer;
+use wiremock::ResponseTemplate;
+use wiremock::matchers::header;
+use wiremock::matchers::method;
+use wiremock::matchers::path;
+
+const DEFAULT_READ_TIMEOUT: std::time::Duration = std::time::Duration::from_secs(10);
+const INVALID_REQUEST_ERROR_CODE: i64 = -32600;
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
+async fn get_account_rate_limits_requires_auth() -> Result<()> {
+    let codex_home = TempDir::new()?;
+
+    let mut mcp = McpProcess::new_with_env(codex_home.path(), &[("OPENAI_API_KEY", None)]).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
+
+    let request_id = mcp.send_get_account_rate_limits_request().await?;
+
+    let error: JSONRPCError = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_error_message(RequestId::Integer(request_id)),
+    )
+    .await??;
+
+    assert_eq!(error.id, RequestId::Integer(request_id));
+    assert_eq!(error.error.code, INVALID_REQUEST_ERROR_CODE);
+    assert_eq!(
+        error.error.message,
+        "codex account authentication required to read rate limits"
+    );
+
+    Ok(())
+}
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
+async fn get_account_rate_limits_requires_chatgpt_auth() -> Result<()> {
+    let codex_home = TempDir::new()?;
+
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
+
+    login_with_api_key(&mut mcp, "sk-test-key").await?;
+
+    let request_id = mcp.send_get_account_rate_limits_request().await?;
+
+    let error: JSONRPCError = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_error_message(RequestId::Integer(request_id)),
+    )
+    .await??;
+
+    assert_eq!(error.id, RequestId::Integer(request_id));
+    assert_eq!(error.error.code, INVALID_REQUEST_ERROR_CODE);
+    assert_eq!(
+        error.error.message,
+        "chatgpt authentication required to read rate limits"
+    );
+
+    Ok(())
+}
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
+async fn get_account_rate_limits_returns_snapshot() -> Result<()> {
+    let codex_home = TempDir::new()?;
+    write_chatgpt_auth(
+        codex_home.path(),
+        ChatGptAuthFixture::new("chatgpt-token")
+            .account_id("account-123")
+            .plan_type("pro"),
+        AuthCredentialsStoreMode::File,
+    )?;
+
+    let server = MockServer::start().await;
+    let server_url = server.uri();
+    write_chatgpt_base_url(codex_home.path(), &server_url)?;
+
+    let primary_reset_timestamp = chrono::DateTime::parse_from_rfc3339("2025-01-01T00:02:00Z")
+        .expect("parse primary reset timestamp")
+        .timestamp();
+    let secondary_reset_timestamp = chrono::DateTime::parse_from_rfc3339("2025-01-01T01:00:00Z")
+        .expect("parse secondary reset timestamp")
+        .timestamp();
+    let response_body = json!({
+        "plan_type": "pro",
+        "rate_limit": {
+            "allowed": true,
+            "limit_reached": false,
+            "primary_window": {
+                "used_percent": 42,
+                "limit_window_seconds": 3600,
+                "reset_after_seconds": 120,
+                "reset_at": primary_reset_timestamp,
+            },
+            "secondary_window": {
+                "used_percent": 5,
+                "limit_window_seconds": 86400,
+                "reset_after_seconds": 43200,
+                "reset_at": secondary_reset_timestamp,
+            }
+        }
+    });
+
+    Mock::given(method("GET"))
+        .and(path("/api/codex/usage"))
+        .and(header("authorization", "Bearer chatgpt-token"))
+        .and(header("chatgpt-account-id", "account-123"))
+        .respond_with(ResponseTemplate::new(200).set_body_json(response_body))
+        .mount(&server)
+        .await;
+
+    let mut mcp = McpProcess::new_with_env(codex_home.path(), &[("OPENAI_API_KEY", None)]).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
+
+    let request_id = mcp.send_get_account_rate_limits_request().await?;
+
+    let response: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(request_id)),
+    )
+    .await??;
+
+    let received: GetAccountRateLimitsResponse = to_response(response)?;
+
+    let expected = GetAccountRateLimitsResponse {
+        rate_limits: RateLimitSnapshot {
+            primary: Some(RateLimitWindow {
+                used_percent: 42.0,
+                window_minutes: Some(60),
+                resets_at: Some(primary_reset_timestamp),
+            }),
+            secondary: Some(RateLimitWindow {
+                used_percent: 5.0,
+                window_minutes: Some(1440),
+                resets_at: Some(secondary_reset_timestamp),
+            }),
+        },
+    };
+    assert_eq!(received, expected);
+
+    Ok(())
+}
+
+async fn login_with_api_key(mcp: &mut McpProcess, api_key: &str) -> Result<()> {
+    let request_id = mcp
+        .send_login_api_key_request(LoginApiKeyParams {
+            api_key: api_key.to_string(),
+        })
+        .await?;
+
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(request_id)),
+    )
+    .await??;
+
+    Ok(())
+}
+
+fn write_chatgpt_base_url(codex_home: &Path, base_url: &str) -> std::io::Result<()> {
+    let config_toml = codex_home.join("config.toml");
+    std::fs::write(config_toml, format!("chatgpt_base_url = \"{base_url}\"\n"))
+}

codex-rs/app-server/tests/suite/send_message.rs

@@ -0,0 +1,391 @@
+use anyhow::Result;
+use app_test_support::McpProcess;
+use app_test_support::create_final_assistant_message_sse_response;
+use app_test_support::create_mock_chat_completions_server;
+use app_test_support::to_response;
+use codex_app_server_protocol::AddConversationListenerParams;
+use codex_app_server_protocol::AddConversationSubscriptionResponse;
+use codex_app_server_protocol::InputItem;
+use codex_app_server_protocol::JSONRPCNotification;
+use codex_app_server_protocol::JSONRPCResponse;
+use codex_app_server_protocol::NewConversationParams;
+use codex_app_server_protocol::NewConversationResponse;
+use codex_app_server_protocol::RequestId;
+use codex_app_server_protocol::SendUserMessageParams;
+use codex_app_server_protocol::SendUserMessageResponse;
+use codex_protocol::ConversationId;
+use codex_protocol::models::ContentItem;
+use codex_protocol::models::ResponseItem;
+use codex_protocol::protocol::RawResponseItemEvent;
+use pretty_assertions::assert_eq;
+use std::path::Path;
+use tempfile::TempDir;
+use tokio::time::timeout;
+
+const DEFAULT_READ_TIMEOUT: std::time::Duration = std::time::Duration::from_secs(10);
+
+#[tokio::test]
+async fn test_send_message_success() -> Result<()> {
+    // Spin up a mock completions server that immediately ends the Codex turn.
+    // Two Codex turns hit the mock model (session start + send-user-message). Provide two SSE responses.
+    let responses = vec![
+        create_final_assistant_message_sse_response("Done")?,
+        create_final_assistant_message_sse_response("Done")?,
+    ];
+    let server = create_mock_chat_completions_server(responses).await;
+
+    // Create a temporary Codex home with config pointing at the mock server.
+    let codex_home = TempDir::new()?;
+    create_config_toml(codex_home.path(), &server.uri())?;
+
+    // Start MCP server process and initialize.
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
+
+    // Start a conversation using the new wire API.
+    let new_conv_id = mcp
+        .send_new_conversation_request(NewConversationParams {
+            ..Default::default()
+        })
+        .await?;
+    let new_conv_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(new_conv_id)),
+    )
+    .await??;
+    let NewConversationResponse {
+        conversation_id, ..
+    } = to_response::<_>(new_conv_resp)?;
+
+    // 2) addConversationListener
+    let add_listener_id = mcp
+        .send_add_conversation_listener_request(AddConversationListenerParams {
+            conversation_id,
+            experimental_raw_events: false,
+        })
+        .await?;
+    let add_listener_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(add_listener_id)),
+    )
+    .await??;
+    let AddConversationSubscriptionResponse { subscription_id: _ } =
+        to_response::<_>(add_listener_resp)?;
+
+    // Now exercise sendUserMessage twice.
+    send_message("Hello", conversation_id, &mut mcp).await?;
+    send_message("Hello again", conversation_id, &mut mcp).await?;
+    Ok(())
+}
+
+#[expect(clippy::expect_used)]
+async fn send_message(
+    message: &str,
+    conversation_id: ConversationId,
+    mcp: &mut McpProcess,
+) -> Result<()> {
+    // Now exercise sendUserMessage.
+    let send_id = mcp
+        .send_send_user_message_request(SendUserMessageParams {
+            conversation_id,
+            items: vec![InputItem::Text {
+                text: message.to_string(),
+            }],
+        })
+        .await?;
+
+    let response: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(send_id)),
+    )
+    .await??;
+
+    let _ok: SendUserMessageResponse = to_response::<SendUserMessageResponse>(response)?;
+
+    // Verify the task_finished notification is received.
+    // Note this also ensures that the final request to the server was made.
+    let task_finished_notification: JSONRPCNotification = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("codex/event/task_complete"),
+    )
+    .await??;
+    let serde_json::Value::Object(map) = task_finished_notification
+        .params
+        .expect("notification should have params")
+    else {
+        panic!("task_finished_notification should have params");
+    };
+    assert_eq!(
+        map.get("conversationId")
+            .expect("should have conversationId"),
+        &serde_json::Value::String(conversation_id.to_string())
+    );
+
+    let raw_attempt = tokio::time::timeout(
+        std::time::Duration::from_millis(200),
+        mcp.read_stream_until_notification_message("codex/event/raw_response_item"),
+    )
+    .await;
+    assert!(
+        raw_attempt.is_err(),
+        "unexpected raw item notification when not opted in"
+    );
+    Ok(())
+}
+
+#[tokio::test]
+async fn test_send_message_raw_notifications_opt_in() -> Result<()> {
+    let responses = vec![create_final_assistant_message_sse_response("Done")?];
+    let server = create_mock_chat_completions_server(responses).await;
+
+    let codex_home = TempDir::new()?;
+    create_config_toml(codex_home.path(), &server.uri())?;
+
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
+
+    let new_conv_id = mcp
+        .send_new_conversation_request(NewConversationParams {
+            developer_instructions: Some("Use the test harness tools.".to_string()),
+            ..Default::default()
+        })
+        .await?;
+    let new_conv_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(new_conv_id)),
+    )
+    .await??;
+    let NewConversationResponse {
+        conversation_id, ..
+    } = to_response::<_>(new_conv_resp)?;
+
+    let add_listener_id = mcp
+        .send_add_conversation_listener_request(AddConversationListenerParams {
+            conversation_id,
+            experimental_raw_events: true,
+        })
+        .await?;
+    let add_listener_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(add_listener_id)),
+    )
+    .await??;
+    let AddConversationSubscriptionResponse { subscription_id: _ } =
+        to_response::<_>(add_listener_resp)?;
+
+    let send_id = mcp
+        .send_send_user_message_request(SendUserMessageParams {
+            conversation_id,
+            items: vec![InputItem::Text {
+                text: "Hello".to_string(),
+            }],
+        })
+        .await?;
+
+    let developer = read_raw_response_item(&mut mcp, conversation_id).await;
+    assert_developer_message(&developer, "Use the test harness tools.");
+
+    let instructions = read_raw_response_item(&mut mcp, conversation_id).await;
+    assert_instructions_message(&instructions);
+
+    let environment = read_raw_response_item(&mut mcp, conversation_id).await;
+    assert_environment_message(&environment);
+
+    let response: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(send_id)),
+    )
+    .await??;
+    let _ok: SendUserMessageResponse = to_response::<SendUserMessageResponse>(response)?;
+
+    let user_message = read_raw_response_item(&mut mcp, conversation_id).await;
+    assert_user_message(&user_message, "Hello");
+
+    let assistant_message = read_raw_response_item(&mut mcp, conversation_id).await;
+    assert_assistant_message(&assistant_message, "Done");
+
+    let _ = tokio::time::timeout(
+        std::time::Duration::from_millis(250),
+        mcp.read_stream_until_notification_message("codex/event/task_complete"),
+    )
+    .await;
+
+    Ok(())
+}
+
+#[tokio::test]
+async fn test_send_message_session_not_found() -> Result<()> {
+    // Start MCP without creating a Codex session
+    let codex_home = TempDir::new()?;
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
+
+    let unknown = ConversationId::new();
+    let req_id = mcp
+        .send_send_user_message_request(SendUserMessageParams {
+            conversation_id: unknown,
+            items: vec![InputItem::Text {
+                text: "ping".to_string(),
+            }],
+        })
+        .await?;
+
+    // Expect an error response for unknown conversation.
+    let err = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_error_message(RequestId::Integer(req_id)),
+    )
+    .await??;
+    assert_eq!(err.id, RequestId::Integer(req_id));
+    Ok(())
+}
+
+// ---------------------------------------------------------------------------
+// Helpers
+// ---------------------------------------------------------------------------
+
+fn create_config_toml(codex_home: &Path, server_uri: &str) -> std::io::Result<()> {
+    let config_toml = codex_home.join("config.toml");
+    std::fs::write(
+        config_toml,
+        format!(
+            r#"
+model = "mock-model"
+approval_policy = "never"
+sandbox_mode = "danger-full-access"
+
+model_provider = "mock_provider"
+
+[model_providers.mock_provider]
+name = "Mock provider for test"
+base_url = "{server_uri}/v1"
+wire_api = "chat"
+request_max_retries = 0
+stream_max_retries = 0
+"#
+        ),
+    )
+}
+
+#[expect(clippy::expect_used)]
+async fn read_raw_response_item(
+    mcp: &mut McpProcess,
+    conversation_id: ConversationId,
+) -> ResponseItem {
+    let raw_notification: JSONRPCNotification = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("codex/event/raw_response_item"),
+    )
+    .await
+    .expect("codex/event/raw_response_item notification timeout")
+    .expect("codex/event/raw_response_item notification resp");
+
+    let serde_json::Value::Object(params) = raw_notification
+        .params
+        .expect("codex/event/raw_response_item should have params")
+    else {
+        panic!("codex/event/raw_response_item should have params");
+    };
+
+    let conversation_id_value = params
+        .get("conversationId")
+        .and_then(|value| value.as_str())
+        .expect("raw response item should include conversationId");
+
+    assert_eq!(
+        conversation_id_value,
+        conversation_id.to_string(),
+        "raw response item conversation mismatch"
+    );
+
+    let msg_value = params
+        .get("msg")
+        .cloned()
+        .expect("raw response item should include msg payload");
+
+    let event: RawResponseItemEvent =
+        serde_json::from_value(msg_value).expect("deserialize raw response item");
+    event.item
+}
+
+fn assert_instructions_message(item: &ResponseItem) {
+    match item {
+        ResponseItem::Message { role, content, .. } => {
+            assert_eq!(role, "user");
+            let texts = content_texts(content);
+            let is_instructions = texts
+                .iter()
+                .any(|text| text.starts_with("# AGENTS.md instructions for "));
+            assert!(
+                is_instructions,
+                "expected instructions message, got {texts:?}"
+            );
+        }
+        other => panic!("expected instructions message, got {other:?}"),
+    }
+}
+
+fn assert_developer_message(item: &ResponseItem, expected_text: &str) {
+    match item {
+        ResponseItem::Message { role, content, .. } => {
+            assert_eq!(role, "developer");
+            let texts = content_texts(content);
+            assert_eq!(
+                texts,
+                vec![expected_text],
+                "expected developer instructions message, got {texts:?}"
+            );
+        }
+        other => panic!("expected developer instructions message, got {other:?}"),
+    }
+}
+
+fn assert_environment_message(item: &ResponseItem) {
+    match item {
+        ResponseItem::Message { role, content, .. } => {
+            assert_eq!(role, "user");
+            let texts = content_texts(content);
+            assert!(
+                texts
+                    .iter()
+                    .any(|text| text.contains("<environment_context>")),
+                "expected environment context message, got {texts:?}"
+            );
+        }
+        other => panic!("expected environment message, got {other:?}"),
+    }
+}
+
+fn assert_user_message(item: &ResponseItem, expected_text: &str) {
+    match item {
+        ResponseItem::Message { role, content, .. } => {
+            assert_eq!(role, "user");
+            let texts = content_texts(content);
+            assert_eq!(texts, vec![expected_text]);
+        }
+        other => panic!("expected user message, got {other:?}"),
+    }
+}
+
+fn assert_assistant_message(item: &ResponseItem, expected_text: &str) {
+    match item {
+        ResponseItem::Message { role, content, .. } => {
+            assert_eq!(role, "assistant");
+            let texts = content_texts(content);
+            assert_eq!(texts, vec![expected_text]);
+        }
+        other => panic!("expected assistant message, got {other:?}"),
+    }
+}
+
+fn content_texts(content: &[ContentItem]) -> Vec<&str> {
+    content
+        .iter()
+        .filter_map(|item| match item {
+            ContentItem::InputText { text } | ContentItem::OutputText { text } => {
+                Some(text.as_str())
+            }
+            _ => None,
+        })
+        .collect()
+}

codex-rs/app-server/tests/suite/set_default_model.rs

@@ -1,57 +1,44 @@
-use std::path::Path;
-
+use anyhow::Result;
+use app_test_support::McpProcess;
+use app_test_support::to_response;
+use codex_app_server_protocol::JSONRPCResponse;
+use codex_app_server_protocol::RequestId;
+use codex_app_server_protocol::SetDefaultModelParams;
+use codex_app_server_protocol::SetDefaultModelResponse;
 use codex_core::config::ConfigToml;
-use codex_protocol::mcp_protocol::SetDefaultModelParams;
-use codex_protocol::mcp_protocol::SetDefaultModelResponse;
-use mcp_test_support::McpProcess;
-use mcp_test_support::to_response;
-use mcp_types::JSONRPCResponse;
-use mcp_types::RequestId;
 use pretty_assertions::assert_eq;
+use std::path::Path;
 use tempfile::TempDir;
 use tokio::time::timeout;
 
 const DEFAULT_READ_TIMEOUT: std::time::Duration = std::time::Duration::from_secs(10);
 
 #[tokio::test(flavor = "multi_thread", worker_threads = 2)]
-async fn set_default_model_persists_overrides() {
-    let codex_home = TempDir::new().expect("create tempdir");
-    create_config_toml(codex_home.path()).expect("write config.toml");
+async fn set_default_model_persists_overrides() -> Result<()> {
+    let codex_home = TempDir::new()?;
+    create_config_toml(codex_home.path())?;
 
-    let mut mcp = McpProcess::new(codex_home.path())
-        .await
-        .expect("spawn mcp process");
-    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize())
-        .await
-        .expect("init timeout")
-        .expect("init failed");
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
 
     let params = SetDefaultModelParams {
         model: Some("gpt-4.1".to_string()),
         reasoning_effort: None,
     };
 
-    let request_id = mcp
-        .send_set_default_model_request(params)
-        .await
-        .expect("send setDefaultModel");
+    let request_id = mcp.send_set_default_model_request(params).await?;
 
     let resp: JSONRPCResponse = timeout(
         DEFAULT_READ_TIMEOUT,
         mcp.read_stream_until_response_message(RequestId::Integer(request_id)),
     )
-    .await
-    .expect("setDefaultModel timeout")
-    .expect("setDefaultModel response");
+    .await??;
 
-    let _: SetDefaultModelResponse =
-        to_response(resp).expect("deserialize setDefaultModel response");
+    let _: SetDefaultModelResponse = to_response(resp)?;
 
     let config_path = codex_home.path().join("config.toml");
-    let config_contents = tokio::fs::read_to_string(&config_path)
-        .await
-        .expect("read config.toml");
-    let config_toml: ConfigToml = toml::from_str(&config_contents).expect("parse config.toml");
+    let config_contents = tokio::fs::read_to_string(&config_path).await?;
+    let config_toml: ConfigToml = toml::from_str(&config_contents)?;
 
     assert_eq!(
         ConfigToml {
@@ -61,6 +48,7 @@ async fn set_default_model_persists_overrides() {
         },
         config_toml,
     );
+    Ok(())
 }
 
 // Helper to create a config.toml; mirrors create_conversation.rs

codex-rs/app-server/tests/suite/user_agent.rs

@@ -0,0 +1,41 @@
+use anyhow::Result;
+use app_test_support::McpProcess;
+use app_test_support::to_response;
+use codex_app_server_protocol::GetUserAgentResponse;
+use codex_app_server_protocol::JSONRPCResponse;
+use codex_app_server_protocol::RequestId;
+use pretty_assertions::assert_eq;
+use tempfile::TempDir;
+use tokio::time::timeout;
+
+const DEFAULT_READ_TIMEOUT: std::time::Duration = std::time::Duration::from_secs(10);
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
+async fn get_user_agent_returns_current_codex_user_agent() -> Result<()> {
+    let codex_home = TempDir::new()?;
+
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
+
+    let request_id = mcp.send_get_user_agent_request().await?;
+    let response: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(request_id)),
+    )
+    .await??;
+
+    let os_info = os_info::get();
+    let user_agent = format!(
+        "codex_cli_rs/0.0.0 ({} {}; {}) {} (codex-app-server-tests; 0.1.0)",
+        os_info.os_type(),
+        os_info.version(),
+        os_info.architecture().unwrap_or("unknown"),
+        codex_core::terminal::user_agent()
+    );
+
+    let received: GetUserAgentResponse = to_response(response)?;
+    let expected = GetUserAgentResponse { user_agent };
+
+    assert_eq!(received, expected);
+    Ok(())
+}

codex-rs/app-server/tests/suite/user_info.rs

@@ -0,0 +1,46 @@
+use anyhow::Result;
+use app_test_support::ChatGptAuthFixture;
+use app_test_support::McpProcess;
+use app_test_support::to_response;
+use app_test_support::write_chatgpt_auth;
+use codex_app_server_protocol::JSONRPCResponse;
+use codex_app_server_protocol::RequestId;
+use codex_app_server_protocol::UserInfoResponse;
+use codex_core::auth::AuthCredentialsStoreMode;
+use pretty_assertions::assert_eq;
+use std::time::Duration;
+use tempfile::TempDir;
+use tokio::time::timeout;
+
+const DEFAULT_READ_TIMEOUT: Duration = Duration::from_secs(10);
+
+#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
+async fn user_info_returns_email_from_auth_json() -> Result<()> {
+    let codex_home = TempDir::new()?;
+
+    write_chatgpt_auth(
+        codex_home.path(),
+        ChatGptAuthFixture::new("access")
+            .refresh_token("refresh")
+            .email("user@example.com"),
+        AuthCredentialsStoreMode::File,
+    )?;
+
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
+
+    let request_id = mcp.send_user_info_request().await?;
+    let response: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(request_id)),
+    )
+    .await??;
+
+    let received: UserInfoResponse = to_response(response)?;
+    let expected = UserInfoResponse {
+        alleged_user_email: Some("user@example.com".to_string()),
+    };
+
+    assert_eq!(received, expected);
+    Ok(())
+}

codex-rs/apply-patch/Cargo.toml

@@ -23,5 +23,6 @@ tree-sitter-bash = { workspace = true }
 
 [dev-dependencies]
 assert_cmd = { workspace = true }
+assert_matches = { workspace = true }
 pretty_assertions = { workspace = true }
 tempfile = { workspace = true }

codex-rs/apply-patch/src/lib.rs

@@ -843,6 +843,7 @@ pub fn print_summary(
 #[cfg(test)]
 mod tests {
     use super::*;
+    use assert_matches::assert_matches;
     use pretty_assertions::assert_eq;
     use std::fs;
     use std::string::ToString;
@@ -894,10 +895,10 @@ mod tests {
 
     fn assert_not_match(script: &str) {
         let args = args_bash(script);
-        assert!(matches!(
+        assert_matches!(
             maybe_parse_apply_patch(&args),
             MaybeApplyPatch::NotApplyPatch
-        ));
+        );
     }
 
     #[test]
@@ -905,10 +906,10 @@ mod tests {
         let patch = "*** Begin Patch\n*** Add File: foo\n+hi\n*** End Patch".to_string();
         let args = vec![patch];
         let dir = tempdir().unwrap();
-        assert!(matches!(
+        assert_matches!(
             maybe_parse_apply_patch_verified(&args, dir.path()),
             MaybeApplyPatchVerified::CorrectnessError(ApplyPatchError::ImplicitInvocation)
-        ));
+        );
     }
 
     #[test]
@@ -916,10 +917,10 @@ mod tests {
         let script = "*** Begin Patch\n*** Add File: foo\n+hi\n*** End Patch";
         let args = args_bash(script);
         let dir = tempdir().unwrap();
-        assert!(matches!(
+        assert_matches!(
             maybe_parse_apply_patch_verified(&args, dir.path()),
             MaybeApplyPatchVerified::CorrectnessError(ApplyPatchError::ImplicitInvocation)
-        ));
+        );
     }
 
     #[test]

codex-rs/apply-patch/tests/suite/mod.rs

@@ -1 +1,3 @@
 mod cli;
+#[cfg(not(target_os = "windows"))]
+mod tool;

codex-rs/apply-patch/tests/suite/tool.rs

@@ -0,0 +1,257 @@
+use assert_cmd::Command;
+use pretty_assertions::assert_eq;
+use std::fs;
+use std::path::Path;
+use tempfile::tempdir;
+
+fn run_apply_patch_in_dir(dir: &Path, patch: &str) -> anyhow::Result<assert_cmd::assert::Assert> {
+    let mut cmd = Command::cargo_bin("apply_patch")?;
+    cmd.current_dir(dir);
+    Ok(cmd.arg(patch).assert())
+}
+
+fn apply_patch_command(dir: &Path) -> anyhow::Result<Command> {
+    let mut cmd = Command::cargo_bin("apply_patch")?;
+    cmd.current_dir(dir);
+    Ok(cmd)
+}
+
+#[test]
+fn test_apply_patch_cli_applies_multiple_operations() -> anyhow::Result<()> {
+    let tmp = tempdir()?;
+    let modify_path = tmp.path().join("modify.txt");
+    let delete_path = tmp.path().join("delete.txt");
+
+    fs::write(&modify_path, "line1\nline2\n")?;
+    fs::write(&delete_path, "obsolete\n")?;
+
+    let patch = "*** Begin Patch\n*** Add File: nested/new.txt\n+created\n*** Delete File: delete.txt\n*** Update File: modify.txt\n@@\n-line2\n+changed\n*** End Patch";
+
+    run_apply_patch_in_dir(tmp.path(), patch)?.success().stdout(
+        "Success. Updated the following files:\nA nested/new.txt\nM modify.txt\nD delete.txt\n",
+    );
+
+    assert_eq!(
+        fs::read_to_string(tmp.path().join("nested/new.txt"))?,
+        "created\n"
+    );
+    assert_eq!(fs::read_to_string(&modify_path)?, "line1\nchanged\n");
+    assert!(!delete_path.exists());
+
+    Ok(())
+}
+
+#[test]
+fn test_apply_patch_cli_applies_multiple_chunks() -> anyhow::Result<()> {
+    let tmp = tempdir()?;
+    let target_path = tmp.path().join("multi.txt");
+    fs::write(&target_path, "line1\nline2\nline3\nline4\n")?;
+
+    let patch = "*** Begin Patch\n*** Update File: multi.txt\n@@\n-line2\n+changed2\n@@\n-line4\n+changed4\n*** End Patch";
+
+    run_apply_patch_in_dir(tmp.path(), patch)?
+        .success()
+        .stdout("Success. Updated the following files:\nM multi.txt\n");
+
+    assert_eq!(
+        fs::read_to_string(&target_path)?,
+        "line1\nchanged2\nline3\nchanged4\n"
+    );
+
+    Ok(())
+}
+
+#[test]
+fn test_apply_patch_cli_moves_file_to_new_directory() -> anyhow::Result<()> {
+    let tmp = tempdir()?;
+    let original_path = tmp.path().join("old/name.txt");
+    let new_path = tmp.path().join("renamed/dir/name.txt");
+    fs::create_dir_all(original_path.parent().expect("parent should exist"))?;
+    fs::write(&original_path, "old content\n")?;
+
+    let patch = "*** Begin Patch\n*** Update File: old/name.txt\n*** Move to: renamed/dir/name.txt\n@@\n-old content\n+new content\n*** End Patch";
+
+    run_apply_patch_in_dir(tmp.path(), patch)?
+        .success()
+        .stdout("Success. Updated the following files:\nM renamed/dir/name.txt\n");
+
+    assert!(!original_path.exists());
+    assert_eq!(fs::read_to_string(&new_path)?, "new content\n");
+
+    Ok(())
+}
+
+#[test]
+fn test_apply_patch_cli_rejects_empty_patch() -> anyhow::Result<()> {
+    let tmp = tempdir()?;
+
+    apply_patch_command(tmp.path())?
+        .arg("*** Begin Patch\n*** End Patch")
+        .assert()
+        .failure()
+        .stderr("No files were modified.\n");
+
+    Ok(())
+}
+
+#[test]
+fn test_apply_patch_cli_reports_missing_context() -> anyhow::Result<()> {
+    let tmp = tempdir()?;
+    let target_path = tmp.path().join("modify.txt");
+    fs::write(&target_path, "line1\nline2\n")?;
+
+    apply_patch_command(tmp.path())?
+        .arg("*** Begin Patch\n*** Update File: modify.txt\n@@\n-missing\n+changed\n*** End Patch")
+        .assert()
+        .failure()
+        .stderr("Failed to find expected lines in modify.txt:\nmissing\n");
+    assert_eq!(fs::read_to_string(&target_path)?, "line1\nline2\n");
+
+    Ok(())
+}
+
+#[test]
+fn test_apply_patch_cli_rejects_missing_file_delete() -> anyhow::Result<()> {
+    let tmp = tempdir()?;
+
+    apply_patch_command(tmp.path())?
+        .arg("*** Begin Patch\n*** Delete File: missing.txt\n*** End Patch")
+        .assert()
+        .failure()
+        .stderr("Failed to delete file missing.txt\n");
+
+    Ok(())
+}
+
+#[test]
+fn test_apply_patch_cli_rejects_empty_update_hunk() -> anyhow::Result<()> {
+    let tmp = tempdir()?;
+
+    apply_patch_command(tmp.path())?
+        .arg("*** Begin Patch\n*** Update File: foo.txt\n*** End Patch")
+        .assert()
+        .failure()
+        .stderr("Invalid patch hunk on line 2: Update file hunk for path 'foo.txt' is empty\n");
+
+    Ok(())
+}
+
+#[test]
+fn test_apply_patch_cli_requires_existing_file_for_update() -> anyhow::Result<()> {
+    let tmp = tempdir()?;
+
+    apply_patch_command(tmp.path())?
+        .arg("*** Begin Patch\n*** Update File: missing.txt\n@@\n-old\n+new\n*** End Patch")
+        .assert()
+        .failure()
+        .stderr(
+            "Failed to read file to update missing.txt: No such file or directory (os error 2)\n",
+        );
+
+    Ok(())
+}
+
+#[test]
+fn test_apply_patch_cli_move_overwrites_existing_destination() -> anyhow::Result<()> {
+    let tmp = tempdir()?;
+    let original_path = tmp.path().join("old/name.txt");
+    let destination = tmp.path().join("renamed/dir/name.txt");
+    fs::create_dir_all(original_path.parent().expect("parent should exist"))?;
+    fs::create_dir_all(destination.parent().expect("parent should exist"))?;
+    fs::write(&original_path, "from\n")?;
+    fs::write(&destination, "existing\n")?;
+
+    run_apply_patch_in_dir(
+        tmp.path(),
+        "*** Begin Patch\n*** Update File: old/name.txt\n*** Move to: renamed/dir/name.txt\n@@\n-from\n+new\n*** End Patch",
+    )?
+    .success()
+    .stdout("Success. Updated the following files:\nM renamed/dir/name.txt\n");
+
+    assert!(!original_path.exists());
+    assert_eq!(fs::read_to_string(&destination)?, "new\n");
+
+    Ok(())
+}
+
+#[test]
+fn test_apply_patch_cli_add_overwrites_existing_file() -> anyhow::Result<()> {
+    let tmp = tempdir()?;
+    let path = tmp.path().join("duplicate.txt");
+    fs::write(&path, "old content\n")?;
+
+    run_apply_patch_in_dir(
+        tmp.path(),
+        "*** Begin Patch\n*** Add File: duplicate.txt\n+new content\n*** End Patch",
+    )?
+    .success()
+    .stdout("Success. Updated the following files:\nA duplicate.txt\n");
+
+    assert_eq!(fs::read_to_string(&path)?, "new content\n");
+
+    Ok(())
+}
+
+#[test]
+fn test_apply_patch_cli_delete_directory_fails() -> anyhow::Result<()> {
+    let tmp = tempdir()?;
+    fs::create_dir(tmp.path().join("dir"))?;
+
+    apply_patch_command(tmp.path())?
+        .arg("*** Begin Patch\n*** Delete File: dir\n*** End Patch")
+        .assert()
+        .failure()
+        .stderr("Failed to delete file dir\n");
+
+    Ok(())
+}
+
+#[test]
+fn test_apply_patch_cli_rejects_invalid_hunk_header() -> anyhow::Result<()> {
+    let tmp = tempdir()?;
+
+    apply_patch_command(tmp.path())?
+        .arg("*** Begin Patch\n*** Frobnicate File: foo\n*** End Patch")
+        .assert()
+        .failure()
+        .stderr("Invalid patch hunk on line 2: '*** Frobnicate File: foo' is not a valid hunk header. Valid hunk headers: '*** Add File: {path}', '*** Delete File: {path}', '*** Update File: {path}'\n");
+
+    Ok(())
+}
+
+#[test]
+fn test_apply_patch_cli_updates_file_appends_trailing_newline() -> anyhow::Result<()> {
+    let tmp = tempdir()?;
+    let target_path = tmp.path().join("no_newline.txt");
+    fs::write(&target_path, "no newline at end")?;
+
+    run_apply_patch_in_dir(
+        tmp.path(),
+        "*** Begin Patch\n*** Update File: no_newline.txt\n@@\n-no newline at end\n+first line\n+second line\n*** End Patch",
+    )?
+    .success()
+    .stdout("Success. Updated the following files:\nM no_newline.txt\n");
+
+    let contents = fs::read_to_string(&target_path)?;
+    assert!(contents.ends_with('\n'));
+    assert_eq!(contents, "first line\nsecond line\n");
+
+    Ok(())
+}
+
+#[test]
+fn test_apply_patch_cli_failure_after_partial_success_leaves_changes() -> anyhow::Result<()> {
+    let tmp = tempdir()?;
+    let new_file = tmp.path().join("created.txt");
+
+    apply_patch_command(tmp.path())?
+        .arg("*** Begin Patch\n*** Add File: created.txt\n+hello\n*** Update File: missing.txt\n@@\n-old\n+new\n*** End Patch")
+        .assert()
+        .failure()
+        .stdout("")
+        .stderr("Failed to read file to update missing.txt: No such file or directory (os error 2)\n");
+
+    assert_eq!(fs::read_to_string(&new_file)?, "hello\n");
+
+    Ok(())
+}

codex-rs/async-utils/Cargo.toml

@@ -0,0 +1,15 @@
+[package]
+edition.workspace = true
+name = "codex-async-utils"
+version.workspace = true
+
+[lints]
+workspace = true
+
+[dependencies]
+async-trait.workspace = true
+tokio = { workspace = true, features = ["macros", "rt", "rt-multi-thread", "time"] }
+tokio-util.workspace = true
+
+[dev-dependencies]
+pretty_assertions.workspace = true

codex-rs/async-utils/src/lib.rs

@@ -0,0 +1,86 @@
+use async_trait::async_trait;
+use std::future::Future;
+use tokio_util::sync::CancellationToken;
+
+#[derive(Debug, PartialEq, Eq)]
+pub enum CancelErr {
+    Cancelled,
+}
+
+#[async_trait]
+pub trait OrCancelExt: Sized {
+    type Output;
+
+    async fn or_cancel(self, token: &CancellationToken) -> Result<Self::Output, CancelErr>;
+}
+
+#[async_trait]
+impl<F> OrCancelExt for F
+where
+    F: Future + Send,
+    F::Output: Send,
+{
+    type Output = F::Output;
+
+    async fn or_cancel(self, token: &CancellationToken) -> Result<Self::Output, CancelErr> {
+        tokio::select! {
+            _ = token.cancelled() => Err(CancelErr::Cancelled),
+            res = self => Ok(res),
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use pretty_assertions::assert_eq;
+    use std::time::Duration;
+    use tokio::task;
+    use tokio::time::sleep;
+
+    #[tokio::test]
+    async fn returns_ok_when_future_completes_first() {
+        let token = CancellationToken::new();
+        let value = async { 42 };
+
+        let result = value.or_cancel(&token).await;
+
+        assert_eq!(Ok(42), result);
+    }
+
+    #[tokio::test]
+    async fn returns_err_when_token_cancelled_first() {
+        let token = CancellationToken::new();
+        let token_clone = token.clone();
+
+        let cancel_handle = task::spawn(async move {
+            sleep(Duration::from_millis(10)).await;
+            token_clone.cancel();
+        });
+
+        let result = async {
+            sleep(Duration::from_millis(100)).await;
+            7
+        }
+        .or_cancel(&token)
+        .await;
+
+        cancel_handle.await.expect("cancel task panicked");
+        assert_eq!(Err(CancelErr::Cancelled), result);
+    }
+
+    #[tokio::test]
+    async fn returns_err_when_token_already_cancelled() {
+        let token = CancellationToken::new();
+        token.cancel();
+
+        let result = async {
+            sleep(Duration::from_millis(50)).await;
+            5
+        }
+        .or_cancel(&token)
+        .await;
+
+        assert_eq!(Err(CancelErr::Cancelled), result);
+    }
+}

codex-rs/backend-client/Cargo.toml

@@ -0,0 +1,20 @@
+[package]
+name = "codex-backend-client"
+version = "0.0.0"
+edition = "2024"
+publish = false
+
+[lib]
+path = "src/lib.rs"
+
+[dependencies]
+anyhow = "1"
+serde = { version = "1", features = ["derive"] }
+serde_json = "1"
+reqwest = { version = "0.12", default-features = false, features = ["json", "rustls-tls"] }
+codex-backend-openapi-models = { path = "../codex-backend-openapi-models" }
+codex-protocol = { workspace = true }
+codex-core = { workspace = true }
+
+[dev-dependencies]
+pretty_assertions = "1"

codex-rs/backend-client/src/client.rs

@@ -0,0 +1,317 @@
+use crate::types::CodeTaskDetailsResponse;
+use crate::types::PaginatedListTaskListItem;
+use crate::types::RateLimitStatusPayload;
+use crate::types::RateLimitWindowSnapshot;
+use crate::types::TurnAttemptsSiblingTurnsResponse;
+use anyhow::Result;
+use codex_core::auth::CodexAuth;
+use codex_core::default_client::get_codex_user_agent;
+use codex_protocol::protocol::RateLimitSnapshot;
+use codex_protocol::protocol::RateLimitWindow;
+use reqwest::header::AUTHORIZATION;
+use reqwest::header::CONTENT_TYPE;
+use reqwest::header::HeaderMap;
+use reqwest::header::HeaderName;
+use reqwest::header::HeaderValue;
+use reqwest::header::USER_AGENT;
+use serde::de::DeserializeOwned;
+
+#[derive(Clone, Copy, Debug, PartialEq, Eq)]
+pub enum PathStyle {
+    /// /api/codex/…
+    CodexApi,
+    /// /wham/…
+    ChatGptApi,
+}
+
+impl PathStyle {
+    pub fn from_base_url(base_url: &str) -> Self {
+        if base_url.contains("/backend-api") {
+            PathStyle::ChatGptApi
+        } else {
+            PathStyle::CodexApi
+        }
+    }
+}
+
+#[derive(Clone, Debug)]
+pub struct Client {
+    base_url: String,
+    http: reqwest::Client,
+    bearer_token: Option<String>,
+    user_agent: Option<HeaderValue>,
+    chatgpt_account_id: Option<String>,
+    path_style: PathStyle,
+}
+
+impl Client {
+    pub fn new(base_url: impl Into<String>) -> Result<Self> {
+        let mut base_url = base_url.into();
+        // Normalize common ChatGPT hostnames to include /backend-api so we hit the WHAM paths.
+        // Also trim trailing slashes for consistent URL building.
+        while base_url.ends_with('/') {
+            base_url.pop();
+        }
+        if (base_url.starts_with("https://chatgpt.com")
+            || base_url.starts_with("https://chat.openai.com"))
+            && !base_url.contains("/backend-api")
+        {
+            base_url = format!("{base_url}/backend-api");
+        }
+        let http = reqwest::Client::builder().build()?;
+        let path_style = PathStyle::from_base_url(&base_url);
+        Ok(Self {
+            base_url,
+            http,
+            bearer_token: None,
+            user_agent: None,
+            chatgpt_account_id: None,
+            path_style,
+        })
+    }
+
+    pub async fn from_auth(base_url: impl Into<String>, auth: &CodexAuth) -> Result<Self> {
+        let token = auth.get_token().await.map_err(anyhow::Error::from)?;
+        let mut client = Self::new(base_url)?
+            .with_user_agent(get_codex_user_agent())
+            .with_bearer_token(token);
+        if let Some(account_id) = auth.get_account_id() {
+            client = client.with_chatgpt_account_id(account_id);
+        }
+        Ok(client)
+    }
+
+    pub fn with_bearer_token(mut self, token: impl Into<String>) -> Self {
+        self.bearer_token = Some(token.into());
+        self
+    }
+
+    pub fn with_user_agent(mut self, ua: impl Into<String>) -> Self {
+        if let Ok(hv) = HeaderValue::from_str(&ua.into()) {
+            self.user_agent = Some(hv);
+        }
+        self
+    }
+
+    pub fn with_chatgpt_account_id(mut self, account_id: impl Into<String>) -> Self {
+        self.chatgpt_account_id = Some(account_id.into());
+        self
+    }
+
+    pub fn with_path_style(mut self, style: PathStyle) -> Self {
+        self.path_style = style;
+        self
+    }
+
+    fn headers(&self) -> HeaderMap {
+        let mut h = HeaderMap::new();
+        if let Some(ua) = &self.user_agent {
+            h.insert(USER_AGENT, ua.clone());
+        } else {
+            h.insert(USER_AGENT, HeaderValue::from_static("codex-cli"));
+        }
+        if let Some(token) = &self.bearer_token {
+            let value = format!("Bearer {token}");
+            if let Ok(hv) = HeaderValue::from_str(&value) {
+                h.insert(AUTHORIZATION, hv);
+            }
+        }
+        if let Some(acc) = &self.chatgpt_account_id
+            && let Ok(name) = HeaderName::from_bytes(b"ChatGPT-Account-Id")
+            && let Ok(hv) = HeaderValue::from_str(acc)
+        {
+            h.insert(name, hv);
+        }
+        h
+    }
+
+    async fn exec_request(
+        &self,
+        req: reqwest::RequestBuilder,
+        method: &str,
+        url: &str,
+    ) -> Result<(String, String)> {
+        let res = req.send().await?;
+        let status = res.status();
+        let ct = res
+            .headers()
+            .get(CONTENT_TYPE)
+            .and_then(|v| v.to_str().ok())
+            .unwrap_or("")
+            .to_string();
+        let body = res.text().await.unwrap_or_default();
+        if !status.is_success() {
+            anyhow::bail!("{method} {url} failed: {status}; content-type={ct}; body={body}");
+        }
+        Ok((body, ct))
+    }
+
+    fn decode_json<T: DeserializeOwned>(&self, url: &str, ct: &str, body: &str) -> Result<T> {
+        match serde_json::from_str::<T>(body) {
+            Ok(v) => Ok(v),
+            Err(e) => {
+                anyhow::bail!("Decode error for {url}: {e}; content-type={ct}; body={body}");
+            }
+        }
+    }
+
+    pub async fn get_rate_limits(&self) -> Result<RateLimitSnapshot> {
+        let url = match self.path_style {
+            PathStyle::CodexApi => format!("{}/api/codex/usage", self.base_url),
+            PathStyle::ChatGptApi => format!("{}/wham/usage", self.base_url),
+        };
+        let req = self.http.get(&url).headers(self.headers());
+        let (body, ct) = self.exec_request(req, "GET", &url).await?;
+        let payload: RateLimitStatusPayload = self.decode_json(&url, &ct, &body)?;
+        Ok(Self::rate_limit_snapshot_from_payload(payload))
+    }
+
+    pub async fn list_tasks(
+        &self,
+        limit: Option<i32>,
+        task_filter: Option<&str>,
+        environment_id: Option<&str>,
+    ) -> Result<PaginatedListTaskListItem> {
+        let url = match self.path_style {
+            PathStyle::CodexApi => format!("{}/api/codex/tasks/list", self.base_url),
+            PathStyle::ChatGptApi => format!("{}/wham/tasks/list", self.base_url),
+        };
+        let req = self.http.get(&url).headers(self.headers());
+        let req = if let Some(lim) = limit {
+            req.query(&[("limit", lim)])
+        } else {
+            req
+        };
+        let req = if let Some(tf) = task_filter {
+            req.query(&[("task_filter", tf)])
+        } else {
+            req
+        };
+        let req = if let Some(id) = environment_id {
+            req.query(&[("environment_id", id)])
+        } else {
+            req
+        };
+        let (body, ct) = self.exec_request(req, "GET", &url).await?;
+        self.decode_json::<PaginatedListTaskListItem>(&url, &ct, &body)
+    }
+
+    pub async fn get_task_details(&self, task_id: &str) -> Result<CodeTaskDetailsResponse> {
+        let (parsed, _body, _ct) = self.get_task_details_with_body(task_id).await?;
+        Ok(parsed)
+    }
+
+    pub async fn get_task_details_with_body(
+        &self,
+        task_id: &str,
+    ) -> Result<(CodeTaskDetailsResponse, String, String)> {
+        let url = match self.path_style {
+            PathStyle::CodexApi => format!("{}/api/codex/tasks/{}", self.base_url, task_id),
+            PathStyle::ChatGptApi => format!("{}/wham/tasks/{}", self.base_url, task_id),
+        };
+        let req = self.http.get(&url).headers(self.headers());
+        let (body, ct) = self.exec_request(req, "GET", &url).await?;
+        let parsed: CodeTaskDetailsResponse = self.decode_json(&url, &ct, &body)?;
+        Ok((parsed, body, ct))
+    }
+
+    pub async fn list_sibling_turns(
+        &self,
+        task_id: &str,
+        turn_id: &str,
+    ) -> Result<TurnAttemptsSiblingTurnsResponse> {
+        let url = match self.path_style {
+            PathStyle::CodexApi => format!(
+                "{}/api/codex/tasks/{}/turns/{}/sibling_turns",
+                self.base_url, task_id, turn_id
+            ),
+            PathStyle::ChatGptApi => format!(
+                "{}/wham/tasks/{}/turns/{}/sibling_turns",
+                self.base_url, task_id, turn_id
+            ),
+        };
+        let req = self.http.get(&url).headers(self.headers());
+        let (body, ct) = self.exec_request(req, "GET", &url).await?;
+        self.decode_json::<TurnAttemptsSiblingTurnsResponse>(&url, &ct, &body)
+    }
+
+    /// Create a new task (user turn) by POSTing to the appropriate backend path
+    /// based on `path_style`. Returns the created task id.
+    pub async fn create_task(&self, request_body: serde_json::Value) -> Result<String> {
+        let url = match self.path_style {
+            PathStyle::CodexApi => format!("{}/api/codex/tasks", self.base_url),
+            PathStyle::ChatGptApi => format!("{}/wham/tasks", self.base_url),
+        };
+        let req = self
+            .http
+            .post(&url)
+            .headers(self.headers())
+            .header(CONTENT_TYPE, HeaderValue::from_static("application/json"))
+            .json(&request_body);
+        let (body, ct) = self.exec_request(req, "POST", &url).await?;
+        // Extract id from JSON: prefer `task.id`; fallback to top-level `id` when present.
+        match serde_json::from_str::<serde_json::Value>(&body) {
+            Ok(v) => {
+                if let Some(id) = v
+                    .get("task")
+                    .and_then(|t| t.get("id"))
+                    .and_then(|s| s.as_str())
+                {
+                    Ok(id.to_string())
+                } else if let Some(id) = v.get("id").and_then(|s| s.as_str()) {
+                    Ok(id.to_string())
+                } else {
+                    anyhow::bail!(
+                        "POST {url} succeeded but no task id found; content-type={ct}; body={body}"
+                    );
+                }
+            }
+            Err(e) => anyhow::bail!("Decode error for {url}: {e}; content-type={ct}; body={body}"),
+        }
+    }
+
+    // rate limit helpers
+    fn rate_limit_snapshot_from_payload(payload: RateLimitStatusPayload) -> RateLimitSnapshot {
+        let Some(details) = payload
+            .rate_limit
+            .and_then(|inner| inner.map(|boxed| *boxed))
+        else {
+            return RateLimitSnapshot {
+                primary: None,
+                secondary: None,
+            };
+        };
+
+        RateLimitSnapshot {
+            primary: Self::map_rate_limit_window(details.primary_window),
+            secondary: Self::map_rate_limit_window(details.secondary_window),
+        }
+    }
+
+    fn map_rate_limit_window(
+        window: Option<Option<Box<RateLimitWindowSnapshot>>>,
+    ) -> Option<RateLimitWindow> {
+        let snapshot = match window {
+            Some(Some(snapshot)) => *snapshot,
+            _ => return None,
+        };
+
+        let used_percent = f64::from(snapshot.used_percent);
+        let window_minutes = Self::window_minutes_from_seconds(snapshot.limit_window_seconds);
+        let resets_at = Some(i64::from(snapshot.reset_at));
+        Some(RateLimitWindow {
+            used_percent,
+            window_minutes,
+            resets_at,
+        })
+    }
+
+    fn window_minutes_from_seconds(seconds: i32) -> Option<i64> {
+        if seconds <= 0 {
+            return None;
+        }
+
+        let seconds_i64 = i64::from(seconds);
+        Some((seconds_i64 + 59) / 60)
+    }
+}

codex-rs/backend-client/src/lib.rs

@@ -0,0 +1,9 @@
+mod client;
+pub mod types;
+
+pub use client::Client;
+pub use types::CodeTaskDetailsResponse;
+pub use types::CodeTaskDetailsResponseExt;
+pub use types::PaginatedListTaskListItem;
+pub use types::TaskListItem;
+pub use types::TurnAttemptsSiblingTurnsResponse;

codex-rs/backend-client/src/types.rs

@@ -0,0 +1,373 @@
+pub use codex_backend_openapi_models::models::PaginatedListTaskListItem;
+pub use codex_backend_openapi_models::models::PlanType;
+pub use codex_backend_openapi_models::models::RateLimitStatusDetails;
+pub use codex_backend_openapi_models::models::RateLimitStatusPayload;
+pub use codex_backend_openapi_models::models::RateLimitWindowSnapshot;
+pub use codex_backend_openapi_models::models::TaskListItem;
+
+use serde::Deserialize;
+use serde::de::Deserializer;
+use serde_json::Value;
+use std::collections::HashMap;
+
+/// Hand-rolled models for the Cloud Tasks task-details response.
+/// The generated OpenAPI models are pretty bad. This is a half-step
+/// towards hand-rolling them.
+#[derive(Clone, Debug, Deserialize)]
+pub struct CodeTaskDetailsResponse {
+    #[serde(default)]
+    pub current_user_turn: Option<Turn>,
+    #[serde(default)]
+    pub current_assistant_turn: Option<Turn>,
+    #[serde(default)]
+    pub current_diff_task_turn: Option<Turn>,
+}
+
+#[derive(Clone, Debug, Default, Deserialize)]
+pub struct Turn {
+    #[serde(default)]
+    pub id: Option<String>,
+    #[serde(default)]
+    pub attempt_placement: Option<i64>,
+    #[serde(default, rename = "turn_status")]
+    pub turn_status: Option<String>,
+    #[serde(default, deserialize_with = "deserialize_vec")]
+    pub sibling_turn_ids: Vec<String>,
+    #[serde(default, deserialize_with = "deserialize_vec")]
+    pub input_items: Vec<TurnItem>,
+    #[serde(default, deserialize_with = "deserialize_vec")]
+    pub output_items: Vec<TurnItem>,
+    #[serde(default)]
+    pub worklog: Option<Worklog>,
+    #[serde(default)]
+    pub error: Option<TurnError>,
+}
+
+#[derive(Clone, Debug, Default, Deserialize)]
+pub struct TurnItem {
+    #[serde(rename = "type", default)]
+    pub kind: String,
+    #[serde(default)]
+    pub role: Option<String>,
+    #[serde(default, deserialize_with = "deserialize_vec")]
+    pub content: Vec<ContentFragment>,
+    #[serde(default)]
+    pub diff: Option<String>,
+    #[serde(default)]
+    pub output_diff: Option<DiffPayload>,
+}
+
+#[derive(Clone, Debug, Deserialize)]
+#[serde(untagged)]
+pub enum ContentFragment {
+    Structured(StructuredContent),
+    Text(String),
+}
+
+#[derive(Clone, Debug, Default, Deserialize)]
+pub struct StructuredContent {
+    #[serde(rename = "content_type", default)]
+    pub content_type: Option<String>,
+    #[serde(default)]
+    pub text: Option<String>,
+}
+
+#[derive(Clone, Debug, Default, Deserialize)]
+pub struct DiffPayload {
+    #[serde(default)]
+    pub diff: Option<String>,
+}
+
+#[derive(Clone, Debug, Default, Deserialize)]
+pub struct Worklog {
+    #[serde(default, deserialize_with = "deserialize_vec")]
+    pub messages: Vec<WorklogMessage>,
+}
+
+#[derive(Clone, Debug, Default, Deserialize)]
+pub struct WorklogMessage {
+    #[serde(default)]
+    pub author: Option<Author>,
+    #[serde(default)]
+    pub content: Option<WorklogContent>,
+}
+
+#[derive(Clone, Debug, Default, Deserialize)]
+pub struct Author {
+    #[serde(default)]
+    pub role: Option<String>,
+}
+
+#[derive(Clone, Debug, Default, Deserialize)]
+pub struct WorklogContent {
+    #[serde(default)]
+    pub parts: Vec<ContentFragment>,
+}
+
+#[derive(Clone, Debug, Default, Deserialize)]
+pub struct TurnError {
+    #[serde(default)]
+    pub code: Option<String>,
+    #[serde(default)]
+    pub message: Option<String>,
+}
+
+impl ContentFragment {
+    fn text(&self) -> Option<&str> {
+        match self {
+            ContentFragment::Structured(inner) => {
+                if inner
+                    .content_type
+                    .as_deref()
+                    .map(|ct| ct.eq_ignore_ascii_case("text"))
+                    .unwrap_or(false)
+                {
+                    inner.text.as_deref().filter(|s| !s.is_empty())
+                } else {
+                    None
+                }
+            }
+            ContentFragment::Text(raw) => {
+                if raw.trim().is_empty() {
+                    None
+                } else {
+                    Some(raw.as_str())
+                }
+            }
+        }
+    }
+}
+
+impl TurnItem {
+    fn text_values(&self) -> Vec<String> {
+        self.content
+            .iter()
+            .filter_map(|fragment| fragment.text().map(str::to_string))
+            .collect()
+    }
+
+    fn diff_text(&self) -> Option<String> {
+        if self.kind == "output_diff" {
+            if let Some(diff) = &self.diff
+                && !diff.is_empty()
+            {
+                return Some(diff.clone());
+            }
+        } else if self.kind == "pr"
+            && let Some(payload) = &self.output_diff
+            && let Some(diff) = &payload.diff
+            && !diff.is_empty()
+        {
+            return Some(diff.clone());
+        }
+        None
+    }
+}
+
+impl Turn {
+    fn unified_diff(&self) -> Option<String> {
+        self.output_items.iter().find_map(TurnItem::diff_text)
+    }
+
+    fn message_texts(&self) -> Vec<String> {
+        let mut out: Vec<String> = self
+            .output_items
+            .iter()
+            .filter(|item| item.kind == "message")
+            .flat_map(TurnItem::text_values)
+            .collect();
+
+        if let Some(log) = &self.worklog {
+            for message in &log.messages {
+                if message.is_assistant() {
+                    out.extend(message.text_values());
+                }
+            }
+        }
+
+        out
+    }
+
+    fn user_prompt(&self) -> Option<String> {
+        let parts: Vec<String> = self
+            .input_items
+            .iter()
+            .filter(|item| item.kind == "message")
+            .filter(|item| {
+                item.role
+                    .as_deref()
+                    .map(|r| r.eq_ignore_ascii_case("user"))
+                    .unwrap_or(true)
+            })
+            .flat_map(TurnItem::text_values)
+            .collect();
+
+        if parts.is_empty() {
+            None
+        } else {
+            Some(parts.join(
+                "
+
+",
+            ))
+        }
+    }
+
+    fn error_summary(&self) -> Option<String> {
+        self.error.as_ref().and_then(TurnError::summary)
+    }
+}
+
+impl WorklogMessage {
+    fn is_assistant(&self) -> bool {
+        self.author
+            .as_ref()
+            .and_then(|a| a.role.as_deref())
+            .map(|role| role.eq_ignore_ascii_case("assistant"))
+            .unwrap_or(false)
+    }
+
+    fn text_values(&self) -> Vec<String> {
+        self.content
+            .as_ref()
+            .map(|content| {
+                content
+                    .parts
+                    .iter()
+                    .filter_map(|fragment| fragment.text().map(str::to_string))
+                    .collect()
+            })
+            .unwrap_or_default()
+    }
+}
+
+impl TurnError {
+    fn summary(&self) -> Option<String> {
+        let code = self.code.as_deref().unwrap_or("");
+        let message = self.message.as_deref().unwrap_or("");
+        match (code.is_empty(), message.is_empty()) {
+            (true, true) => None,
+            (false, true) => Some(code.to_string()),
+            (true, false) => Some(message.to_string()),
+            (false, false) => Some(format!("{code}: {message}")),
+        }
+    }
+}
+
+pub trait CodeTaskDetailsResponseExt {
+    /// Attempt to extract a unified diff string from the assistant or diff turn.
+    fn unified_diff(&self) -> Option<String>;
+    /// Extract assistant text output messages (no diff) from current turns.
+    fn assistant_text_messages(&self) -> Vec<String>;
+    /// Extract the user's prompt text from the current user turn, when present.
+    fn user_text_prompt(&self) -> Option<String>;
+    /// Extract an assistant error message (if the turn failed and provided one).
+    fn assistant_error_message(&self) -> Option<String>;
+}
+
+impl CodeTaskDetailsResponseExt for CodeTaskDetailsResponse {
+    fn unified_diff(&self) -> Option<String> {
+        [
+            self.current_diff_task_turn.as_ref(),
+            self.current_assistant_turn.as_ref(),
+        ]
+        .into_iter()
+        .flatten()
+        .find_map(Turn::unified_diff)
+    }
+
+    fn assistant_text_messages(&self) -> Vec<String> {
+        let mut out = Vec::new();
+        for turn in [
+            self.current_diff_task_turn.as_ref(),
+            self.current_assistant_turn.as_ref(),
+        ]
+        .into_iter()
+        .flatten()
+        {
+            out.extend(turn.message_texts());
+        }
+        out
+    }
+
+    fn user_text_prompt(&self) -> Option<String> {
+        self.current_user_turn.as_ref().and_then(Turn::user_prompt)
+    }
+
+    fn assistant_error_message(&self) -> Option<String> {
+        self.current_assistant_turn
+            .as_ref()
+            .and_then(Turn::error_summary)
+    }
+}
+
+fn deserialize_vec<'de, D, T>(deserializer: D) -> Result<Vec<T>, D::Error>
+where
+    D: Deserializer<'de>,
+    T: Deserialize<'de>,
+{
+    Option::<Vec<T>>::deserialize(deserializer).map(|opt| opt.unwrap_or_default())
+}
+
+#[derive(Clone, Debug, Deserialize)]
+pub struct TurnAttemptsSiblingTurnsResponse {
+    #[serde(default)]
+    pub sibling_turns: Vec<HashMap<String, Value>>,
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use pretty_assertions::assert_eq;
+
+    fn fixture(name: &str) -> CodeTaskDetailsResponse {
+        let json = match name {
+            "diff" => include_str!("../tests/fixtures/task_details_with_diff.json"),
+            "error" => include_str!("../tests/fixtures/task_details_with_error.json"),
+            other => panic!("unknown fixture {other}"),
+        };
+        serde_json::from_str(json).expect("fixture should deserialize")
+    }
+
+    #[test]
+    fn unified_diff_prefers_current_diff_task_turn() {
+        let details = fixture("diff");
+        let diff = details.unified_diff().expect("diff present");
+        assert!(diff.contains("diff --git"));
+    }
+
+    #[test]
+    fn unified_diff_falls_back_to_pr_output_diff() {
+        let details = fixture("error");
+        let diff = details.unified_diff().expect("diff from pr output");
+        assert!(diff.contains("lib.rs"));
+    }
+
+    #[test]
+    fn assistant_text_messages_extracts_text_content() {
+        let details = fixture("diff");
+        let messages = details.assistant_text_messages();
+        assert_eq!(messages, vec!["Assistant response".to_string()]);
+    }
+
+    #[test]
+    fn user_text_prompt_joins_parts_with_spacing() {
+        let details = fixture("diff");
+        let prompt = details.user_text_prompt().expect("prompt present");
+        assert_eq!(
+            prompt,
+            "First line
+
+Second line"
+        );
+    }
+
+    #[test]
+    fn assistant_error_message_combines_code_and_message() {
+        let details = fixture("error");
+        let msg = details
+            .assistant_error_message()
+            .expect("error should be present");
+        assert_eq!(msg, "APPLY_FAILED: Patch could not be applied");
+    }
+}

codex-rs/backend-client/tests/fixtures/task_details_with_diff.json

@@ -0,0 +1,38 @@
+{
+  "task": {
+    "id": "task_123",
+    "title": "Refactor cloud task client",
+    "archived": false,
+    "external_pull_requests": []
+  },
+  "current_user_turn": {
+    "input_items": [
+      {
+        "type": "message",
+        "role": "user",
+        "content": [
+          { "content_type": "text", "text": "First line" },
+          { "content_type": "text", "text": "Second line" }
+        ]
+      }
+    ]
+  },
+  "current_assistant_turn": {
+    "output_items": [
+      {
+        "type": "message",
+        "content": [
+          { "content_type": "text", "text": "Assistant response" }
+        ]
+      }
+    ]
+  },
+  "current_diff_task_turn": {
+    "output_items": [
+      {
+        "type": "output_diff",
+        "diff": "diff --git a/src/main.rs b/src/main.rs\n+fn main() { println!(\"hi\"); }\n"
+      }
+    ]
+  }
+}

codex-rs/backend-client/tests/fixtures/task_details_with_error.json

@@ -0,0 +1,22 @@
+{
+  "task": {
+    "id": "task_456",
+    "title": "Investigate failure",
+    "archived": false,
+    "external_pull_requests": []
+  },
+  "current_assistant_turn": {
+    "output_items": [
+      {
+        "type": "pr",
+        "output_diff": {
+          "diff": "diff --git a/lib.rs b/lib.rs\n+pub fn hello() {}\n"
+        }
+      }
+    ],
+    "error": {
+      "code": "APPLY_FAILED",
+      "message": "Patch could not be applied"
+    }
+  }
+}

codex-rs/chatgpt/Cargo.toml

@@ -14,6 +14,7 @@ codex-core = { workspace = true }
 serde = { workspace = true, features = ["derive"] }
 serde_json = { workspace = true }
 tokio = { workspace = true, features = ["full"] }
+codex-git = { workspace = true }
 
 [dev-dependencies]
 tempfile = { workspace = true }

codex-rs/chatgpt/src/apply_command.rs

@@ -29,9 +29,11 @@ pub async fn run_apply_command(
             .parse_overrides()
             .map_err(anyhow::Error::msg)?,
         ConfigOverrides::default(),
-    )?;
+    )
+    .await?;
 
-    init_chatgpt_token_from_auth(&config.codex_home).await?;
+    init_chatgpt_token_from_auth(&config.codex_home, config.cli_auth_credentials_store_mode)
+        .await?;
 
     let task_response = get_task(&config, apply_cli.task_id).await?;
     apply_diff_from_task(task_response, cwd).await
@@ -56,46 +58,24 @@ pub async fn apply_diff_from_task(
 }
 
 async fn apply_diff(diff: &str, cwd: Option<PathBuf>) -> anyhow::Result<()> {
-    let mut cmd = tokio::process::Command::new("git");
-    if let Some(cwd) = cwd {
-        cmd.current_dir(cwd);
-    }
-    let toplevel_output = cmd
-        .args(vec!["rev-parse", "--show-toplevel"])
-        .output()
-        .await?;
-
-    if !toplevel_output.status.success() {
-        anyhow::bail!("apply must be run from a git repository.");
-    }
-
-    let repo_root = String::from_utf8(toplevel_output.stdout)?
-        .trim()
-        .to_string();
-
-    let mut git_apply_cmd = tokio::process::Command::new("git")
-        .args(vec!["apply", "--3way"])
-        .current_dir(&repo_root)
-        .stdin(std::process::Stdio::piped())
-        .stdout(std::process::Stdio::piped())
-        .stderr(std::process::Stdio::piped())
-        .spawn()?;
-
-    if let Some(mut stdin) = git_apply_cmd.stdin.take() {
-        tokio::io::AsyncWriteExt::write_all(&mut stdin, diff.as_bytes()).await?;
-        drop(stdin);
-    }
-
-    let output = git_apply_cmd.wait_with_output().await?;
-
-    if !output.status.success() {
+    let cwd = cwd.unwrap_or(std::env::current_dir().unwrap_or_else(|_| std::env::temp_dir()));
+    let req = codex_git::ApplyGitRequest {
+        cwd,
+        diff: diff.to_string(),
+        revert: false,
+        preflight: false,
+    };
+    let res = codex_git::apply_git_patch(&req)?;
+    if res.exit_code != 0 {
         anyhow::bail!(
-            "Git apply failed with status {}: {}",
-            output.status,
-            String::from_utf8_lossy(&output.stderr)
+            "Git apply failed (applied={}, skipped={}, conflicts={})\nstdout:\n{}\nstderr:\n{}",
+            res.applied_paths.len(),
+            res.skipped_paths.len(),
+            res.conflicted_paths.len(),
+            res.stdout,
+            res.stderr
         );
     }
-
     println!("Successfully applied diff");
     Ok(())
 }

codex-rs/chatgpt/src/chatgpt_client.rs

@@ -13,7 +13,8 @@ pub(crate) async fn chatgpt_get_request<T: DeserializeOwned>(
     path: String,
 ) -> anyhow::Result<T> {
     let chatgpt_base_url = &config.chatgpt_base_url;
-    init_chatgpt_token_from_auth(&config.codex_home).await?;
+    init_chatgpt_token_from_auth(&config.codex_home, config.cli_auth_credentials_store_mode)
+        .await?;
 
     // Make direct HTTP request to ChatGPT backend API with the token
     let client = create_client();
@@ -44,6 +45,6 @@ pub(crate) async fn chatgpt_get_request<T: DeserializeOwned>(
     } else {
         let status = response.status();
         let body = response.text().await.unwrap_or_default();
-        anyhow::bail!("Request failed with status {}: {}", status, body)
+        anyhow::bail!("Request failed with status {status}: {body}")
     }
 }

codex-rs/chatgpt/src/chatgpt_token.rs

@@ -3,6 +3,7 @@ use std::path::Path;
 use std::sync::LazyLock;
 use std::sync::RwLock;
 
+use codex_core::auth::AuthCredentialsStoreMode;
 use codex_core::token_data::TokenData;
 
 static CHATGPT_TOKEN: LazyLock<RwLock<Option<TokenData>>> = LazyLock::new(|| RwLock::new(None));
@@ -18,8 +19,11 @@ pub fn set_chatgpt_token_data(value: TokenData) {
 }
 
 /// Initialize the ChatGPT token from auth.json file
-pub async fn init_chatgpt_token_from_auth(codex_home: &Path) -> std::io::Result<()> {
-    let auth = CodexAuth::from_codex_home(codex_home)?;
+pub async fn init_chatgpt_token_from_auth(
+    codex_home: &Path,
+    auth_credentials_store_mode: AuthCredentialsStoreMode,
+) -> std::io::Result<()> {
+    let auth = CodexAuth::from_auth_storage(codex_home, auth_credentials_store_mode)?;
     if let Some(auth) = auth {
         let token_data = auth.get_token_data().await?;
         set_chatgpt_token_data(token_data);

codex-rs/cli/Cargo.toml

@@ -18,16 +18,22 @@ workspace = true
 anyhow = { workspace = true }
 clap = { workspace = true, features = ["derive"] }
 clap_complete = { workspace = true }
+codex-app-server = { workspace = true }
+codex-app-server-protocol = { workspace = true }
 codex-arg0 = { workspace = true }
 codex-chatgpt = { workspace = true }
+codex-cloud-tasks = { path = "../cloud-tasks" }
 codex-common = { workspace = true, features = ["cli"] }
 codex-core = { workspace = true }
 codex-exec = { workspace = true }
 codex-login = { workspace = true }
 codex-mcp-server = { workspace = true }
+codex-process-hardening = { workspace = true }
 codex-protocol = { workspace = true }
 codex-protocol-ts = { workspace = true }
 codex-responses-api-proxy = { workspace = true }
+codex-rmcp-client = { workspace = true }
+codex-stdio-to-uds = { workspace = true }
 codex-tui = { workspace = true }
 ctor = { workspace = true }
 owo-colors = { workspace = true }
@@ -40,20 +46,13 @@ tokio = { workspace = true, features = [
     "rt-multi-thread",
     "signal",
 ] }
-tracing = { workspace = true }
-tracing-subscriber = { workspace = true }
 
-[target.'cfg(target_os = "linux")'.dependencies]
-libc = { workspace = true }
-
-[target.'cfg(target_os = "android")'.dependencies]
-libc = { workspace = true }
-
-[target.'cfg(target_os = "macos")'.dependencies]
-libc = { workspace = true }
+[target.'cfg(target_os = "windows")'.dependencies]
+codex_windows_sandbox = { package = "codex-windows-sandbox", path = "../windows-sandbox-rs" }
 
 [dev-dependencies]
 assert_cmd = { workspace = true }
+assert_matches = { workspace = true }
 predicates = { workspace = true }
 pretty_assertions = { workspace = true }
 tempfile = { workspace = true }

codex-rs/cli/src/debug_sandbox.rs

@@ -11,6 +11,7 @@ use codex_protocol::config_types::SandboxMode;
 
 use crate::LandlockCommand;
 use crate::SeatbeltCommand;
+use crate::WindowsCommand;
 use crate::exit_status::handle_exit_status;
 
 pub async fn run_command_under_seatbelt(
@@ -51,9 +52,29 @@ pub async fn run_command_under_landlock(
     .await
 }
 
+pub async fn run_command_under_windows(
+    command: WindowsCommand,
+    codex_linux_sandbox_exe: Option<PathBuf>,
+) -> anyhow::Result<()> {
+    let WindowsCommand {
+        full_auto,
+        config_overrides,
+        command,
+    } = command;
+    run_command_under_sandbox(
+        full_auto,
+        command,
+        config_overrides,
+        codex_linux_sandbox_exe,
+        SandboxType::Windows,
+    )
+    .await
+}
+
 enum SandboxType {
     Seatbelt,
     Landlock,
+    Windows,
 }
 
 async fn run_command_under_sandbox(
@@ -73,7 +94,8 @@ async fn run_command_under_sandbox(
             codex_linux_sandbox_exe,
             ..Default::default()
         },
-    )?;
+    )
+    .await?;
 
     // In practice, this should be `std::env::current_dir()` because this CLI
     // does not support `--cwd`, but let's use the config value for consistency.
@@ -86,6 +108,63 @@ async fn run_command_under_sandbox(
     let stdio_policy = StdioPolicy::Inherit;
     let env = create_env(&config.shell_environment_policy);
 
+    // Special-case Windows sandbox: execute and exit the process to emulate inherited stdio.
+    if let SandboxType::Windows = sandbox_type {
+        #[cfg(target_os = "windows")]
+        {
+            use codex_windows_sandbox::run_windows_sandbox_capture;
+
+            let policy_str = match &config.sandbox_policy {
+                codex_core::protocol::SandboxPolicy::DangerFullAccess => "workspace-write",
+                codex_core::protocol::SandboxPolicy::ReadOnly => "read-only",
+                codex_core::protocol::SandboxPolicy::WorkspaceWrite { .. } => "workspace-write",
+            };
+
+            let sandbox_cwd = sandbox_policy_cwd.clone();
+            let cwd_clone = cwd.clone();
+            let env_map = env.clone();
+            let command_vec = command.clone();
+            let res = tokio::task::spawn_blocking(move || {
+                run_windows_sandbox_capture(
+                    policy_str,
+                    &sandbox_cwd,
+                    command_vec,
+                    &cwd_clone,
+                    env_map,
+                    None,
+                )
+            })
+            .await;
+
+            let capture = match res {
+                Ok(Ok(v)) => v,
+                Ok(Err(err)) => {
+                    eprintln!("windows sandbox failed: {err}");
+                    std::process::exit(1);
+                }
+                Err(join_err) => {
+                    eprintln!("windows sandbox join error: {join_err}");
+                    std::process::exit(1);
+                }
+            };
+
+            if !capture.stdout.is_empty() {
+                use std::io::Write;
+                let _ = std::io::stdout().write_all(&capture.stdout);
+            }
+            if !capture.stderr.is_empty() {
+                use std::io::Write;
+                let _ = std::io::stderr().write_all(&capture.stderr);
+            }
+
+            std::process::exit(capture.exit_code);
+        }
+        #[cfg(not(target_os = "windows"))]
+        {
+            anyhow::bail!("Windows sandbox is only available on Windows");
+        }
+    }
+
     let mut child = match sandbox_type {
         SandboxType::Seatbelt => {
             spawn_command_under_seatbelt(
@@ -114,6 +193,9 @@ async fn run_command_under_sandbox(
             )
             .await?
         }
+        SandboxType::Windows => {
+            unreachable!("Windows sandbox should have been handled above");
+        }
     };
     let status = child.wait().await?;
 

codex-rs/cli/src/lib.rs

@@ -1,7 +1,6 @@
 pub mod debug_sandbox;
 mod exit_status;
 pub mod login;
-pub mod proto;
 
 use clap::Parser;
 use codex_common::CliConfigOverrides;
@@ -33,3 +32,17 @@ pub struct LandlockCommand {
     #[arg(trailing_var_arg = true)]
     pub command: Vec<String>,
 }
+
+#[derive(Debug, Parser)]
+pub struct WindowsCommand {
+    /// Convenience alias for low-friction sandboxed automatic execution (network-disabled sandbox that can write to cwd and TMPDIR)
+    #[arg(long = "full-auto", default_value_t = false)]
+    pub full_auto: bool,
+
+    #[clap(skip)]
+    pub config_overrides: CliConfigOverrides,
+
+    /// Full command args to run under Windows restricted token sandbox.
+    #[arg(trailing_var_arg = true)]
+    pub command: Vec<String>,
+}

codex-rs/cli/src/login.rs

@@ -1,17 +1,31 @@
+use codex_app_server_protocol::AuthMode;
 use codex_common::CliConfigOverrides;
 use codex_core::CodexAuth;
+use codex_core::auth::AuthCredentialsStoreMode;
 use codex_core::auth::CLIENT_ID;
 use codex_core::auth::login_with_api_key;
 use codex_core::auth::logout;
 use codex_core::config::Config;
 use codex_core::config::ConfigOverrides;
 use codex_login::ServerOptions;
+use codex_login::run_device_code_login;
 use codex_login::run_login_server;
-use codex_protocol::mcp_protocol::AuthMode;
+use codex_protocol::config_types::ForcedLoginMethod;
+use std::io::IsTerminal;
+use std::io::Read;
 use std::path::PathBuf;
 
-pub async fn login_with_chatgpt(codex_home: PathBuf) -> std::io::Result<()> {
-    let opts = ServerOptions::new(codex_home, CLIENT_ID.to_string());
+pub async fn login_with_chatgpt(
+    codex_home: PathBuf,
+    forced_chatgpt_workspace_id: Option<String>,
+    cli_auth_credentials_store_mode: AuthCredentialsStoreMode,
+) -> std::io::Result<()> {
+    let opts = ServerOptions::new(
+        codex_home,
+        CLIENT_ID.to_string(),
+        forced_chatgpt_workspace_id,
+        cli_auth_credentials_store_mode,
+    );
     let server = run_login_server(opts)?;
 
     eprintln!(
@@ -23,9 +37,22 @@ pub async fn login_with_chatgpt(codex_home: PathBuf) -> std::io::Result<()> {
 }
 
 pub async fn run_login_with_chatgpt(cli_config_overrides: CliConfigOverrides) -> ! {
-    let config = load_config_or_exit(cli_config_overrides);
+    let config = load_config_or_exit(cli_config_overrides).await;
 
-    match login_with_chatgpt(config.codex_home).await {
+    if matches!(config.forced_login_method, Some(ForcedLoginMethod::Api)) {
+        eprintln!("ChatGPT login is disabled. Use API key login instead.");
+        std::process::exit(1);
+    }
+
+    let forced_chatgpt_workspace_id = config.forced_chatgpt_workspace_id.clone();
+
+    match login_with_chatgpt(
+        config.codex_home,
+        forced_chatgpt_workspace_id,
+        config.cli_auth_credentials_store_mode,
+    )
+    .await
+    {
         Ok(_) => {
             eprintln!("Successfully logged in");
             std::process::exit(0);
@@ -41,9 +68,18 @@ pub async fn run_login_with_api_key(
     cli_config_overrides: CliConfigOverrides,
     api_key: String,
 ) -> ! {
-    let config = load_config_or_exit(cli_config_overrides);
+    let config = load_config_or_exit(cli_config_overrides).await;
 
-    match login_with_api_key(&config.codex_home, &api_key) {
+    if matches!(config.forced_login_method, Some(ForcedLoginMethod::Chatgpt)) {
+        eprintln!("API key login is disabled. Use ChatGPT login instead.");
+        std::process::exit(1);
+    }
+
+    match login_with_api_key(
+        &config.codex_home,
+        &api_key,
+        config.cli_auth_credentials_store_mode,
+    ) {
         Ok(_) => {
             eprintln!("Successfully logged in");
             std::process::exit(0);
@@ -55,10 +91,70 @@ pub async fn run_login_with_api_key(
     }
 }
 
-pub async fn run_login_status(cli_config_overrides: CliConfigOverrides) -> ! {
-    let config = load_config_or_exit(cli_config_overrides);
+pub fn read_api_key_from_stdin() -> String {
+    let mut stdin = std::io::stdin();
 
-    match CodexAuth::from_codex_home(&config.codex_home) {
+    if stdin.is_terminal() {
+        eprintln!(
+            "--with-api-key expects the API key on stdin. Try piping it, e.g. `printenv OPENAI_API_KEY | codex login --with-api-key`."
+        );
+        std::process::exit(1);
+    }
+
+    eprintln!("Reading API key from stdin...");
+
+    let mut buffer = String::new();
+    if let Err(err) = stdin.read_to_string(&mut buffer) {
+        eprintln!("Failed to read API key from stdin: {err}");
+        std::process::exit(1);
+    }
+
+    let api_key = buffer.trim().to_string();
+    if api_key.is_empty() {
+        eprintln!("No API key provided via stdin.");
+        std::process::exit(1);
+    }
+
+    api_key
+}
+
+/// Login using the OAuth device code flow.
+pub async fn run_login_with_device_code(
+    cli_config_overrides: CliConfigOverrides,
+    issuer_base_url: Option<String>,
+    client_id: Option<String>,
+) -> ! {
+    let config = load_config_or_exit(cli_config_overrides).await;
+    if matches!(config.forced_login_method, Some(ForcedLoginMethod::Api)) {
+        eprintln!("ChatGPT login is disabled. Use API key login instead.");
+        std::process::exit(1);
+    }
+    let forced_chatgpt_workspace_id = config.forced_chatgpt_workspace_id.clone();
+    let mut opts = ServerOptions::new(
+        config.codex_home,
+        client_id.unwrap_or(CLIENT_ID.to_string()),
+        forced_chatgpt_workspace_id,
+        config.cli_auth_credentials_store_mode,
+    );
+    if let Some(iss) = issuer_base_url {
+        opts.issuer = iss;
+    }
+    match run_device_code_login(opts).await {
+        Ok(()) => {
+            eprintln!("Successfully logged in");
+            std::process::exit(0);
+        }
+        Err(e) => {
+            eprintln!("Error logging in with device code: {e}");
+            std::process::exit(1);
+        }
+    }
+}
+
+pub async fn run_login_status(cli_config_overrides: CliConfigOverrides) -> ! {
+    let config = load_config_or_exit(cli_config_overrides).await;
+
+    match CodexAuth::from_auth_storage(&config.codex_home, config.cli_auth_credentials_store_mode) {
         Ok(Some(auth)) => match auth.mode {
             AuthMode::ApiKey => match auth.get_token().await {
                 Ok(api_key) => {
@@ -87,9 +183,9 @@ pub async fn run_login_status(cli_config_overrides: CliConfigOverrides) -> ! {
 }
 
 pub async fn run_logout(cli_config_overrides: CliConfigOverrides) -> ! {
-    let config = load_config_or_exit(cli_config_overrides);
+    let config = load_config_or_exit(cli_config_overrides).await;
 
-    match logout(&config.codex_home) {
+    match logout(&config.codex_home, config.cli_auth_credentials_store_mode) {
         Ok(true) => {
             eprintln!("Successfully logged out");
             std::process::exit(0);
@@ -105,7 +201,7 @@ pub async fn run_logout(cli_config_overrides: CliConfigOverrides) -> ! {
     }
 }
 
-fn load_config_or_exit(cli_config_overrides: CliConfigOverrides) -> Config {
+async fn load_config_or_exit(cli_config_overrides: CliConfigOverrides) -> Config {
     let cli_overrides = match cli_config_overrides.parse_overrides() {
         Ok(v) => v,
         Err(e) => {
@@ -115,7 +211,7 @@ fn load_config_or_exit(cli_config_overrides: CliConfigOverrides) -> Config {
     };
 
     let config_overrides = ConfigOverrides::default();
-    match Config::load_with_cli_overrides(cli_overrides, config_overrides) {
+    match Config::load_with_cli_overrides(cli_overrides, config_overrides).await {
         Ok(config) => config,
         Err(e) => {
             eprintln!("Error loading configuration: {e}");

codex-rs/cli/src/main.rs

@@ -1,4 +1,3 @@
-use anyhow::Context;
 use clap::CommandFactory;
 use clap::Parser;
 use clap_complete::Shell;
@@ -8,25 +7,30 @@ use codex_chatgpt::apply_command::ApplyCommand;
 use codex_chatgpt::apply_command::run_apply_command;
 use codex_cli::LandlockCommand;
 use codex_cli::SeatbeltCommand;
+use codex_cli::WindowsCommand;
+use codex_cli::login::read_api_key_from_stdin;
 use codex_cli::login::run_login_status;
 use codex_cli::login::run_login_with_api_key;
 use codex_cli::login::run_login_with_chatgpt;
+use codex_cli::login::run_login_with_device_code;
 use codex_cli::login::run_logout;
-use codex_cli::proto;
+use codex_cloud_tasks::Cli as CloudTasksCli;
 use codex_common::CliConfigOverrides;
 use codex_exec::Cli as ExecCli;
 use codex_responses_api_proxy::Args as ResponsesApiProxyArgs;
 use codex_tui::AppExitInfo;
 use codex_tui::Cli as TuiCli;
+use codex_tui::updates::UpdateAction;
 use owo_colors::OwoColorize;
 use std::path::PathBuf;
 use supports_color::Stream;
 
 mod mcp_cmd;
-mod pre_main_hardening;
 
 use crate::mcp_cmd::McpCli;
-use crate::proto::ProtoCli;
+use codex_core::config::Config;
+use codex_core::config::ConfigOverrides;
+use codex_core::features::is_known_feature_key;
 
 /// Codex CLI
 ///
@@ -40,12 +44,16 @@ use crate::proto::ProtoCli;
     // The executable is sometimes invoked via a platform‑specific name like
     // `codex-x86_64-unknown-linux-musl`, but the help output should always use
     // the generic `codex` command name that users run.
-    bin_name = "codex"
+    bin_name = "codex",
+    override_usage = "codex [OPTIONS] [PROMPT]\n       codex [OPTIONS] <COMMAND> [ARGS]"
 )]
 struct MultitoolCli {
     #[clap(flatten)]
     pub config_overrides: CliConfigOverrides,
 
+    #[clap(flatten)]
+    pub feature_toggles: FeatureToggles,
+
     #[clap(flatten)]
     interactive: TuiCli,
 
@@ -68,15 +76,18 @@ enum Subcommand {
     /// [experimental] Run Codex as an MCP server and manage MCP servers.
     Mcp(McpCli),
 
-    /// Run the Protocol stream via stdin/stdout
-    #[clap(visible_alias = "p")]
-    Proto(ProtoCli),
+    /// [experimental] Run the Codex MCP server (stdio transport).
+    McpServer,
+
+    /// [experimental] Run the app server.
+    AppServer,
 
     /// Generate shell completion scripts.
     Completion(CompletionCommand),
 
-    /// Internal debugging commands.
-    Debug(DebugArgs),
+    /// Run commands within a Codex-provided sandbox.
+    #[clap(visible_alias = "debug")]
+    Sandbox(SandboxArgs),
 
     /// Apply the latest diff produced by Codex agent as a `git apply` to your local working tree.
     #[clap(visible_alias = "a")]
@@ -88,10 +99,20 @@ enum Subcommand {
     /// Internal: generate TypeScript protocol bindings.
     #[clap(hide = true)]
     GenerateTs(GenerateTsCommand),
+    /// [EXPERIMENTAL] Browse tasks from Codex Cloud and apply changes locally.
+    #[clap(name = "cloud", alias = "cloud-tasks")]
+    Cloud(CloudTasksCli),
 
     /// Internal: run the responses API proxy.
     #[clap(hide = true)]
     ResponsesApiProxy(ResponsesApiProxyArgs),
+
+    /// Internal: relay stdio to a Unix domain socket.
+    #[clap(hide = true, name = "stdio-to-uds")]
+    StdioToUds(StdioToUdsCommand),
+
+    /// Inspect feature flags.
+    Features(FeaturesCli),
 }
 
 #[derive(Debug, Parser)]
@@ -117,18 +138,23 @@ struct ResumeCommand {
 }
 
 #[derive(Debug, Parser)]
-struct DebugArgs {
+struct SandboxArgs {
     #[command(subcommand)]
-    cmd: DebugCommand,
+    cmd: SandboxCommand,
 }
 
 #[derive(Debug, clap::Subcommand)]
-enum DebugCommand {
+enum SandboxCommand {
     /// Run a command under Seatbelt (macOS only).
-    Seatbelt(SeatbeltCommand),
+    #[clap(visible_alias = "seatbelt")]
+    Macos(SeatbeltCommand),
 
     /// Run a command under Landlock+seccomp (Linux only).
-    Landlock(LandlockCommand),
+    #[clap(visible_alias = "landlock")]
+    Linux(LandlockCommand),
+
+    /// Run a command under Windows restricted token (Windows only).
+    Windows(WindowsCommand),
 }
 
 #[derive(Debug, Parser)]
@@ -136,9 +162,32 @@ struct LoginCommand {
     #[clap(skip)]
     config_overrides: CliConfigOverrides,
 
-    #[arg(long = "api-key", value_name = "API_KEY")]
+    #[arg(
+        long = "with-api-key",
+        help = "Read the API key from stdin (e.g. `printenv OPENAI_API_KEY | codex login --with-api-key`)"
+    )]
+    with_api_key: bool,
+
+    #[arg(
+        long = "api-key",
+        value_name = "API_KEY",
+        help = "(deprecated) Previously accepted the API key directly; now exits with guidance to use --with-api-key",
+        hide = true
+    )]
     api_key: Option<String>,
 
+    #[arg(long = "device-auth")]
+    use_device_code: bool,
+
+    /// EXPERIMENTAL: Use custom OAuth issuer base URL (advanced)
+    /// Override the OAuth issuer base URL (advanced)
+    #[arg(long = "experimental_issuer", value_name = "URL", hide = true)]
+    issuer_base_url: Option<String>,
+
+    /// EXPERIMENTAL: Use custom OAuth client ID (advanced)
+    #[arg(long = "experimental_client-id", value_name = "CLIENT_ID", hide = true)]
+    client_id: Option<String>,
+
     #[command(subcommand)]
     action: Option<LoginSubcommand>,
 }
@@ -166,10 +215,18 @@ struct GenerateTsCommand {
     prettier: Option<PathBuf>,
 }
 
+#[derive(Debug, Parser)]
+struct StdioToUdsCommand {
+    /// Path to the Unix domain socket to connect to.
+    #[arg(value_name = "SOCKET_PATH")]
+    socket_path: PathBuf,
+}
+
 fn format_exit_messages(exit_info: AppExitInfo, color_enabled: bool) -> Vec<String> {
     let AppExitInfo {
         token_usage,
         conversation_id,
+        ..
     } = exit_info;
 
     if token_usage.is_zero() {
@@ -188,45 +245,103 @@ fn format_exit_messages(exit_info: AppExitInfo, color_enabled: bool) -> Vec<Stri
         } else {
             resume_cmd
         };
-        lines.push(format!("To continue this session, run {command}."));
+        lines.push(format!("To continue this session, run {command}"));
     }
 
     lines
 }
 
-fn print_exit_messages(exit_info: AppExitInfo) {
+/// Handle the app exit and print the results. Optionally run the update action.
+fn handle_app_exit(exit_info: AppExitInfo) -> anyhow::Result<()> {
+    let update_action = exit_info.update_action;
     let color_enabled = supports_color::on(Stream::Stdout).is_some();
     for line in format_exit_messages(exit_info, color_enabled) {
         println!("{line}");
     }
+    if let Some(action) = update_action {
+        run_update_action(action)?;
+    }
+    Ok(())
 }
 
-pub(crate) const CODEX_SECURE_MODE_ENV_VAR: &str = "CODEX_SECURE_MODE";
+/// Run the update action and print the result.
+fn run_update_action(action: UpdateAction) -> anyhow::Result<()> {
+    println!();
+    let (cmd, args) = action.command_args();
+    let cmd_str = action.command_str();
+    println!("Updating Codex via `{cmd_str}`...");
+    let status = std::process::Command::new(cmd).args(args).status()?;
+    if !status.success() {
+        anyhow::bail!("`{cmd_str}` failed with status {status}");
+    }
+    println!();
+    println!("🎉 Update ran successfully! Please restart Codex.");
+    Ok(())
+}
 
-/// As early as possible in the process lifecycle, apply hardening measures
-/// if the CODEX_SECURE_MODE environment variable is set to "1".
+#[derive(Debug, Default, Parser, Clone)]
+struct FeatureToggles {
+    /// Enable a feature (repeatable). Equivalent to `-c features.<name>=true`.
+    #[arg(long = "enable", value_name = "FEATURE", action = clap::ArgAction::Append, global = true)]
+    enable: Vec<String>,
+
+    /// Disable a feature (repeatable). Equivalent to `-c features.<name>=false`.
+    #[arg(long = "disable", value_name = "FEATURE", action = clap::ArgAction::Append, global = true)]
+    disable: Vec<String>,
+}
+
+impl FeatureToggles {
+    fn to_overrides(&self) -> anyhow::Result<Vec<String>> {
+        let mut v = Vec::new();
+        for feature in &self.enable {
+            Self::validate_feature(feature)?;
+            v.push(format!("features.{feature}=true"));
+        }
+        for feature in &self.disable {
+            Self::validate_feature(feature)?;
+            v.push(format!("features.{feature}=false"));
+        }
+        Ok(v)
+    }
+
+    fn validate_feature(feature: &str) -> anyhow::Result<()> {
+        if is_known_feature_key(feature) {
+            Ok(())
+        } else {
+            anyhow::bail!("Unknown feature flag: {feature}")
+        }
+    }
+}
+
+#[derive(Debug, Parser)]
+struct FeaturesCli {
+    #[command(subcommand)]
+    sub: FeaturesSubcommand,
+}
+
+#[derive(Debug, Parser)]
+enum FeaturesSubcommand {
+    /// List known features with their stage and effective state.
+    List,
+}
+
+fn stage_str(stage: codex_core::features::Stage) -> &'static str {
+    use codex_core::features::Stage;
+    match stage {
+        Stage::Experimental => "experimental",
+        Stage::Beta => "beta",
+        Stage::Stable => "stable",
+        Stage::Deprecated => "deprecated",
+        Stage::Removed => "removed",
+    }
+}
+
+/// As early as possible in the process lifecycle, apply hardening measures. We
+/// skip this in debug builds to avoid interfering with debugging.
 #[ctor::ctor]
+#[cfg(not(debug_assertions))]
 fn pre_main_hardening() {
-    let secure_mode = match std::env::var(CODEX_SECURE_MODE_ENV_VAR) {
-        Ok(value) => value,
-        Err(_) => return,
-    };
-
-    if secure_mode == "1" {
-        #[cfg(any(target_os = "linux", target_os = "android"))]
-        crate::pre_main_hardening::pre_main_hardening_linux();
-
-        #[cfg(target_os = "macos")]
-        crate::pre_main_hardening::pre_main_hardening_macos();
-
-        #[cfg(windows)]
-        crate::pre_main_hardening::pre_main_hardening_windows();
-    }
-
-    // Always clear this env var so child processes don't inherit it.
-    unsafe {
-        std::env::remove_var(CODEX_SECURE_MODE_ENV_VAR);
-    }
+    codex_process_hardening::pre_main_hardening();
 }
 
 fn main() -> anyhow::Result<()> {
@@ -238,11 +353,16 @@ fn main() -> anyhow::Result<()> {
 
 async fn cli_main(codex_linux_sandbox_exe: Option<PathBuf>) -> anyhow::Result<()> {
     let MultitoolCli {
-        config_overrides: root_config_overrides,
+        config_overrides: mut root_config_overrides,
+        feature_toggles,
         mut interactive,
         subcommand,
     } = MultitoolCli::parse();
 
+    // Fold --enable/--disable into config overrides so they flow to all subcommands.
+    let toggle_overrides = feature_toggles.to_overrides()?;
+    root_config_overrides.raw_overrides.extend(toggle_overrides);
+
     match subcommand {
         None => {
             prepend_config_flags(
@@ -250,7 +370,7 @@ async fn cli_main(codex_linux_sandbox_exe: Option<PathBuf>) -> anyhow::Result<()
                 root_config_overrides.clone(),
             );
             let exit_info = codex_tui::run_main(interactive, codex_linux_sandbox_exe).await?;
-            print_exit_messages(exit_info);
+            handle_app_exit(exit_info)?;
         }
         Some(Subcommand::Exec(mut exec_cli)) => {
             prepend_config_flags(
@@ -259,10 +379,16 @@ async fn cli_main(codex_linux_sandbox_exe: Option<PathBuf>) -> anyhow::Result<()
             );
             codex_exec::run_main(exec_cli, codex_linux_sandbox_exe).await?;
         }
+        Some(Subcommand::McpServer) => {
+            codex_mcp_server::run_main(codex_linux_sandbox_exe, root_config_overrides).await?;
+        }
         Some(Subcommand::Mcp(mut mcp_cli)) => {
             // Propagate any root-level config overrides (e.g. `-c key=value`).
             prepend_config_flags(&mut mcp_cli.config_overrides, root_config_overrides.clone());
-            mcp_cli.run(codex_linux_sandbox_exe).await?;
+            mcp_cli.run().await?;
+        }
+        Some(Subcommand::AppServer) => {
+            codex_app_server::run_main(codex_linux_sandbox_exe, root_config_overrides).await?;
         }
         Some(Subcommand::Resume(ResumeCommand {
             session_id,
@@ -276,7 +402,8 @@ async fn cli_main(codex_linux_sandbox_exe: Option<PathBuf>) -> anyhow::Result<()
                 last,
                 config_overrides,
             );
-            codex_tui::run_main(interactive, codex_linux_sandbox_exe).await?;
+            let exit_info = codex_tui::run_main(interactive, codex_linux_sandbox_exe).await?;
+            handle_app_exit(exit_info)?;
         }
         Some(Subcommand::Login(mut login_cli)) => {
             prepend_config_flags(
@@ -288,7 +415,20 @@ async fn cli_main(codex_linux_sandbox_exe: Option<PathBuf>) -> anyhow::Result<()
                     run_login_status(login_cli.config_overrides).await;
                 }
                 None => {
-                    if let Some(api_key) = login_cli.api_key {
+                    if login_cli.use_device_code {
+                        run_login_with_device_code(
+                            login_cli.config_overrides,
+                            login_cli.issuer_base_url,
+                            login_cli.client_id,
+                        )
+                        .await;
+                    } else if login_cli.api_key.is_some() {
+                        eprintln!(
+                            "The --api-key flag is no longer supported. Pipe the key instead, e.g. `printenv OPENAI_API_KEY | codex login --with-api-key`."
+                        );
+                        std::process::exit(1);
+                    } else if login_cli.with_api_key {
+                        let api_key = read_api_key_from_stdin();
                         run_login_with_api_key(login_cli.config_overrides, api_key).await;
                     } else {
                         run_login_with_chatgpt(login_cli.config_overrides).await;
@@ -303,18 +443,18 @@ async fn cli_main(codex_linux_sandbox_exe: Option<PathBuf>) -> anyhow::Result<()
             );
             run_logout(logout_cli.config_overrides).await;
         }
-        Some(Subcommand::Proto(mut proto_cli)) => {
-            prepend_config_flags(
-                &mut proto_cli.config_overrides,
-                root_config_overrides.clone(),
-            );
-            proto::run_main(proto_cli).await?;
-        }
         Some(Subcommand::Completion(completion_cli)) => {
             print_completion(completion_cli);
         }
-        Some(Subcommand::Debug(debug_args)) => match debug_args.cmd {
-            DebugCommand::Seatbelt(mut seatbelt_cli) => {
+        Some(Subcommand::Cloud(mut cloud_cli)) => {
+            prepend_config_flags(
+                &mut cloud_cli.config_overrides,
+                root_config_overrides.clone(),
+            );
+            codex_cloud_tasks::run_main(cloud_cli, codex_linux_sandbox_exe).await?;
+        }
+        Some(Subcommand::Sandbox(sandbox_args)) => match sandbox_args.cmd {
+            SandboxCommand::Macos(mut seatbelt_cli) => {
                 prepend_config_flags(
                     &mut seatbelt_cli.config_overrides,
                     root_config_overrides.clone(),
@@ -325,7 +465,7 @@ async fn cli_main(codex_linux_sandbox_exe: Option<PathBuf>) -> anyhow::Result<()
                 )
                 .await?;
             }
-            DebugCommand::Landlock(mut landlock_cli) => {
+            SandboxCommand::Linux(mut landlock_cli) => {
                 prepend_config_flags(
                     &mut landlock_cli.config_overrides,
                     root_config_overrides.clone(),
@@ -336,6 +476,17 @@ async fn cli_main(codex_linux_sandbox_exe: Option<PathBuf>) -> anyhow::Result<()
                 )
                 .await?;
             }
+            SandboxCommand::Windows(mut windows_cli) => {
+                prepend_config_flags(
+                    &mut windows_cli.config_overrides,
+                    root_config_overrides.clone(),
+                );
+                codex_cli::debug_sandbox::run_command_under_windows(
+                    windows_cli,
+                    codex_linux_sandbox_exe,
+                )
+                .await?;
+            }
         },
         Some(Subcommand::Apply(mut apply_cli)) => {
             prepend_config_flags(
@@ -344,14 +495,42 @@ async fn cli_main(codex_linux_sandbox_exe: Option<PathBuf>) -> anyhow::Result<()
             );
             run_apply_command(apply_cli, None).await?;
         }
+        Some(Subcommand::ResponsesApiProxy(args)) => {
+            tokio::task::spawn_blocking(move || codex_responses_api_proxy::run_main(args))
+                .await??;
+        }
+        Some(Subcommand::StdioToUds(cmd)) => {
+            let socket_path = cmd.socket_path;
+            tokio::task::spawn_blocking(move || codex_stdio_to_uds::run(socket_path.as_path()))
+                .await??;
+        }
         Some(Subcommand::GenerateTs(gen_cli)) => {
             codex_protocol_ts::generate_ts(&gen_cli.out_dir, gen_cli.prettier.as_deref())?;
         }
-        Some(Subcommand::ResponsesApiProxy(args)) => {
-            tokio::task::spawn_blocking(move || codex_responses_api_proxy::run_main(args))
-                .await
-                .context("responses-api-proxy blocking task panicked")??;
-        }
+        Some(Subcommand::Features(FeaturesCli { sub })) => match sub {
+            FeaturesSubcommand::List => {
+                // Respect root-level `-c` overrides plus top-level flags like `--profile`.
+                let cli_kv_overrides = root_config_overrides
+                    .parse_overrides()
+                    .map_err(anyhow::Error::msg)?;
+
+                // Thread through relevant top-level flags (at minimum, `--profile`).
+                // Also honor `--search` since it maps to a feature toggle.
+                let overrides = ConfigOverrides {
+                    config_profile: interactive.config_profile.clone(),
+                    tools_web_search_request: interactive.web_search.then_some(true),
+                    ..Default::default()
+                };
+
+                let config = Config::load_with_cli_overrides(cli_kv_overrides, overrides).await?;
+                for def in codex_core::features::FEATURES.iter() {
+                    let name = def.key;
+                    let stage = stage_str(def.stage);
+                    let enabled = config.features.enabled(def.id);
+                    println!("{name}\t{stage}\t{enabled}");
+                }
+            }
+        },
     }
 
     Ok(())
@@ -426,6 +605,9 @@ fn merge_resume_cli_flags(interactive: &mut TuiCli, resume_cli: TuiCli) {
     if !resume_cli.images.is_empty() {
         interactive.images = resume_cli.images;
     }
+    if !resume_cli.add_dir.is_empty() {
+        interactive.add_dir.extend(resume_cli.add_dir);
+    }
     if let Some(prompt) = resume_cli.prompt {
         interactive.prompt = Some(prompt);
     }
@@ -445,8 +627,10 @@ fn print_completion(cmd: CompletionCommand) {
 #[cfg(test)]
 mod tests {
     use super::*;
+    use assert_matches::assert_matches;
     use codex_core::protocol::TokenUsage;
-    use codex_protocol::mcp_protocol::ConversationId;
+    use codex_protocol::ConversationId;
+    use pretty_assertions::assert_eq;
 
     fn finalize_from_args(args: &[&str]) -> TuiCli {
         let cli = MultitoolCli::try_parse_from(args).expect("parse");
@@ -454,6 +638,7 @@ mod tests {
             interactive,
             config_overrides: root_overrides,
             subcommand,
+            feature_toggles: _,
         } = cli;
 
         let Subcommand::Resume(ResumeCommand {
@@ -479,6 +664,7 @@ mod tests {
             conversation_id: conversation
                 .map(ConversationId::from_string)
                 .map(Result::unwrap),
+            update_action: None,
         }
     }
 
@@ -487,6 +673,7 @@ mod tests {
         let exit_info = AppExitInfo {
             token_usage: TokenUsage::default(),
             conversation_id: None,
+            update_action: None,
         };
         let lines = format_exit_messages(exit_info, false);
         assert!(lines.is_empty());
@@ -500,7 +687,7 @@ mod tests {
             lines,
             vec![
                 "Token usage: total=2 input=0 output=2".to_string(),
-                "To continue this session, run codex resume 123e4567-e89b-12d3-a456-426614174000."
+                "To continue this session, run codex resume 123e4567-e89b-12d3-a456-426614174000"
                     .to_string(),
             ]
         );
@@ -577,14 +764,14 @@ mod tests {
         assert_eq!(interactive.model.as_deref(), Some("gpt-5-test"));
         assert!(interactive.oss);
         assert_eq!(interactive.config_profile.as_deref(), Some("my-profile"));
-        assert!(matches!(
+        assert_matches!(
             interactive.sandbox_mode,
             Some(codex_common::SandboxModeCliArg::WorkspaceWrite)
-        ));
-        assert!(matches!(
+        );
+        assert_matches!(
             interactive.approval_policy,
             Some(codex_common::ApprovalModeCliArg::OnRequest)
-        ));
+        );
         assert!(interactive.full_auto);
         assert_eq!(
             interactive.cwd.as_deref(),
@@ -620,4 +807,32 @@ mod tests {
         assert!(!interactive.resume_last);
         assert_eq!(interactive.resume_session_id, None);
     }
+
+    #[test]
+    fn feature_toggles_known_features_generate_overrides() {
+        let toggles = FeatureToggles {
+            enable: vec!["web_search_request".to_string()],
+            disable: vec!["unified_exec".to_string()],
+        };
+        let overrides = toggles.to_overrides().expect("valid features");
+        assert_eq!(
+            overrides,
+            vec![
+                "features.web_search_request=true".to_string(),
+                "features.unified_exec=false".to_string(),
+            ]
+        );
+    }
+
+    #[test]
+    fn feature_toggles_unknown_feature_errors() {
+        let toggles = FeatureToggles {
+            enable: vec!["does_not_exist".to_string()],
+            disable: Vec::new(),
+        };
+        let err = toggles
+            .to_overrides()
+            .expect_err("feature should be rejected");
+        assert_eq!(err.to_string(), "Unknown feature flag: does_not_exist");
+    }
 }

codex-rs/cli/src/mcp_cmd.rs

@@ -1,18 +1,25 @@
-use std::collections::BTreeMap;
 use std::collections::HashMap;
-use std::path::PathBuf;
 
 use anyhow::Context;
 use anyhow::Result;
 use anyhow::anyhow;
 use anyhow::bail;
+use clap::ArgGroup;
 use codex_common::CliConfigOverrides;
+use codex_common::format_env_display::format_env_display;
 use codex_core::config::Config;
 use codex_core::config::ConfigOverrides;
+use codex_core::config::edit::ConfigEditsBuilder;
 use codex_core::config::find_codex_home;
 use codex_core::config::load_global_mcp_servers;
-use codex_core::config::write_global_mcp_servers;
-use codex_core::config_types::McpServerConfig;
+use codex_core::config::types::McpServerConfig;
+use codex_core::config::types::McpServerTransportConfig;
+use codex_core::features::Feature;
+use codex_core::mcp::auth::compute_auth_statuses;
+use codex_core::protocol::McpAuthStatus;
+use codex_rmcp_client::delete_oauth_tokens;
+use codex_rmcp_client::perform_oauth_login;
+use codex_rmcp_client::supports_oauth_login;
 
 /// [experimental] Launch Codex as an MCP server or manage configured MCP servers.
 ///
@@ -28,14 +35,11 @@ pub struct McpCli {
     pub config_overrides: CliConfigOverrides,
 
     #[command(subcommand)]
-    pub cmd: Option<McpSubcommand>,
+    pub subcommand: McpSubcommand,
 }
 
 #[derive(Debug, clap::Subcommand)]
 pub enum McpSubcommand {
-    /// [experimental] Run the Codex MCP server (stdio transport).
-    Serve,
-
     /// [experimental] List configured MCP servers.
     List(ListArgs),
 
@@ -47,6 +51,14 @@ pub enum McpSubcommand {
 
     /// [experimental] Remove a global MCP server entry.
     Remove(RemoveArgs),
+
+    /// [experimental] Authenticate with a configured MCP server via OAuth.
+    /// Requires experimental_use_rmcp_client = true in config.toml.
+    Login(LoginArgs),
+
+    /// [experimental] Remove stored OAuth credentials for a server.
+    /// Requires experimental_use_rmcp_client = true in config.toml.
+    Logout(LogoutArgs),
 }
 
 #[derive(Debug, clap::Parser)]
@@ -71,13 +83,61 @@ pub struct AddArgs {
     /// Name for the MCP server configuration.
     pub name: String,
 
-    /// Environment variables to set when launching the server.
-    #[arg(long, value_parser = parse_env_pair, value_name = "KEY=VALUE")]
-    pub env: Vec<(String, String)>,
+    #[command(flatten)]
+    pub transport_args: AddMcpTransportArgs,
+}
 
+#[derive(Debug, clap::Args)]
+#[command(
+    group(
+        ArgGroup::new("transport")
+            .args(["command", "url"])
+            .required(true)
+            .multiple(false)
+    )
+)]
+pub struct AddMcpTransportArgs {
+    #[command(flatten)]
+    pub stdio: Option<AddMcpStdioArgs>,
+
+    #[command(flatten)]
+    pub streamable_http: Option<AddMcpStreamableHttpArgs>,
+}
+
+#[derive(Debug, clap::Args)]
+pub struct AddMcpStdioArgs {
     /// Command to launch the MCP server.
-    #[arg(trailing_var_arg = true, num_args = 1..)]
+    /// Use --url for a streamable HTTP server.
+    #[arg(
+            trailing_var_arg = true,
+            num_args = 0..,
+        )]
     pub command: Vec<String>,
+
+    /// Environment variables to set when launching the server.
+    /// Only valid with stdio servers.
+    #[arg(
+        long,
+        value_parser = parse_env_pair,
+        value_name = "KEY=VALUE",
+    )]
+    pub env: Vec<(String, String)>,
+}
+
+#[derive(Debug, clap::Args)]
+pub struct AddMcpStreamableHttpArgs {
+    /// URL for a streamable HTTP MCP server.
+    #[arg(long)]
+    pub url: String,
+
+    /// Optional environment variable to read for a bearer token.
+    /// Only valid with streamable HTTP servers.
+    #[arg(
+        long = "bearer-token-env-var",
+        value_name = "ENV_VAR",
+        requires = "url"
+    )]
+    pub bearer_token_env_var: Option<String>,
 }
 
 #[derive(Debug, clap::Parser)]
@@ -86,29 +146,47 @@ pub struct RemoveArgs {
     pub name: String,
 }
 
+#[derive(Debug, clap::Parser)]
+pub struct LoginArgs {
+    /// Name of the MCP server to authenticate with oauth.
+    pub name: String,
+
+    /// Comma-separated list of OAuth scopes to request.
+    #[arg(long, value_delimiter = ',', value_name = "SCOPE,SCOPE")]
+    pub scopes: Vec<String>,
+}
+
+#[derive(Debug, clap::Parser)]
+pub struct LogoutArgs {
+    /// Name of the MCP server to deauthenticate.
+    pub name: String,
+}
+
 impl McpCli {
-    pub async fn run(self, codex_linux_sandbox_exe: Option<PathBuf>) -> Result<()> {
+    pub async fn run(self) -> Result<()> {
         let McpCli {
             config_overrides,
-            cmd,
+            subcommand,
         } = self;
-        let subcommand = cmd.unwrap_or(McpSubcommand::Serve);
 
         match subcommand {
-            McpSubcommand::Serve => {
-                codex_mcp_server::run_main(codex_linux_sandbox_exe, config_overrides).await?;
-            }
             McpSubcommand::List(args) => {
-                run_list(&config_overrides, args)?;
+                run_list(&config_overrides, args).await?;
             }
             McpSubcommand::Get(args) => {
-                run_get(&config_overrides, args)?;
+                run_get(&config_overrides, args).await?;
             }
             McpSubcommand::Add(args) => {
-                run_add(&config_overrides, args)?;
+                run_add(&config_overrides, args).await?;
             }
             McpSubcommand::Remove(args) => {
-                run_remove(&config_overrides, args)?;
+                run_remove(&config_overrides, args).await?;
+            }
+            McpSubcommand::Login(args) => {
+                run_login(&config_overrides, args).await?;
+            }
+            McpSubcommand::Logout(args) => {
+                run_logout(&config_overrides, args).await?;
             }
         }
 
@@ -116,54 +194,127 @@ impl McpCli {
     }
 }
 
-fn run_add(config_overrides: &CliConfigOverrides, add_args: AddArgs) -> Result<()> {
+async fn run_add(config_overrides: &CliConfigOverrides, add_args: AddArgs) -> Result<()> {
     // Validate any provided overrides even though they are not currently applied.
-    config_overrides.parse_overrides().map_err(|e| anyhow!(e))?;
+    let overrides = config_overrides
+        .parse_overrides()
+        .map_err(anyhow::Error::msg)?;
+    let config = Config::load_with_cli_overrides(overrides, ConfigOverrides::default())
+        .await
+        .context("failed to load configuration")?;
 
-    let AddArgs { name, env, command } = add_args;
+    let AddArgs {
+        name,
+        transport_args,
+    } = add_args;
 
     validate_server_name(&name)?;
 
-    let mut command_parts = command.into_iter();
-    let command_bin = command_parts
-        .next()
-        .ok_or_else(|| anyhow!("command is required"))?;
-    let command_args: Vec<String> = command_parts.collect();
-
-    let env_map = if env.is_empty() {
-        None
-    } else {
-        let mut map = HashMap::new();
-        for (key, value) in env {
-            map.insert(key, value);
-        }
-        Some(map)
-    };
-
     let codex_home = find_codex_home().context("failed to resolve CODEX_HOME")?;
     let mut servers = load_global_mcp_servers(&codex_home)
+        .await
         .with_context(|| format!("failed to load MCP servers from {}", codex_home.display()))?;
 
+    let transport = match transport_args {
+        AddMcpTransportArgs {
+            stdio: Some(stdio), ..
+        } => {
+            let mut command_parts = stdio.command.into_iter();
+            let command_bin = command_parts
+                .next()
+                .ok_or_else(|| anyhow!("command is required"))?;
+            let command_args: Vec<String> = command_parts.collect();
+
+            let env_map = if stdio.env.is_empty() {
+                None
+            } else {
+                Some(stdio.env.into_iter().collect::<HashMap<_, _>>())
+            };
+            McpServerTransportConfig::Stdio {
+                command: command_bin,
+                args: command_args,
+                env: env_map,
+                env_vars: Vec::new(),
+                cwd: None,
+            }
+        }
+        AddMcpTransportArgs {
+            streamable_http:
+                Some(AddMcpStreamableHttpArgs {
+                    url,
+                    bearer_token_env_var,
+                }),
+            ..
+        } => McpServerTransportConfig::StreamableHttp {
+            url,
+            bearer_token_env_var,
+            http_headers: None,
+            env_http_headers: None,
+        },
+        AddMcpTransportArgs { .. } => bail!("exactly one of --command or --url must be provided"),
+    };
+
     let new_entry = McpServerConfig {
-        command: command_bin,
-        args: command_args,
-        env: env_map,
+        transport: transport.clone(),
+        enabled: true,
         startup_timeout_sec: None,
         tool_timeout_sec: None,
+        enabled_tools: None,
+        disabled_tools: None,
     };
 
     servers.insert(name.clone(), new_entry);
 
-    write_global_mcp_servers(&codex_home, &servers)
+    ConfigEditsBuilder::new(&codex_home)
+        .replace_mcp_servers(&servers)
+        .apply()
+        .await
         .with_context(|| format!("failed to write MCP servers to {}", codex_home.display()))?;
 
     println!("Added global MCP server '{name}'.");
 
+    if let McpServerTransportConfig::StreamableHttp {
+        url,
+        bearer_token_env_var: None,
+        http_headers,
+        env_http_headers,
+    } = transport
+    {
+        match supports_oauth_login(&url).await {
+            Ok(true) => {
+                if !config.features.enabled(Feature::RmcpClient) {
+                    println!(
+                        "MCP server supports login. Add `experimental_use_rmcp_client = true` \
+                         to your config.toml and run `codex mcp login {name}` to login."
+                    );
+                } else {
+                    println!("Detected OAuth support. Starting OAuth flow…");
+                    perform_oauth_login(
+                        &name,
+                        &url,
+                        config.mcp_oauth_credentials_store_mode,
+                        http_headers.clone(),
+                        env_http_headers.clone(),
+                        &Vec::new(),
+                    )
+                    .await?;
+                    println!("Successfully logged in.");
+                }
+            }
+            Ok(false) => {}
+            Err(_) => println!(
+                "MCP server may or may not require login. Run `codex mcp login {name}` to login."
+            ),
+        }
+    }
+
     Ok(())
 }
 
-fn run_remove(config_overrides: &CliConfigOverrides, remove_args: RemoveArgs) -> Result<()> {
-    config_overrides.parse_overrides().map_err(|e| anyhow!(e))?;
+async fn run_remove(config_overrides: &CliConfigOverrides, remove_args: RemoveArgs) -> Result<()> {
+    config_overrides
+        .parse_overrides()
+        .map_err(anyhow::Error::msg)?;
 
     let RemoveArgs { name } = remove_args;
 
@@ -171,12 +322,16 @@ fn run_remove(config_overrides: &CliConfigOverrides, remove_args: RemoveArgs) ->
 
     let codex_home = find_codex_home().context("failed to resolve CODEX_HOME")?;
     let mut servers = load_global_mcp_servers(&codex_home)
+        .await
         .with_context(|| format!("failed to load MCP servers from {}", codex_home.display()))?;
 
     let removed = servers.remove(&name).is_some();
 
     if removed {
-        write_global_mcp_servers(&codex_home, &servers)
+        ConfigEditsBuilder::new(&codex_home)
+            .replace_mcp_servers(&servers)
+            .apply()
+            .await
             .with_context(|| format!("failed to write MCP servers to {}", codex_home.display()))?;
     }
 
@@ -189,34 +344,142 @@ fn run_remove(config_overrides: &CliConfigOverrides, remove_args: RemoveArgs) ->
     Ok(())
 }
 
-fn run_list(config_overrides: &CliConfigOverrides, list_args: ListArgs) -> Result<()> {
-    let overrides = config_overrides.parse_overrides().map_err(|e| anyhow!(e))?;
+async fn run_login(config_overrides: &CliConfigOverrides, login_args: LoginArgs) -> Result<()> {
+    let overrides = config_overrides
+        .parse_overrides()
+        .map_err(anyhow::Error::msg)?;
     let config = Config::load_with_cli_overrides(overrides, ConfigOverrides::default())
+        .await
+        .context("failed to load configuration")?;
+
+    if !config.features.enabled(Feature::RmcpClient) {
+        bail!("OAuth login is only supported when [feature].rmcp_client is true in config.toml.");
+    }
+
+    let LoginArgs { name, scopes } = login_args;
+
+    let Some(server) = config.mcp_servers.get(&name) else {
+        bail!("No MCP server named '{name}' found.");
+    };
+
+    let (url, http_headers, env_http_headers) = match &server.transport {
+        McpServerTransportConfig::StreamableHttp {
+            url,
+            http_headers,
+            env_http_headers,
+            ..
+        } => (url.clone(), http_headers.clone(), env_http_headers.clone()),
+        _ => bail!("OAuth login is only supported for streamable HTTP servers."),
+    };
+
+    perform_oauth_login(
+        &name,
+        &url,
+        config.mcp_oauth_credentials_store_mode,
+        http_headers,
+        env_http_headers,
+        &scopes,
+    )
+    .await?;
+    println!("Successfully logged in to MCP server '{name}'.");
+    Ok(())
+}
+
+async fn run_logout(config_overrides: &CliConfigOverrides, logout_args: LogoutArgs) -> Result<()> {
+    let overrides = config_overrides
+        .parse_overrides()
+        .map_err(anyhow::Error::msg)?;
+    let config = Config::load_with_cli_overrides(overrides, ConfigOverrides::default())
+        .await
+        .context("failed to load configuration")?;
+
+    let LogoutArgs { name } = logout_args;
+
+    let server = config
+        .mcp_servers
+        .get(&name)
+        .ok_or_else(|| anyhow!("No MCP server named '{name}' found in configuration."))?;
+
+    let url = match &server.transport {
+        McpServerTransportConfig::StreamableHttp { url, .. } => url.clone(),
+        _ => bail!("OAuth logout is only supported for streamable_http transports."),
+    };
+
+    match delete_oauth_tokens(&name, &url, config.mcp_oauth_credentials_store_mode) {
+        Ok(true) => println!("Removed OAuth credentials for '{name}'."),
+        Ok(false) => println!("No OAuth credentials stored for '{name}'."),
+        Err(err) => return Err(anyhow!("failed to delete OAuth credentials: {err}")),
+    }
+
+    Ok(())
+}
+
+async fn run_list(config_overrides: &CliConfigOverrides, list_args: ListArgs) -> Result<()> {
+    let overrides = config_overrides
+        .parse_overrides()
+        .map_err(anyhow::Error::msg)?;
+    let config = Config::load_with_cli_overrides(overrides, ConfigOverrides::default())
+        .await
         .context("failed to load configuration")?;
 
     let mut entries: Vec<_> = config.mcp_servers.iter().collect();
     entries.sort_by(|(a, _), (b, _)| a.cmp(b));
+    let auth_statuses = compute_auth_statuses(
+        config.mcp_servers.iter(),
+        config.mcp_oauth_credentials_store_mode,
+    )
+    .await;
 
     if list_args.json {
         let json_entries: Vec<_> = entries
             .into_iter()
             .map(|(name, cfg)| {
-                let env = cfg.env.as_ref().map(|env| {
-                    env.iter()
-                        .map(|(k, v)| (k.clone(), v.clone()))
-                        .collect::<BTreeMap<_, _>>()
-                });
+                let auth_status = auth_statuses
+                    .get(name.as_str())
+                    .map(|entry| entry.auth_status)
+                    .unwrap_or(McpAuthStatus::Unsupported);
+                let transport = match &cfg.transport {
+                    McpServerTransportConfig::Stdio {
+                        command,
+                        args,
+                        env,
+                        env_vars,
+                        cwd,
+                    } => serde_json::json!({
+                        "type": "stdio",
+                        "command": command,
+                        "args": args,
+                        "env": env,
+                        "env_vars": env_vars,
+                        "cwd": cwd,
+                    }),
+                    McpServerTransportConfig::StreamableHttp {
+                        url,
+                        bearer_token_env_var,
+                        http_headers,
+                        env_http_headers,
+                    } => {
+                        serde_json::json!({
+                            "type": "streamable_http",
+                            "url": url,
+                            "bearer_token_env_var": bearer_token_env_var,
+                            "http_headers": http_headers,
+                            "env_http_headers": env_http_headers,
+                        })
+                    }
+                };
+
                 serde_json::json!({
                     "name": name,
-                    "command": cfg.command,
-                    "args": cfg.args,
-                    "env": env,
+                    "enabled": cfg.enabled,
+                    "transport": transport,
                     "startup_timeout_sec": cfg
                         .startup_timeout_sec
                         .map(|timeout| timeout.as_secs_f64()),
                     "tool_timeout_sec": cfg
                         .tool_timeout_sec
                         .map(|timeout| timeout.as_secs_f64()),
+                    "auth_status": auth_status,
                 })
             })
             .collect();
@@ -230,70 +493,190 @@ fn run_list(config_overrides: &CliConfigOverrides, list_args: ListArgs) -> Resul
         return Ok(());
     }
 
-    let mut rows: Vec<[String; 4]> = Vec::new();
+    let mut stdio_rows: Vec<[String; 7]> = Vec::new();
+    let mut http_rows: Vec<[String; 5]> = Vec::new();
+
     for (name, cfg) in entries {
-        let args = if cfg.args.is_empty() {
-            "-".to_string()
-        } else {
-            cfg.args.join(" ")
-        };
-
-        let env = match cfg.env.as_ref() {
-            None => "-".to_string(),
-            Some(map) if map.is_empty() => "-".to_string(),
-            Some(map) => {
-                let mut pairs: Vec<_> = map.iter().collect();
-                pairs.sort_by(|(a, _), (b, _)| a.cmp(b));
-                pairs
-                    .into_iter()
-                    .map(|(k, v)| format!("{k}={v}"))
-                    .collect::<Vec<_>>()
-                    .join(", ")
+        match &cfg.transport {
+            McpServerTransportConfig::Stdio {
+                command,
+                args,
+                env,
+                env_vars,
+                cwd,
+            } => {
+                let args_display = if args.is_empty() {
+                    "-".to_string()
+                } else {
+                    args.join(" ")
+                };
+                let env_display = format_env_display(env.as_ref(), env_vars);
+                let cwd_display = cwd
+                    .as_ref()
+                    .map(|path| path.display().to_string())
+                    .filter(|value| !value.is_empty())
+                    .unwrap_or_else(|| "-".to_string());
+                let status = if cfg.enabled {
+                    "enabled".to_string()
+                } else {
+                    "disabled".to_string()
+                };
+                let auth_status = auth_statuses
+                    .get(name.as_str())
+                    .map(|entry| entry.auth_status)
+                    .unwrap_or(McpAuthStatus::Unsupported)
+                    .to_string();
+                stdio_rows.push([
+                    name.clone(),
+                    command.clone(),
+                    args_display,
+                    env_display,
+                    cwd_display,
+                    status,
+                    auth_status,
+                ]);
+            }
+            McpServerTransportConfig::StreamableHttp {
+                url,
+                bearer_token_env_var,
+                ..
+            } => {
+                let status = if cfg.enabled {
+                    "enabled".to_string()
+                } else {
+                    "disabled".to_string()
+                };
+                let auth_status = auth_statuses
+                    .get(name.as_str())
+                    .map(|entry| entry.auth_status)
+                    .unwrap_or(McpAuthStatus::Unsupported)
+                    .to_string();
+                let bearer_token_display =
+                    bearer_token_env_var.as_deref().unwrap_or("-").to_string();
+                http_rows.push([
+                    name.clone(),
+                    url.clone(),
+                    bearer_token_display,
+                    status,
+                    auth_status,
+                ]);
             }
-        };
-
-        rows.push([name.clone(), cfg.command.clone(), args, env]);
-    }
-
-    let mut widths = ["Name".len(), "Command".len(), "Args".len(), "Env".len()];
-    for row in &rows {
-        for (i, cell) in row.iter().enumerate() {
-            widths[i] = widths[i].max(cell.len());
         }
     }
 
-    println!(
-        "{:<name_w$}  {:<cmd_w$}  {:<args_w$}  {:<env_w$}",
-        "Name",
-        "Command",
-        "Args",
-        "Env",
-        name_w = widths[0],
-        cmd_w = widths[1],
-        args_w = widths[2],
-        env_w = widths[3],
-    );
+    if !stdio_rows.is_empty() {
+        let mut widths = [
+            "Name".len(),
+            "Command".len(),
+            "Args".len(),
+            "Env".len(),
+            "Cwd".len(),
+            "Status".len(),
+            "Auth".len(),
+        ];
+        for row in &stdio_rows {
+            for (i, cell) in row.iter().enumerate() {
+                widths[i] = widths[i].max(cell.len());
+            }
+        }
 
-    for row in rows {
         println!(
-            "{:<name_w$}  {:<cmd_w$}  {:<args_w$}  {:<env_w$}",
-            row[0],
-            row[1],
-            row[2],
-            row[3],
+            "{name:<name_w$}  {command:<cmd_w$}  {args:<args_w$}  {env:<env_w$}  {cwd:<cwd_w$}  {status:<status_w$}  {auth:<auth_w$}",
+            name = "Name",
+            command = "Command",
+            args = "Args",
+            env = "Env",
+            cwd = "Cwd",
+            status = "Status",
+            auth = "Auth",
             name_w = widths[0],
             cmd_w = widths[1],
             args_w = widths[2],
             env_w = widths[3],
+            cwd_w = widths[4],
+            status_w = widths[5],
+            auth_w = widths[6],
         );
+
+        for row in &stdio_rows {
+            println!(
+                "{name:<name_w$}  {command:<cmd_w$}  {args:<args_w$}  {env:<env_w$}  {cwd:<cwd_w$}  {status:<status_w$}  {auth:<auth_w$}",
+                name = row[0].as_str(),
+                command = row[1].as_str(),
+                args = row[2].as_str(),
+                env = row[3].as_str(),
+                cwd = row[4].as_str(),
+                status = row[5].as_str(),
+                auth = row[6].as_str(),
+                name_w = widths[0],
+                cmd_w = widths[1],
+                args_w = widths[2],
+                env_w = widths[3],
+                cwd_w = widths[4],
+                status_w = widths[5],
+                auth_w = widths[6],
+            );
+        }
+    }
+
+    if !stdio_rows.is_empty() && !http_rows.is_empty() {
+        println!();
+    }
+
+    if !http_rows.is_empty() {
+        let mut widths = [
+            "Name".len(),
+            "Url".len(),
+            "Bearer Token Env Var".len(),
+            "Status".len(),
+            "Auth".len(),
+        ];
+        for row in &http_rows {
+            for (i, cell) in row.iter().enumerate() {
+                widths[i] = widths[i].max(cell.len());
+            }
+        }
+
+        println!(
+            "{name:<name_w$}  {url:<url_w$}  {token:<token_w$}  {status:<status_w$}  {auth:<auth_w$}",
+            name = "Name",
+            url = "Url",
+            token = "Bearer Token Env Var",
+            status = "Status",
+            auth = "Auth",
+            name_w = widths[0],
+            url_w = widths[1],
+            token_w = widths[2],
+            status_w = widths[3],
+            auth_w = widths[4],
+        );
+
+        for row in &http_rows {
+            println!(
+                "{name:<name_w$}  {url:<url_w$}  {token:<token_w$}  {status:<status_w$}  {auth:<auth_w$}",
+                name = row[0].as_str(),
+                url = row[1].as_str(),
+                token = row[2].as_str(),
+                status = row[3].as_str(),
+                auth = row[4].as_str(),
+                name_w = widths[0],
+                url_w = widths[1],
+                token_w = widths[2],
+                status_w = widths[3],
+                auth_w = widths[4],
+            );
+        }
     }
 
     Ok(())
 }
 
-fn run_get(config_overrides: &CliConfigOverrides, get_args: GetArgs) -> Result<()> {
-    let overrides = config_overrides.parse_overrides().map_err(|e| anyhow!(e))?;
+async fn run_get(config_overrides: &CliConfigOverrides, get_args: GetArgs) -> Result<()> {
+    let overrides = config_overrides
+        .parse_overrides()
+        .map_err(anyhow::Error::msg)?;
     let config = Config::load_with_cli_overrides(overrides, ConfigOverrides::default())
+        .await
         .context("failed to load configuration")?;
 
     let Some(server) = config.mcp_servers.get(&get_args.name) else {
@@ -301,16 +684,40 @@ fn run_get(config_overrides: &CliConfigOverrides, get_args: GetArgs) -> Result<(
     };
 
     if get_args.json {
-        let env = server.env.as_ref().map(|env| {
-            env.iter()
-                .map(|(k, v)| (k.clone(), v.clone()))
-                .collect::<BTreeMap<_, _>>()
-        });
+        let transport = match &server.transport {
+            McpServerTransportConfig::Stdio {
+                command,
+                args,
+                env,
+                env_vars,
+                cwd,
+            } => serde_json::json!({
+                "type": "stdio",
+                "command": command,
+                "args": args,
+                "env": env,
+                "env_vars": env_vars,
+                "cwd": cwd,
+            }),
+            McpServerTransportConfig::StreamableHttp {
+                url,
+                bearer_token_env_var,
+                http_headers,
+                env_http_headers,
+            } => serde_json::json!({
+                "type": "streamable_http",
+                "url": url,
+                "bearer_token_env_var": bearer_token_env_var,
+                "http_headers": http_headers,
+                "env_http_headers": env_http_headers,
+            }),
+        };
         let output = serde_json::to_string_pretty(&serde_json::json!({
             "name": get_args.name,
-            "command": server.command,
-            "args": server.args,
-            "env": env,
+            "enabled": server.enabled,
+            "transport": transport,
+            "enabled_tools": server.enabled_tools.clone(),
+            "disabled_tools": server.disabled_tools.clone(),
             "startup_timeout_sec": server
                 .startup_timeout_sec
                 .map(|timeout| timeout.as_secs_f64()),
@@ -322,28 +729,91 @@ fn run_get(config_overrides: &CliConfigOverrides, get_args: GetArgs) -> Result<(
         return Ok(());
     }
 
+    if !server.enabled {
+        println!("{} (disabled)", get_args.name);
+        return Ok(());
+    }
+
     println!("{}", get_args.name);
-    println!("  command: {}", server.command);
-    let args = if server.args.is_empty() {
-        "-".to_string()
-    } else {
-        server.args.join(" ")
-    };
-    println!("  args: {args}");
-    let env_display = match server.env.as_ref() {
-        None => "-".to_string(),
-        Some(map) if map.is_empty() => "-".to_string(),
-        Some(map) => {
-            let mut pairs: Vec<_> = map.iter().collect();
-            pairs.sort_by(|(a, _), (b, _)| a.cmp(b));
-            pairs
-                .into_iter()
-                .map(|(k, v)| format!("{k}={v}"))
-                .collect::<Vec<_>>()
-                .join(", ")
+    println!("  enabled: {}", server.enabled);
+    let format_tool_list = |tools: &Option<Vec<String>>| -> String {
+        match tools {
+            Some(list) if list.is_empty() => "[]".to_string(),
+            Some(list) => list.join(", "),
+            None => "-".to_string(),
         }
     };
-    println!("  env: {env_display}");
+    if server.enabled_tools.is_some() {
+        let enabled_tools_display = format_tool_list(&server.enabled_tools);
+        println!("  enabled_tools: {enabled_tools_display}");
+    }
+    if server.disabled_tools.is_some() {
+        let disabled_tools_display = format_tool_list(&server.disabled_tools);
+        println!("  disabled_tools: {disabled_tools_display}");
+    }
+    match &server.transport {
+        McpServerTransportConfig::Stdio {
+            command,
+            args,
+            env,
+            env_vars,
+            cwd,
+        } => {
+            println!("  transport: stdio");
+            println!("  command: {command}");
+            let args_display = if args.is_empty() {
+                "-".to_string()
+            } else {
+                args.join(" ")
+            };
+            println!("  args: {args_display}");
+            let cwd_display = cwd
+                .as_ref()
+                .map(|path| path.display().to_string())
+                .filter(|value| !value.is_empty())
+                .unwrap_or_else(|| "-".to_string());
+            println!("  cwd: {cwd_display}");
+            let env_display = format_env_display(env.as_ref(), env_vars);
+            println!("  env: {env_display}");
+        }
+        McpServerTransportConfig::StreamableHttp {
+            url,
+            bearer_token_env_var,
+            http_headers,
+            env_http_headers,
+        } => {
+            println!("  transport: streamable_http");
+            println!("  url: {url}");
+            let bearer_token_display = bearer_token_env_var.as_deref().unwrap_or("-");
+            println!("  bearer_token_env_var: {bearer_token_display}");
+            let headers_display = match http_headers {
+                Some(map) if !map.is_empty() => {
+                    let mut pairs: Vec<_> = map.iter().collect();
+                    pairs.sort_by(|(a, _), (b, _)| a.cmp(b));
+                    pairs
+                        .into_iter()
+                        .map(|(k, _)| format!("{k}=*****"))
+                        .collect::<Vec<_>>()
+                        .join(", ")
+                }
+                _ => "-".to_string(),
+            };
+            println!("  http_headers: {headers_display}");
+            let env_headers_display = match env_http_headers {
+                Some(map) if !map.is_empty() => {
+                    let mut pairs: Vec<_> = map.iter().collect();
+                    pairs.sort_by(|(a, _), (b, _)| a.cmp(b));
+                    pairs
+                        .into_iter()
+                        .map(|(k, var)| format!("{k}={var}"))
+                        .collect::<Vec<_>>()
+                        .join(", ")
+                }
+                _ => "-".to_string(),
+            };
+            println!("  env_http_headers: {env_headers_display}");
+        }
+    }
     if let Some(timeout) = server.startup_timeout_sec {
         println!("  startup_timeout_sec: {}", timeout.as_secs_f64());
     }

codex-rs/cli/src/proto.rs

@@ -1,133 +0,0 @@
-use std::io::IsTerminal;
-
-use clap::Parser;
-use codex_common::CliConfigOverrides;
-use codex_core::AuthManager;
-use codex_core::ConversationManager;
-use codex_core::NewConversation;
-use codex_core::config::Config;
-use codex_core::config::ConfigOverrides;
-use codex_core::protocol::Event;
-use codex_core::protocol::EventMsg;
-use codex_core::protocol::Submission;
-use tokio::io::AsyncBufReadExt;
-use tokio::io::BufReader;
-use tracing::error;
-use tracing::info;
-
-#[derive(Debug, Parser)]
-pub struct ProtoCli {
-    #[clap(skip)]
-    pub config_overrides: CliConfigOverrides,
-}
-
-pub async fn run_main(opts: ProtoCli) -> anyhow::Result<()> {
-    if std::io::stdin().is_terminal() {
-        anyhow::bail!("Protocol mode expects stdin to be a pipe, not a terminal");
-    }
-
-    tracing_subscriber::fmt()
-        .with_writer(std::io::stderr)
-        .init();
-
-    let ProtoCli { config_overrides } = opts;
-    let overrides_vec = config_overrides
-        .parse_overrides()
-        .map_err(anyhow::Error::msg)?;
-
-    let config = Config::load_with_cli_overrides(overrides_vec, ConfigOverrides::default())?;
-    // Use conversation_manager API to start a conversation
-    let conversation_manager =
-        ConversationManager::new(AuthManager::shared(config.codex_home.clone()));
-    let NewConversation {
-        conversation_id: _,
-        conversation,
-        session_configured,
-    } = conversation_manager.new_conversation(config).await?;
-
-    // Simulate streaming the session_configured event.
-    let synthetic_event = Event {
-        // Fake id value.
-        id: "".to_string(),
-        msg: EventMsg::SessionConfigured(session_configured),
-    };
-    let session_configured_event = match serde_json::to_string(&synthetic_event) {
-        Ok(s) => s,
-        Err(e) => {
-            error!("Failed to serialize session_configured: {e}");
-            return Err(anyhow::Error::from(e));
-        }
-    };
-    println!("{session_configured_event}");
-
-    // Task that reads JSON lines from stdin and forwards to Submission Queue
-    let sq_fut = {
-        let conversation = conversation.clone();
-        async move {
-            let stdin = BufReader::new(tokio::io::stdin());
-            let mut lines = stdin.lines();
-            loop {
-                let result = tokio::select! {
-                    _ = tokio::signal::ctrl_c() => {
-                        break
-                    },
-                    res = lines.next_line() => res,
-                };
-
-                match result {
-                    Ok(Some(line)) => {
-                        let line = line.trim();
-                        if line.is_empty() {
-                            continue;
-                        }
-                        match serde_json::from_str::<Submission>(line) {
-                            Ok(sub) => {
-                                if let Err(e) = conversation.submit_with_id(sub).await {
-                                    error!("{e:#}");
-                                    break;
-                                }
-                            }
-                            Err(e) => {
-                                error!("invalid submission: {e}");
-                            }
-                        }
-                    }
-                    _ => {
-                        info!("Submission queue closed");
-                        break;
-                    }
-                }
-            }
-        }
-    };
-
-    // Task that reads events from the agent and prints them as JSON lines to stdout
-    let eq_fut = async move {
-        loop {
-            let event = tokio::select! {
-                _ = tokio::signal::ctrl_c() => break,
-                event = conversation.next_event() => event,
-            };
-            match event {
-                Ok(event) => {
-                    let event_str = match serde_json::to_string(&event) {
-                        Ok(s) => s,
-                        Err(e) => {
-                            error!("Failed to serialize event: {e}");
-                            continue;
-                        }
-                    };
-                    println!("{event_str}");
-                }
-                Err(e) => {
-                    error!("{e:#}");
-                    break;
-                }
-            }
-        }
-        info!("Event queue closed");
-    };
-
-    tokio::join!(sq_fut, eq_fut);
-    Ok(())
-}

codex-rs/cli/tests/mcp_add_remove.rs

@@ -2,6 +2,7 @@ use std::path::Path;
 
 use anyhow::Result;
 use codex_core::config::load_global_mcp_servers;
+use codex_core::config::types::McpServerTransportConfig;
 use predicates::str::contains;
 use pretty_assertions::assert_eq;
 use tempfile::TempDir;
@@ -12,8 +13,8 @@ fn codex_command(codex_home: &Path) -> Result<assert_cmd::Command> {
     Ok(cmd)
 }
 
-#[test]
-fn add_and_remove_server_updates_global_config() -> Result<()> {
+#[tokio::test]
+async fn add_and_remove_server_updates_global_config() -> Result<()> {
     let codex_home = TempDir::new()?;
 
     let mut add_cmd = codex_command(codex_home.path())?;
@@ -23,12 +24,26 @@ fn add_and_remove_server_updates_global_config() -> Result<()> {
         .success()
         .stdout(contains("Added global MCP server 'docs'."));
 
-    let servers = load_global_mcp_servers(codex_home.path())?;
+    let servers = load_global_mcp_servers(codex_home.path()).await?;
     assert_eq!(servers.len(), 1);
     let docs = servers.get("docs").expect("server should exist");
-    assert_eq!(docs.command, "echo");
-    assert_eq!(docs.args, vec!["hello".to_string()]);
-    assert!(docs.env.is_none());
+    match &docs.transport {
+        McpServerTransportConfig::Stdio {
+            command,
+            args,
+            env,
+            env_vars,
+            cwd,
+        } => {
+            assert_eq!(command, "echo");
+            assert_eq!(args, &vec!["hello".to_string()]);
+            assert!(env.is_none());
+            assert!(env_vars.is_empty());
+            assert!(cwd.is_none());
+        }
+        other => panic!("unexpected transport: {other:?}"),
+    }
+    assert!(docs.enabled);
 
     let mut remove_cmd = codex_command(codex_home.path())?;
     remove_cmd
@@ -37,7 +52,7 @@ fn add_and_remove_server_updates_global_config() -> Result<()> {
         .success()
         .stdout(contains("Removed global MCP server 'docs'."));
 
-    let servers = load_global_mcp_servers(codex_home.path())?;
+    let servers = load_global_mcp_servers(codex_home.path()).await?;
     assert!(servers.is_empty());
 
     let mut remove_again_cmd = codex_command(codex_home.path())?;
@@ -47,14 +62,14 @@ fn add_and_remove_server_updates_global_config() -> Result<()> {
         .success()
         .stdout(contains("No MCP server named 'docs' found."));
 
-    let servers = load_global_mcp_servers(codex_home.path())?;
+    let servers = load_global_mcp_servers(codex_home.path()).await?;
     assert!(servers.is_empty());
 
     Ok(())
 }
 
-#[test]
-fn add_with_env_preserves_key_order_and_values() -> Result<()> {
+#[tokio::test]
+async fn add_with_env_preserves_key_order_and_values() -> Result<()> {
     let codex_home = TempDir::new()?;
 
     let mut add_cmd = codex_command(codex_home.path())?;
@@ -74,13 +89,140 @@ fn add_with_env_preserves_key_order_and_values() -> Result<()> {
         .assert()
         .success();
 
-    let servers = load_global_mcp_servers(codex_home.path())?;
+    let servers = load_global_mcp_servers(codex_home.path()).await?;
     let envy = servers.get("envy").expect("server should exist");
-    let env = envy.env.as_ref().expect("env should be present");
+    let env = match &envy.transport {
+        McpServerTransportConfig::Stdio { env: Some(env), .. } => env,
+        other => panic!("unexpected transport: {other:?}"),
+    };
 
     assert_eq!(env.len(), 2);
     assert_eq!(env.get("FOO"), Some(&"bar".to_string()));
     assert_eq!(env.get("ALPHA"), Some(&"beta".to_string()));
+    assert!(envy.enabled);
+
+    Ok(())
+}
+
+#[tokio::test]
+async fn add_streamable_http_without_manual_token() -> Result<()> {
+    let codex_home = TempDir::new()?;
+
+    let mut add_cmd = codex_command(codex_home.path())?;
+    add_cmd
+        .args(["mcp", "add", "github", "--url", "https://example.com/mcp"])
+        .assert()
+        .success();
+
+    let servers = load_global_mcp_servers(codex_home.path()).await?;
+    let github = servers.get("github").expect("github server should exist");
+    match &github.transport {
+        McpServerTransportConfig::StreamableHttp {
+            url,
+            bearer_token_env_var,
+            http_headers,
+            env_http_headers,
+        } => {
+            assert_eq!(url, "https://example.com/mcp");
+            assert!(bearer_token_env_var.is_none());
+            assert!(http_headers.is_none());
+            assert!(env_http_headers.is_none());
+        }
+        other => panic!("unexpected transport: {other:?}"),
+    }
+    assert!(github.enabled);
+
+    assert!(!codex_home.path().join(".credentials.json").exists());
+    assert!(!codex_home.path().join(".env").exists());
+
+    Ok(())
+}
+
+#[tokio::test]
+async fn add_streamable_http_with_custom_env_var() -> Result<()> {
+    let codex_home = TempDir::new()?;
+
+    let mut add_cmd = codex_command(codex_home.path())?;
+    add_cmd
+        .args([
+            "mcp",
+            "add",
+            "issues",
+            "--url",
+            "https://example.com/issues",
+            "--bearer-token-env-var",
+            "GITHUB_TOKEN",
+        ])
+        .assert()
+        .success();
+
+    let servers = load_global_mcp_servers(codex_home.path()).await?;
+    let issues = servers.get("issues").expect("issues server should exist");
+    match &issues.transport {
+        McpServerTransportConfig::StreamableHttp {
+            url,
+            bearer_token_env_var,
+            http_headers,
+            env_http_headers,
+        } => {
+            assert_eq!(url, "https://example.com/issues");
+            assert_eq!(bearer_token_env_var.as_deref(), Some("GITHUB_TOKEN"));
+            assert!(http_headers.is_none());
+            assert!(env_http_headers.is_none());
+        }
+        other => panic!("unexpected transport: {other:?}"),
+    }
+    assert!(issues.enabled);
+    Ok(())
+}
+
+#[tokio::test]
+async fn add_streamable_http_rejects_removed_flag() -> Result<()> {
+    let codex_home = TempDir::new()?;
+
+    let mut add_cmd = codex_command(codex_home.path())?;
+    add_cmd
+        .args([
+            "mcp",
+            "add",
+            "github",
+            "--url",
+            "https://example.com/mcp",
+            "--with-bearer-token",
+        ])
+        .assert()
+        .failure()
+        .stderr(contains("--with-bearer-token"));
+
+    let servers = load_global_mcp_servers(codex_home.path()).await?;
+    assert!(servers.is_empty());
+
+    Ok(())
+}
+
+#[tokio::test]
+async fn add_cant_add_command_and_url() -> Result<()> {
+    let codex_home = TempDir::new()?;
+
+    let mut add_cmd = codex_command(codex_home.path())?;
+    add_cmd
+        .args([
+            "mcp",
+            "add",
+            "github",
+            "--url",
+            "https://example.com/mcp",
+            "--command",
+            "--",
+            "echo",
+            "hello",
+        ])
+        .assert()
+        .failure()
+        .stderr(contains("unexpected argument '--command' found"));
+
+    let servers = load_global_mcp_servers(codex_home.path()).await?;
+    assert!(servers.is_empty());
 
     Ok(())
 }

codex-rs/cli/tests/mcp_list.rs

@@ -1,9 +1,14 @@
 use std::path::Path;
 
 use anyhow::Result;
+use codex_core::config::edit::ConfigEditsBuilder;
+use codex_core::config::load_global_mcp_servers;
+use codex_core::config::types::McpServerTransportConfig;
+use predicates::prelude::PredicateBooleanExt;
 use predicates::str::contains;
 use pretty_assertions::assert_eq;
 use serde_json::Value as JsonValue;
+use serde_json::json;
 use tempfile::TempDir;
 
 fn codex_command(codex_home: &Path) -> Result<assert_cmd::Command> {
@@ -25,8 +30,8 @@ fn list_shows_empty_state() -> Result<()> {
     Ok(())
 }
 
-#[test]
-fn list_and_get_render_expected_output() -> Result<()> {
+#[tokio::test]
+async fn list_and_get_render_expected_output() -> Result<()> {
     let codex_home = TempDir::new()?;
 
     let mut add = codex_command(codex_home.path())?;
@@ -44,6 +49,20 @@ fn list_and_get_render_expected_output() -> Result<()> {
     .assert()
     .success();
 
+    let mut servers = load_global_mcp_servers(codex_home.path()).await?;
+    let docs_entry = servers
+        .get_mut("docs")
+        .expect("docs server should exist after add");
+    match &mut docs_entry.transport {
+        McpServerTransportConfig::Stdio { env_vars, .. } => {
+            *env_vars = vec!["APP_TOKEN".to_string(), "WORKSPACE_ID".to_string()];
+        }
+        other => panic!("unexpected transport: {other:?}"),
+    }
+    ConfigEditsBuilder::new(codex_home.path())
+        .replace_mcp_servers(&servers)
+        .apply_blocking()?;
+
     let mut list_cmd = codex_command(codex_home.path())?;
     let list_output = list_cmd.args(["mcp", "list"]).output()?;
     assert!(list_output.status.success());
@@ -51,48 +70,61 @@ fn list_and_get_render_expected_output() -> Result<()> {
     assert!(stdout.contains("Name"));
     assert!(stdout.contains("docs"));
     assert!(stdout.contains("docs-server"));
-    assert!(stdout.contains("TOKEN=secret"));
+    assert!(stdout.contains("TOKEN=*****"));
+    assert!(stdout.contains("APP_TOKEN=*****"));
+    assert!(stdout.contains("WORKSPACE_ID=*****"));
+    assert!(stdout.contains("Status"));
+    assert!(stdout.contains("Auth"));
+    assert!(stdout.contains("enabled"));
+    assert!(stdout.contains("Unsupported"));
 
     let mut list_json_cmd = codex_command(codex_home.path())?;
     let json_output = list_json_cmd.args(["mcp", "list", "--json"]).output()?;
     assert!(json_output.status.success());
     let stdout = String::from_utf8(json_output.stdout)?;
     let parsed: JsonValue = serde_json::from_str(&stdout)?;
-    let array = parsed.as_array().expect("expected array");
-    assert_eq!(array.len(), 1);
-    let entry = &array[0];
-    assert_eq!(entry.get("name"), Some(&JsonValue::String("docs".into())));
     assert_eq!(
-        entry.get("command"),
-        Some(&JsonValue::String("docs-server".into()))
-    );
-
-    let args = entry
-        .get("args")
-        .and_then(|v| v.as_array())
-        .expect("args array");
-    assert_eq!(
-        args,
-        &vec![
-            JsonValue::String("--port".into()),
-            JsonValue::String("4000".into())
+        parsed,
+        json!([
+          {
+            "name": "docs",
+            "enabled": true,
+            "transport": {
+              "type": "stdio",
+              "command": "docs-server",
+              "args": [
+                "--port",
+                "4000"
+              ],
+              "env": {
+                "TOKEN": "secret"
+              },
+              "env_vars": [
+                "APP_TOKEN",
+                "WORKSPACE_ID"
+              ],
+              "cwd": null
+            },
+            "startup_timeout_sec": null,
+            "tool_timeout_sec": null,
+            "auth_status": "unsupported"
+          }
         ]
+        )
     );
 
-    let env = entry
-        .get("env")
-        .and_then(|v| v.as_object())
-        .expect("env map");
-    assert_eq!(env.get("TOKEN"), Some(&JsonValue::String("secret".into())));
-
     let mut get_cmd = codex_command(codex_home.path())?;
     let get_output = get_cmd.args(["mcp", "get", "docs"]).output()?;
     assert!(get_output.status.success());
     let stdout = String::from_utf8(get_output.stdout)?;
     assert!(stdout.contains("docs"));
+    assert!(stdout.contains("transport: stdio"));
     assert!(stdout.contains("command: docs-server"));
     assert!(stdout.contains("args: --port 4000"));
-    assert!(stdout.contains("env: TOKEN=secret"));
+    assert!(stdout.contains("env: TOKEN=*****"));
+    assert!(stdout.contains("APP_TOKEN=*****"));
+    assert!(stdout.contains("WORKSPACE_ID=*****"));
+    assert!(stdout.contains("enabled: true"));
     assert!(stdout.contains("remove: codex mcp remove docs"));
 
     let mut get_json_cmd = codex_command(codex_home.path())?;
@@ -100,7 +132,34 @@ fn list_and_get_render_expected_output() -> Result<()> {
         .args(["mcp", "get", "docs", "--json"])
         .assert()
         .success()
-        .stdout(contains("\"name\": \"docs\""));
+        .stdout(contains("\"name\": \"docs\"").and(contains("\"enabled\": true")));
+
+    Ok(())
+}
+
+#[tokio::test]
+async fn get_disabled_server_shows_single_line() -> Result<()> {
+    let codex_home = TempDir::new()?;
+
+    let mut add = codex_command(codex_home.path())?;
+    add.args(["mcp", "add", "docs", "--", "docs-server"])
+        .assert()
+        .success();
+
+    let mut servers = load_global_mcp_servers(codex_home.path()).await?;
+    let docs = servers
+        .get_mut("docs")
+        .expect("docs server should exist after add");
+    docs.enabled = false;
+    ConfigEditsBuilder::new(codex_home.path())
+        .replace_mcp_servers(&servers)
+        .apply_blocking()?;
+
+    let mut get_cmd = codex_command(codex_home.path())?;
+    let get_output = get_cmd.args(["mcp", "get", "docs"]).output()?;
+    assert!(get_output.status.success());
+    let stdout = String::from_utf8(get_output.stdout)?;
+    assert_eq!(stdout.trim_end(), "docs (disabled)");
 
     Ok(())
 }