Extend OTLP loopback test window

Reverts openai/codex#9003

.bazelignore

@@ -0,0 +1,3 @@
+# Without this, Bazel will consider BUILD.bazel files in
+# .git/sl/origbackups (which can be populated by Sapling SCM).
+.git

.bazelrc

@@ -0,0 +1,45 @@
+common --repo_env=BAZEL_DO_NOT_DETECT_CPP_TOOLCHAIN=1
+common --repo_env=BAZEL_NO_APPLE_CPP_TOOLCHAIN=1
+
+common --disk_cache=~/.cache/bazel-disk-cache
+common --repo_contents_cache=~/.cache/bazel-repo-contents-cache
+common --repository_cache=~/.cache/bazel-repo-cache
+
+common --experimental_platform_in_output_dir
+
+common --enable_platform_specific_config
+# TODO(zbarsky): We need to untangle these libc constraints to get linux remote builds working.
+common:linux --host_platform=//:local
+common --@rules_cc//cc/toolchains/args/archiver_flags:use_libtool_on_macos=False
+common --@toolchains_llvm_bootstrapped//config:experimental_stub_libgcc_s
+
+# We need to use the sh toolchain on windows so we don't send host bash paths to the linux executor.
+common:windows --@rules_rust//rust/settings:experimental_use_sh_toolchain_for_bootstrap_process_wrapper
+
+# TODO(zbarsky): rules_rust doesn't implement this flag properly with remote exec...
+# common --@rules_rust//rust/settings:pipelined_compilation
+
+common --incompatible_strict_action_env
+# Not ideal, but We need to allow dotslash to be found
+common --test_env=PATH=/opt/homebrew/bin:/usr/local/bin:/usr/bin:/bin:/usr/sbin:/sbin
+
+common --test_output=errors
+common --bes_results_url=https://app.buildbuddy.io/invocation/
+common --bes_backend=grpcs://remote.buildbuddy.io
+common --remote_cache=grpcs://remote.buildbuddy.io
+common --remote_download_toplevel
+common --nobuild_runfile_links
+common --remote_timeout=3600
+common --noexperimental_throttle_remote_action_building
+common --experimental_remote_execution_keepalive
+common --grpc_keepalive_time=30s
+
+# This limits both in-flight executions and concurrent downloads. Even with high number
+# of jobs execution will still be limited by CPU cores, so this just pays a bit of
+# memory in exchange for higher download concurrency.
+common --jobs=30
+
+common:remote --extra_execution_platforms=//:rbe
+common:remote --remote_executor=grpcs://remote.buildbuddy.io
+common:remote --jobs=800
+

.github/workflows/Dockerfile.bazel

@@ -0,0 +1,20 @@
+FROM ubuntu:24.04
+
+# TODO(mbolin): Published to docker.io/mbolin491/codex-bazel:latest for
+# initial debugging, but we should publish to a more proper location.
+#
+# docker buildx create --use
+# docker buildx build --platform linux/amd64,linux/arm64 -f .github/workflows/Dockerfile.bazel -t mbolin491/codex-bazel:latest --push .
+
+RUN apt-get update && \
+    apt-get install -y --no-install-recommends \
+    curl git python3 ca-certificates && \
+    rm -rf /var/lib/apt/lists/*
+
+# Install dotslash.
+RUN curl -LSfs "https://github.com/facebook/dotslash/releases/download/v0.5.8/dotslash-ubuntu-22.04.$(uname -m).tar.gz" | tar fxz - -C /usr/local/bin
+
+# Ubuntu 24.04 ships with user 'ubuntu' already created with UID 1000.
+USER ubuntu
+
+WORKDIR /workspace

.github/workflows/bazel.yml

@@ -0,0 +1,110 @@
+name: Bazel (experimental)
+
+# Note this workflow was originally derived from:
+# https://github.com/cerisier/toolchains_llvm_bootstrapped/blob/main/.github/workflows/ci.yaml
+
+on:
+  pull_request: {}
+  push:
+    branches:
+      - main
+  workflow_dispatch:
+
+concurrency:
+  # Cancel previous actions from the same PR or branch except 'main' branch.
+  # See https://docs.github.com/en/actions/using-jobs/using-concurrency and https://docs.github.com/en/actions/learn-github-actions/contexts for more info.
+  group: concurrency-group::${{ github.workflow }}::${{ github.event.pull_request.number > 0 && format('pr-{0}', github.event.pull_request.number) || github.ref_name }}${{ github.ref_name == 'main' && format('::{0}', github.run_id) || ''}}
+  cancel-in-progress: ${{ github.ref_name != 'main' }}
+jobs:
+  test:
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          # macOS
+          - os: macos-15-xlarge
+            target: aarch64-apple-darwin
+          - os: macos-15-xlarge
+            target: x86_64-apple-darwin
+
+          # Linux
+          - os: ubuntu-24.04-arm
+            target: aarch64-unknown-linux-gnu
+          - os: ubuntu-24.04
+            target: x86_64-unknown-linux-gnu
+          - os: ubuntu-24.04-arm
+            target: aarch64-unknown-linux-musl
+          - os: ubuntu-24.04
+            target: x86_64-unknown-linux-musl
+          # TODO: Enable Windows once we fix the toolchain issues there.
+          #- os: windows-latest
+          #  target: x86_64-pc-windows-gnullvm
+    runs-on: ${{ matrix.os }}
+
+    # Configure a human readable name for each job
+    name: Local Bazel build on ${{ matrix.os }} for ${{ matrix.target }}
+
+    steps:
+      - uses: actions/checkout@v6
+
+      # Some integration tests rely on DotSlash being installed.
+      # See https://github.com/openai/codex/pull/7617.
+      - name: Install DotSlash
+        uses: facebook/install-dotslash@v2
+
+      - name: Make DotSlash available in PATH (Unix)
+        if: runner.os != 'Windows'
+        run: cp "$(which dotslash)" /usr/local/bin
+
+      - name: Make DotSlash available in PATH (Windows)
+        if: runner.os == 'Windows'
+        shell: pwsh
+        run: Copy-Item (Get-Command dotslash).Source -Destination "$env:LOCALAPPDATA\Microsoft\WindowsApps\dotslash.exe"
+
+      # Install Bazel via Bazelisk
+      - name: Set up Bazel
+        uses: bazelbuild/setup-bazelisk@v3
+
+      # TODO(mbolin): Bring this back once we have caching working. Currently,
+      # we never seem to get a cache hit but we still end up paying the cost of
+      # uploading at the end of the build, which takes over a minute!
+      #
+      # Cache build and external artifacts so that the next ci build is incremental.
+      # Because github action caches cannot be updated after a build, we need to
+      # store the contents of each build in a unique cache key, then fall back to loading
+      # it on the next ci run. We use hashFiles(...) in the key and restore-keys- with
+      # the prefix to load the most recent cache for the branch on a cache miss. You
+      # should customize the contents of hashFiles to capture any bazel input sources,
+      # although this doesn't need to be perfect. If none of the input sources change
+      # then a cache hit will load an existing cache and bazel won't have to do any work.
+      # In the case of a cache miss, you want the fallback cache to contain most of the
+      # previously built artifacts to minimize build time. The more precise you are with
+      # hashFiles sources the less work bazel will have to do.
+      # - name: Mount bazel caches
+      #   uses: actions/cache@v4
+      #   with:
+      #     path: |
+      #       ~/.cache/bazel-repo-cache
+      #       ~/.cache/bazel-repo-contents-cache
+      #     key: bazel-cache-${{ matrix.os }}-${{ hashFiles('**/BUILD.bazel', '**/*.bzl', 'MODULE.bazel') }}
+      #     restore-keys: |
+      #       bazel-cache-${{ matrix.os }}
+
+      - name: Configure Bazel startup args (Windows)
+        if: runner.os == 'Windows'
+        shell: pwsh
+        run: |
+          # Use a very short path to reduce argv/path length issues.
+          "BAZEL_STARTUP_ARGS=--output_user_root=C:\" | Out-File -FilePath $env:GITHUB_ENV -Encoding utf8 -Append
+
+      - name: bazel test //...
+        env:
+          BUILDBUDDY_API_KEY: ${{ secrets.BUILDBUDDY_API_KEY }}
+        shell: bash
+        run: |
+          bazel $BAZEL_STARTUP_ARGS --bazelrc=.github/workflows/ci.bazelrc test //... \
+            --build_metadata=REPO_URL=https://github.com/openai/codex.git \
+            --build_metadata=COMMIT_SHA=$(git rev-parse HEAD) \
+            --build_metadata=ROLE=CI \
+            --build_metadata=VISIBILITY=PUBLIC \
+            "--remote_header=x-buildbuddy-api-key=$BUILDBUDDY_API_KEY"

.github/workflows/cargo-deny.yml

@@ -20,7 +20,7 @@ jobs:
         uses: dtolnay/rust-toolchain@stable
 
       - name: Run cargo-deny
-        uses: EmbarkStudios/cargo-deny-action@v1
+        uses: EmbarkStudios/cargo-deny-action@v2
         with:
           rust-version: stable
           manifest-path: ./codex-rs/Cargo.toml

.github/workflows/ci.bazelrc

@@ -0,0 +1,20 @@
+common --remote_download_minimal
+common --nobuild_runfile_links
+common --keep_going
+
+# We prefer to run the build actions entirely remotely so we can dial up the concurrency.
+# We have platform-specific tests, so we want to execute the tests on all platforms using the strongest sandboxing available on each platform.
+
+# On linux, we can do a full remote build/test, by targeting the right (x86/arm) runners, so we have coverage of both.
+# Linux crossbuilds don't work until we untangle the libc constraint mess.
+common:linux --config=remote
+common:linux --strategy=remote
+common:linux --platforms=//:rbe
+
+# On mac, we can run all the build actions remotely but test actions locally.
+common:macos --config=remote
+common:macos --strategy=remote
+common:macos --strategy=TestRunner=darwin-sandbox,local
+
+common:windows --strategy=TestRunner=local
+

.github/workflows/close-stale-contributor-prs.yml

@@ -12,6 +12,8 @@ permissions:
 
 jobs:
   close-stale-contributor-prs:
+    # Prevent scheduled runs on forks
+    if: github.repository == 'openai/codex'
     runs-on: ubuntu-latest
     steps:
       - name: Close inactive PRs from contributors

.github/workflows/issue-deduplicator.yml

@@ -9,7 +9,8 @@ on:
 jobs:
   gather-duplicates:
     name: Identify potential duplicates
-    if: ${{ github.event.action == 'opened' || (github.event.action == 'labeled' && github.event.label.name == 'codex-deduplicate') }}
+    # Prevent runs on forks (requires OpenAI API key, wastes Actions minutes)
+    if: github.repository == 'openai/codex' && (github.event.action == 'opened' || (github.event.action == 'labeled' && github.event.label.name == 'codex-deduplicate'))
     runs-on: ubuntu-latest
     permissions:
       contents: read

.github/workflows/issue-labeler.yml

@@ -9,7 +9,8 @@ on:
 jobs:
   gather-labels:
     name: Generate label suggestions
-    if: ${{ github.event.action == 'opened' || (github.event.action == 'labeled' && github.event.label.name == 'codex-label') }}
+    # Prevent runs on forks (requires OpenAI API key, wastes Actions minutes)
+    if: github.repository == 'openai/codex' && (github.event.action == 'opened' || (github.event.action == 'labeled' && github.event.label.name == 'codex-label'))
     runs-on: ubuntu-latest
     permissions:
       contents: read

.github/workflows/rust-release-prepare.yml

@@ -14,6 +14,8 @@ permissions:
 
 jobs:
   prepare:
+    # Prevent scheduled runs on forks (no secrets, wastes Actions minutes)
+    if: github.repository == 'openai/codex'
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v6
@@ -41,7 +43,7 @@ jobs:
           curl --http1.1 --fail --show-error --location "${headers[@]}" "${url}" | jq '.' > codex-rs/core/models.json
 
       - name: Open pull request (if changed)
-        uses: peter-evans/create-pull-request@v7
+        uses: peter-evans/create-pull-request@v8
         with:
           commit-message: "Update models.json"
           title: "Update models.json"

.github/workflows/rust-release.yml

@@ -323,6 +323,26 @@ jobs:
       - name: Checkout repository
         uses: actions/checkout@v6
 
+      - name: Generate release notes from tag commit message
+        id: release_notes
+        shell: bash
+        run: |
+          set -euo pipefail
+
+          # On tag pushes, GITHUB_SHA may be a tag object for annotated tags;
+          # peel it to the underlying commit.
+          commit="$(git rev-parse "${GITHUB_SHA}^{commit}")"
+          notes_path="${RUNNER_TEMP}/release-notes.md"
+
+          # Use the commit message for the commit the tag points at (not the
+          # annotated tag message).
+          git log -1 --format=%B "${commit}" > "${notes_path}"
+          # Ensure trailing newline so GitHub's markdown renderer doesn't
+          # occasionally run the last line into subsequent content.
+          echo >> "${notes_path}"
+
+          echo "path=${notes_path}" >> "${GITHUB_OUTPUT}"
+
       - uses: actions/download-artifact@v7
         with:
           path: dist
@@ -395,6 +415,7 @@ jobs:
         with:
           name: ${{ steps.release_name.outputs.name }}
           tag_name: ${{ github.ref_name }}
+          body_path: ${{ steps.release_notes.outputs.path }}
           files: dist/**
           # Mark as prerelease only when the version has a suffix after x.y.z
           # (e.g. -alpha, -beta). Otherwise publish a normal release.

.gitignore

@@ -9,6 +9,7 @@ node_modules
 
 # build
 dist/
+bazel-*
 build/
 out/
 storybook-static/
@@ -85,3 +86,8 @@ CHANGELOG.ignore.md
 # nix related
 .direnv
 .envrc
+
+# Python bytecode files
+__pycache__/
+*.pyc
+

AGENTS.md

@@ -77,6 +77,12 @@ If you don’t have the tool:
 - Prefer deep equals comparisons whenever possible. Perform `assert_eq!()` on entire objects, rather than individual fields.
 - Avoid mutating process environment in tests; prefer passing environment-derived flags or dependencies from above.
 
+### Spawning workspace binaries in tests (Cargo vs Bazel)
+
+- Prefer `codex_utils_cargo_bin::cargo_bin("...")` over `assert_cmd::Command::cargo_bin(...)` or `escargot` when tests need to spawn first-party binaries.
+  - Under Bazel, binaries and resources may live under runfiles; use `codex_utils_cargo_bin::cargo_bin` to resolve absolute paths that remain stable after `chdir`.
+- When locating fixture files or test resources under Bazel, avoid `env!("CARGO_MANIFEST_DIR")`. Prefer `codex_utils_cargo_bin::find_resource!` so paths resolve correctly under both Cargo and Bazel runfiles.
+
 ### Integration tests (core)
 
 - Prefer the utilities in `core_test_support::responses` when writing end-to-end Codex tests.

BUILD.bazel

@@ -0,0 +1,19 @@
+# We mark the local platform as glibc-compatible so that rust can grab a toolchain for us.
+# TODO(zbarsky): Upstream a better libc constraint into rules_rust.
+# We only enable this on linux though for sanity, and because it breaks remote execution.
+platform(
+    name = "local",
+    constraint_values = [
+        "@toolchains_llvm_bootstrapped//constraints/libc:gnu.2.28",
+    ],
+    parents = [
+        "@platforms//host",
+    ],
+)
+
+alias(
+    name = "rbe",
+    actual = "@rbe_platform",
+)
+
+exports_files(["AGENTS.md"])

MODULE.bazel

@@ -0,0 +1,128 @@
+bazel_dep(name = "platforms", version = "1.0.0")
+bazel_dep(name = "toolchains_llvm_bootstrapped", version = "0.3.1")
+archive_override(
+    module_name = "toolchains_llvm_bootstrapped",
+    integrity = "sha256-9ks21bgEqbQWmwUIvqeLA64+Jk6o4ZVjC8KxjVa2Vw8=",
+    strip_prefix = "toolchains_llvm_bootstrapped-e3775e66a7b6d287c705ca0cd24497ef4a77c503",
+    urls = ["https://github.com/cerisier/toolchains_llvm_bootstrapped/archive/e3775e66a7b6d287c705ca0cd24497ef4a77c503/master.tar.gz"],
+    patch_strip = 1,
+    patches = [
+        "//patches:llvm_toolchain_archive_params.patch",
+    ],
+)
+
+osx = use_extension("@toolchains_llvm_bootstrapped//toolchain/extension:osx.bzl", "osx")
+osx.framework(name = "ApplicationServices")
+osx.framework(name = "AppKit")
+osx.framework(name = "ColorSync")
+osx.framework(name = "CoreFoundation")
+osx.framework(name = "CoreGraphics")
+osx.framework(name = "CoreServices")
+osx.framework(name = "CoreText")
+osx.framework(name = "CFNetwork")
+osx.framework(name = "Foundation")
+osx.framework(name = "ImageIO")
+osx.framework(name = "Kernel")
+osx.framework(name = "OSLog")
+osx.framework(name = "Security")
+osx.framework(name = "SystemConfiguration")
+
+register_toolchains(
+    "@toolchains_llvm_bootstrapped//toolchain:all",
+)
+
+bazel_dep(name = "rules_cc", version = "0.2.16")
+bazel_dep(name = "rules_platform", version = "0.1.0")
+bazel_dep(name = "rules_rust", version = "0.68.1")
+single_version_override(
+    module_name = "rules_rust",
+    patch_strip = 1,
+    patches = [
+        "//patches:rules_rust.patch",
+        "//patches:rules_rust_windows_gnu.patch",
+        "//patches:rules_rust_musl.patch",
+    ],
+)
+
+RUST_TRIPLES = [
+    "aarch64-unknown-linux-musl",
+    "aarch64-apple-darwin",
+    "aarch64-pc-windows-gnullvm",
+    "x86_64-unknown-linux-musl",
+    "x86_64-apple-darwin",
+    "x86_64-pc-windows-gnullvm",
+]
+
+rust = use_extension("@rules_rust//rust:extensions.bzl", "rust")
+rust.toolchain(
+    edition = "2024",
+    extra_target_triples = RUST_TRIPLES,
+    versions = ["1.90.0"],
+)
+use_repo(rust, "rust_toolchains")
+
+register_toolchains("@rust_toolchains//:all")
+
+bazel_dep(name = "rules_rs", version = "0.0.23")
+
+crate = use_extension("@rules_rs//rs:extensions.bzl", "crate")
+crate.from_cargo(
+    cargo_lock = "//codex-rs:Cargo.lock",
+    cargo_toml = "//codex-rs:Cargo.toml",
+    platform_triples = RUST_TRIPLES,
+)
+
+bazel_dep(name = "openssl", version = "3.5.4.bcr.0")
+
+crate.annotation(
+    build_script_data = [
+        "@openssl//:gen_dir",
+    ],
+    build_script_env = {
+        "OPENSSL_DIR": "$(execpath @openssl//:gen_dir)",
+        "OPENSSL_NO_VENDOR": "1",
+        "OPENSSL_STATIC": "1",
+    },
+    crate = "openssl-sys",
+    data = ["@openssl//:gen_dir"],
+)
+
+inject_repo(crate, "openssl")
+
+# Fix readme inclusions
+crate.annotation(
+    crate = "windows-link",
+    patch_args = ["-p1"],
+    patches = [
+        "//patches:windows-link.patch"
+    ],
+)
+
+WINDOWS_IMPORT_LIB = """
+load("@rules_cc//cc:defs.bzl", "cc_import")
+
+cc_import(
+    name = "windows_import_lib",
+    static_library = glob(["lib/*.a"])[0],
+)
+"""
+
+crate.annotation(
+    additive_build_file_content = WINDOWS_IMPORT_LIB,
+    crate = "windows_x86_64_gnullvm",
+    gen_build_script = "off",
+    deps = [":windows_import_lib"],
+)
+crate.annotation(
+    additive_build_file_content = WINDOWS_IMPORT_LIB,
+    crate = "windows_aarch64_gnullvm",
+    gen_build_script = "off",
+    deps = [":windows_import_lib"],
+)
+use_repo(crate, "crates")
+
+rbe_platform_repository = use_repo_rule("//:rbe.bzl", "rbe_platform_repository")
+
+rbe_platform_repository(
+    name = "rbe_platform",
+)

README.md

@@ -1,13 +1,11 @@
 <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>
-</br>If you want Codex in your code editor (VS Code, Cursor, Windsurf), <a href="https://developers.openai.com/codex/ide">install in your IDE</a>
-</br>If you are looking for the <em>cloud-based agent</em> from OpenAI, <strong>Codex Web</strong>, go to <a href="https://chatgpt.com/codex">chatgpt.com/codex</a></p>
-
 <p align="center">
   <img src="./.github/codex-cli-splash.png" alt="Codex CLI splash" width="80%" />
-  </p>
+</p>
+</br>
+If you want Codex in your code editor (VS Code, Cursor, Windsurf), <a href="https://developers.openai.com/codex/ide">install in your IDE.</a>
+</br>If you are looking for the <em>cloud-based agent</em> from OpenAI, <strong>Codex Web</strong>, go to <a href="https://chatgpt.com/codex">chatgpt.com/codex</a>.</p>
 
 ---
 
@@ -15,25 +13,19 @@
 
 ### Installing and running Codex CLI
 
-Install globally with your preferred package manager. If you use npm:
+Install globally with your preferred package manager:
 
 ```shell
+# Install using npm
 npm install -g @openai/codex
 ```
 
-Alternatively, if you use Homebrew:
-
 ```shell
+# Install using Homebrew
 brew install --cask codex
 ```
 
-Then simply run `codex` to get started:
-
-```shell
-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).
+Then simply run `codex` to get started.
 
 <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>
@@ -53,60 +45,15 @@ Each archive contains a single entry with the platform baked into the name (e.g.
 
 ### Using Codex with your ChatGPT plan
 
-<p align="center">
-  <img src="./.github/codex-cli-login.png" alt="Codex CLI login" width="80%" />
-  </p>
-
 Run `codex` and select **Sign in with ChatGPT**. We recommend signing into your ChatGPT account to use Codex as part of your Plus, Pro, Team, Edu, or Enterprise plan. [Learn more about what's included in your ChatGPT plan](https://help.openai.com/en/articles/11369540-codex-in-chatgpt).
 
-You can also use Codex with an API key, but this requires [additional setup](./docs/authentication.md#usage-based-billing-alternative-use-an-openai-api-key). If you previously used an API key for usage-based billing, see the [migration steps](./docs/authentication.md#migrating-from-usage-based-billing-api-key). If you're having trouble with login, please comment on [this issue](https://github.com/openai/codex/issues/1243).
+You can also use Codex with an API key, but this requires [additional setup](https://developers.openai.com/codex/auth#sign-in-with-an-api-key).
 
-### Model Context Protocol (MCP)
+## Docs
 
-Codex can access MCP servers. To configure them, refer to the [config docs](./docs/config.md#mcp_servers).
-
-### Configuration
-
-Codex CLI supports a rich set of configuration options, with preferences stored in `~/.codex/config.toml`. For full configuration options, see [Configuration](./docs/config.md).
-
-### Execpolicy
-
-See the [Execpolicy quickstart](./docs/execpolicy.md) to set up rules that govern what commands Codex can execute.
-
-### Docs & FAQ
-
-- [**Getting started**](./docs/getting-started.md)
-  - [CLI usage](./docs/getting-started.md#cli-usage)
-  - [Slash Commands](./docs/slash_commands.md)
-  - [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)
-  - [Example config](./docs/example-config.md)
-- [**Sandbox & approvals**](./docs/sandbox.md)
-- [**Execpolicy quickstart**](./docs/execpolicy.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)
-  - [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)
+- [**Codex Documentation**](https://developers.openai.com/codex)
 - [**Contributing**](./docs/contributing.md)
-- [**Install & build**](./docs/install.md)
-  - [System Requirements](./docs/install.md#system-requirements)
-  - [DotSlash](./docs/install.md#dotslash)
-  - [Build from source](./docs/install.md#build-from-source)
-- [**FAQ**](./docs/faq.md)
+- [**Installing & building**](./docs/install.md)
 - [**Open source fund**](./docs/open-source-fund.md)
 
----
-
-## License
-
 This repository is licensed under the [Apache-2.0 License](LICENSE).

announcement_tip.toml

@@ -0,0 +1,17 @@
+# Example announcement tips for Codex TUI.
+# Each [[announcements]] entry is evaluated in order; the last matching one is shown.
+# Dates are UTC, formatted as YYYY-MM-DD. The from_date is inclusive and the to_date is exclusive.
+# version_regex matches against the CLI version (env!("CARGO_PKG_VERSION")); omit to apply to all versions.
+# target_app specify which app should display the announcement (cli, vsce, ...).
+
+[[announcements]]
+content = "Welcome to Codex! Check out the new onboarding flow."
+from_date = "2024-10-01"
+to_date = "2024-10-15"
+target_app = "cli"
+
+# Test announcement only for local build version until 2026-01-10 excluded (past)
+[[announcements]]
+content = "This is a test announcement"
+version_regex = "^0\\.0\\.0$"
+to_date = "2026-01-10"

codex-cli/bin/codex.js

@@ -95,7 +95,6 @@ function detectPackageManager() {
     return "bun";
   }
 
-
   if (
     __dirname.includes(".bun/install/global") ||
     __dirname.includes(".bun\\install\\global")

codex-cli/scripts/install_native_deps.py

@@ -2,6 +2,7 @@
 """Install Codex native binaries (Rust CLI plus ripgrep helpers)."""
 
 import argparse
+from contextlib import contextmanager
 import json
 import os
 import shutil
@@ -12,6 +13,7 @@ import zipfile
 from dataclasses import dataclass
 from concurrent.futures import ThreadPoolExecutor, as_completed
 from pathlib import Path
+import sys
 from typing import Iterable, Sequence
 from urllib.parse import urlparse
 from urllib.request import urlopen
@@ -77,6 +79,45 @@ RG_TARGET_PLATFORM_PAIRS: list[tuple[str, str]] = [
 RG_TARGET_TO_PLATFORM = {target: platform for target, platform in RG_TARGET_PLATFORM_PAIRS}
 DEFAULT_RG_TARGETS = [target for target, _ in RG_TARGET_PLATFORM_PAIRS]
 
+# urllib.request.urlopen() defaults to no timeout (can hang indefinitely), which is painful in CI.
+DOWNLOAD_TIMEOUT_SECS = 60
+
+
+def _gha_enabled() -> bool:
+    # GitHub Actions supports "workflow commands" (e.g. ::group:: / ::error::) that make logs
+    # much easier to scan: groups collapse noisy sections and error annotations surface the
+    # failure in the UI without changing the actual exception/traceback output.
+    return os.environ.get("GITHUB_ACTIONS") == "true"
+
+
+def _gha_escape(value: str) -> str:
+    # Workflow commands require percent/newline escaping.
+    return value.replace("%", "%25").replace("\r", "%0D").replace("\n", "%0A")
+
+
+def _gha_error(*, title: str, message: str) -> None:
+    # Emit a GitHub Actions error annotation. This does not replace stdout/stderr logs; it just
+    # adds a prominent summary line to the job UI so the root cause is easier to spot.
+    if not _gha_enabled():
+        return
+    print(
+        f"::error title={_gha_escape(title)}::{_gha_escape(message)}",
+        flush=True,
+    )
+
+
+@contextmanager
+def _gha_group(title: str):
+    # Wrap a block in a collapsible log group on GitHub Actions. Outside of GHA this is a no-op
+    # so local output remains unchanged.
+    if _gha_enabled():
+        print(f"::group::{_gha_escape(title)}", flush=True)
+    try:
+        yield
+    finally:
+        if _gha_enabled():
+            print("::endgroup::", flush=True)
+
 
 def parse_args() -> argparse.Namespace:
     parser = argparse.ArgumentParser(description="Install native Codex binaries.")
@@ -131,18 +172,20 @@ def main() -> int:
     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_binary_components(
-            artifacts_dir,
-            vendor_dir,
-            [BINARY_COMPONENTS[name] for name in components if name in BINARY_COMPONENTS],
-        )
+    with _gha_group(f"Download 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_binary_components(
+                artifacts_dir,
+                vendor_dir,
+                [BINARY_COMPONENTS[name] for name in components if name in BINARY_COMPONENTS],
+            )
 
     if "rg" in components:
-        print("Fetching ripgrep binaries...")
-        fetch_rg(vendor_dir, DEFAULT_RG_TARGETS, manifest_path=RG_MANIFEST)
+        with _gha_group("Fetch ripgrep binaries"):
+            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
@@ -203,7 +246,14 @@ def fetch_rg(
 
         for future in as_completed(future_map):
             target = future_map[future]
-            results[target] = future.result()
+            try:
+                results[target] = future.result()
+            except Exception as exc:
+                _gha_error(
+                    title="ripgrep install failed",
+                    message=f"target={target} error={exc!r}",
+                )
+                raise RuntimeError(f"Failed to install ripgrep for target {target}.") from exc
             print(f"  installed ripgrep for {target}")
 
     return [results[target] for target in targets]
@@ -301,6 +351,8 @@ def _fetch_single_rg(
     url = providers[0]["url"]
     archive_format = platform_info.get("format", "zst")
     archive_member = platform_info.get("path")
+    digest = platform_info.get("digest")
+    expected_size = platform_info.get("size")
 
     dest_dir = vendor_dir / target / "path"
     dest_dir.mkdir(parents=True, exist_ok=True)
@@ -313,10 +365,32 @@ def _fetch_single_rg(
         tmp_dir = Path(tmp_dir_str)
         archive_filename = os.path.basename(urlparse(url).path)
         download_path = tmp_dir / archive_filename
-        _download_file(url, download_path)
+        print(
+            f"  downloading ripgrep for {target} ({platform_key}) from {url}",
+            flush=True,
+        )
+        try:
+            _download_file(url, download_path)
+        except Exception as exc:
+            _gha_error(
+                title="ripgrep download failed",
+                message=f"target={target} platform={platform_key} url={url} error={exc!r}",
+            )
+            raise RuntimeError(
+                "Failed to download ripgrep "
+                f"(target={target}, platform={platform_key}, format={archive_format}, "
+                f"expected_size={expected_size!r}, digest={digest!r}, url={url}, dest={download_path})."
+            ) from exc
 
         dest.unlink(missing_ok=True)
-        extract_archive(download_path, archive_format, archive_member, dest)
+        try:
+            extract_archive(download_path, archive_format, archive_member, dest)
+        except Exception as exc:
+            raise RuntimeError(
+                "Failed to extract ripgrep "
+                f"(target={target}, platform={platform_key}, format={archive_format}, "
+                f"member={archive_member!r}, url={url}, archive={download_path})."
+            ) from exc
 
     if not is_windows:
         dest.chmod(0o755)
@@ -326,7 +400,9 @@ def _fetch_single_rg(
 
 def _download_file(url: str, dest: Path) -> None:
     dest.parent.mkdir(parents=True, exist_ok=True)
-    with urlopen(url) as response, open(dest, "wb") as out:
+    dest.unlink(missing_ok=True)
+
+    with urlopen(url, timeout=DOWNLOAD_TIMEOUT_SECS) as response, open(dest, "wb") as out:
         shutil.copyfileobj(response, out)
 
 

codex-rs/BUILD.bazel

@@ -0,0 +1 @@
+

codex-rs/Cargo.toml

@@ -6,6 +6,7 @@ members = [
     "app-server",
     "app-server-protocol",
     "app-server-test-client",
+    "debug-client",
     "apply-patch",
     "arg0",
     "feedback",
@@ -36,6 +37,7 @@ members = [
     "tui",
     "tui2",
     "utils/absolute-path",
+    "utils/cargo-bin",
     "utils/git",
     "utils/cache",
     "utils/image",
@@ -93,6 +95,7 @@ codex-tui = { path = "tui" }
 codex-tui2 = { path = "tui2" }
 codex-utils-absolute-path = { path = "utils/absolute-path" }
 codex-utils-cache = { path = "utils/cache" }
+codex-utils-cargo-bin = { path = "utils/cargo-bin" }
 codex-utils-image = { path = "utils/image" }
 codex-utils-json-to-toml = { path = "utils/json-to-toml" }
 codex-utils-pty = { path = "utils/pty" }
@@ -132,7 +135,6 @@ dunce = "1.0.4"
 encoding_rs = "0.8.35"
 env-flags = "0.1.1"
 env_logger = "0.11.5"
-escargot = "0.5"
 eventsource-stream = "0.2.3"
 futures = { version = "0.3", default-features = false }
 http = "1.3.1"
@@ -143,14 +145,14 @@ ignore = "0.4.23"
 image = { version = "^0.25.9", default-features = false }
 include_dir = "0.7.4"
 indexmap = "2.12.0"
-insta = "1.44.3"
+insta = "1.46.0"
 itertools = "0.14.0"
 keyring = { version = "3.6", default-features = false }
-landlock = "0.4.1"
+landlock = "0.4.4"
 lazy_static = "1"
 libc = "0.2.177"
 log = "0.4"
-lru = "0.16.2"
+lru = "0.16.3"
 maplit = "1.0.2"
 mime_guess = "2.0.5"
 multimap = "0.10.0"
@@ -158,12 +160,12 @@ notify = "8.2.0"
 nucleo-matcher = "0.3.1"
 once_cell = "1.20.2"
 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"
-tracing-opentelemetry = "0.31.0"
+opentelemetry = "0.31.0"
+opentelemetry-appender-tracing = "0.31.0"
+opentelemetry-otlp = "0.31.0"
+opentelemetry-semantic-conventions = "0.31.0"
+opentelemetry_sdk = "0.31.0"
+tracing-opentelemetry = "0.32.0"
 os_info = "3.12.0"
 owo-colors = "4.2.0"
 path-absolutize = "3.1.1"
@@ -174,11 +176,12 @@ pretty_assertions = "1.4.1"
 pulldown-cmark = "0.10"
 rand = "0.9"
 ratatui = "0.29.0"
+ratatui-core = "0.1.0"
 ratatui-macros = "0.6.0"
 regex = "1.12.2"
-regex-lite = "0.1.7"
+regex-lite = "0.1.8"
 reqwest = "0.12"
-rmcp = { version = "0.10.0", default-features = false }
+rmcp = { version = "0.12.0", default-features = false }
 schemars = "0.8.22"
 seccompiler = "0.5.0"
 sentry = "0.46.0"
@@ -198,26 +201,27 @@ strum_macros = "0.27.2"
 supports-color = "3.0.2"
 sys-locale = "0.3.2"
 tempfile = "3.23.0"
-test-log = "0.2.18"
+test-log = "0.2.19"
 textwrap = "0.16.2"
 thiserror = "2.0.17"
 time = "0.3"
 tiny_http = "0.12"
 tokio = "1"
-tokio-stream = "0.1.17"
+tokio-stream = "0.1.18"
 tokio-test = "0.4"
 tokio-util = "0.7.16"
 toml = "0.9.5"
-toml_edit = "0.23.5"
-tonic = "0.13.1"
+toml_edit = "0.24.0"
 tracing = "0.1.43"
 tracing-appender = "0.2.3"
-tracing-subscriber = "0.3.20"
+tracing-subscriber = "0.3.22"
 tracing-test = "0.2.5"
 tree-sitter = "0.25.10"
 tree-sitter-bash = "0.25"
+zstd = "0.13"
 tree-sitter-highlight = "0.25.10"
 ts-rs = "11"
+tui-scrollbar = "0.2.1"
 uds_windows = "1.1.0"
 unicode-segmentation = "1.12.0"
 unicode-width = "0.2"

codex-rs/README.md

@@ -15,8 +15,8 @@ You can also install via Homebrew (`brew install --cask codex`) or download a pl
 
 ## 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).
+- First run with Codex? Start with [`docs/getting-started.md`](../docs/getting-started.md) (links to the walkthrough for prompts, keyboard shortcuts, and session management).
+- Want deeper control? See [`docs/config.md`](../docs/config.md) and [`docs/install.md`](../docs/install.md).
 
 ## What's new in the Rust CLI
 
@@ -30,7 +30,7 @@ Codex supports a rich set of configuration options. Note that the Rust CLI uses
 
 #### MCP client
 
-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.
+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#connecting-to-mcp-servers) for details.
 
 #### MCP server (experimental)
 

codex-rs/ansi-escape/BUILD.bazel

@@ -0,0 +1,6 @@
+load("//:defs.bzl", "codex_rust_crate")
+
+codex_rust_crate(
+    name = "ansi-escape",
+    crate_name = "codex_ansi_escape",
+)

codex-rs/app-server-protocol/BUILD.bazel

@@ -0,0 +1,6 @@
+load("//:defs.bzl", "codex_rust_crate")
+
+codex_rust_crate(
+    name = "app-server-protocol",
+    crate_name = "codex_app_server_protocol",
+)

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

@@ -109,14 +109,26 @@ client_request_definitions! {
         params: v2::ThreadResumeParams,
         response: v2::ThreadResumeResponse,
     },
+    ThreadFork => "thread/fork" {
+        params: v2::ThreadForkParams,
+        response: v2::ThreadForkResponse,
+    },
     ThreadArchive => "thread/archive" {
         params: v2::ThreadArchiveParams,
         response: v2::ThreadArchiveResponse,
     },
+    ThreadRollback => "thread/rollback" {
+        params: v2::ThreadRollbackParams,
+        response: v2::ThreadRollbackResponse,
+    },
     ThreadList => "thread/list" {
         params: v2::ThreadListParams,
         response: v2::ThreadListResponse,
     },
+    ThreadLoadedList => "thread/loaded/list" {
+        params: v2::ThreadLoadedListParams,
+        response: v2::ThreadLoadedListResponse,
+    },
     SkillsList => "skills/list" {
         params: v2::SkillsListParams,
         response: v2::SkillsListResponse,
@@ -193,6 +205,11 @@ client_request_definitions! {
         response: v2::ConfigWriteResponse,
     },
 
+    ConfigRequirementsRead => "configRequirements/read" {
+        params: #[ts(type = "undefined")] #[serde(skip_serializing_if = "Option::is_none")] Option<()>,
+        response: v2::ConfigRequirementsReadResponse,
+    },
+
     GetAccount => "account/read" {
         params: v2::GetAccountParams,
         response: v2::GetAccountResponse,
@@ -217,6 +234,11 @@ client_request_definitions! {
         params: v1::ResumeConversationParams,
         response: v1::ResumeConversationResponse,
     },
+    /// Fork a recorded Codex conversation into a new session.
+    ForkConversation {
+        params: v1::ForkConversationParams,
+        response: v1::ForkConversationResponse,
+    },
     ArchiveConversation {
         params: v1::ArchiveConversationParams,
         response: v1::ArchiveConversationResponse,
@@ -539,6 +561,7 @@ server_notification_definitions! {
     ReasoningSummaryPartAdded => "item/reasoning/summaryPartAdded" (v2::ReasoningSummaryPartAddedNotification),
     ReasoningTextDelta => "item/reasoning/textDelta" (v2::ReasoningTextDeltaNotification),
     ContextCompacted => "thread/compacted" (v2::ContextCompactedNotification),
+    DeprecationNotice => "deprecationNotice" (v2::DeprecationNoticeNotification),
 
     /// Notifies the user of world-writable directories on Windows, which cannot be protected by the sandbox.
     WindowsWorldWritableWarning => "windows/worldWritableWarning" (v2::WindowsWorldWritableWarningNotification),
@@ -564,7 +587,7 @@ client_notification_definitions! {
 mod tests {
     use super::*;
     use anyhow::Result;
-    use codex_protocol::ConversationId;
+    use codex_protocol::ThreadId;
     use codex_protocol::account::PlanType;
     use codex_protocol::parse_command::ParsedCommand;
     use codex_protocol::protocol::AskForApproval;
@@ -613,7 +636,7 @@ mod tests {
 
     #[test]
     fn conversation_id_serializes_as_plain_string() -> Result<()> {
-        let id = ConversationId::from_string("67e55044-10b1-426f-9247-bb680e5fe0c8")?;
+        let id = ThreadId::from_string("67e55044-10b1-426f-9247-bb680e5fe0c8")?;
 
         assert_eq!(
             json!("67e55044-10b1-426f-9247-bb680e5fe0c8"),
@@ -624,11 +647,10 @@ mod tests {
 
     #[test]
     fn conversation_id_deserializes_from_plain_string() -> Result<()> {
-        let id: ConversationId =
-            serde_json::from_value(json!("67e55044-10b1-426f-9247-bb680e5fe0c8"))?;
+        let id: ThreadId = serde_json::from_value(json!("67e55044-10b1-426f-9247-bb680e5fe0c8"))?;
 
         assert_eq!(
-            ConversationId::from_string("67e55044-10b1-426f-9247-bb680e5fe0c8")?,
+            ThreadId::from_string("67e55044-10b1-426f-9247-bb680e5fe0c8")?,
             id,
         );
         Ok(())
@@ -649,7 +671,7 @@ mod tests {
 
     #[test]
     fn serialize_server_request() -> Result<()> {
-        let conversation_id = ConversationId::from_string("67e55044-10b1-426f-9247-bb680e5fe0c8")?;
+        let conversation_id = ThreadId::from_string("67e55044-10b1-426f-9247-bb680e5fe0c8")?;
         let params = v1::ExecCommandApprovalParams {
             conversation_id,
             call_id: "call-42".to_string(),
@@ -707,6 +729,22 @@ mod tests {
         Ok(())
     }
 
+    #[test]
+    fn serialize_config_requirements_read() -> Result<()> {
+        let request = ClientRequest::ConfigRequirementsRead {
+            request_id: RequestId::Integer(1),
+            params: None,
+        };
+        assert_eq!(
+            json!({
+                "method": "configRequirements/read",
+                "id": 1,
+            }),
+            serde_json::to_value(&request)?,
+        );
+        Ok(())
+    }
+
     #[test]
     fn serialize_account_login_api_key() -> Result<()> {
         let request = ClientRequest::LoginAccount {

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

@@ -6,6 +6,7 @@ use crate::protocol::v2::UserInput;
 use codex_protocol::protocol::AgentReasoningEvent;
 use codex_protocol::protocol::AgentReasoningRawContentEvent;
 use codex_protocol::protocol::EventMsg;
+use codex_protocol::protocol::ThreadRolledBackEvent;
 use codex_protocol::protocol::TurnAbortedEvent;
 use codex_protocol::protocol::UserMessageEvent;
 
@@ -57,6 +58,7 @@ impl ThreadHistoryBuilder {
             EventMsg::TokenCount(_) => {}
             EventMsg::EnteredReviewMode(_) => {}
             EventMsg::ExitedReviewMode(_) => {}
+            EventMsg::ThreadRolledBack(payload) => self.handle_thread_rollback(payload),
             EventMsg::UndoCompleted(_) => {}
             EventMsg::TurnAborted(payload) => self.handle_turn_aborted(payload),
             _ => {}
@@ -130,6 +132,23 @@ impl ThreadHistoryBuilder {
         turn.status = TurnStatus::Interrupted;
     }
 
+    fn handle_thread_rollback(&mut self, payload: &ThreadRolledBackEvent) {
+        self.finish_current_turn();
+
+        let n = usize::try_from(payload.num_turns).unwrap_or(usize::MAX);
+        if n >= self.turns.len() {
+            self.turns.clear();
+        } else {
+            self.turns.truncate(self.turns.len().saturating_sub(n));
+        }
+
+        // Re-number subsequent synthetic ids so the pruned history is consistent.
+        self.next_turn_index =
+            i64::try_from(self.turns.len().saturating_add(1)).unwrap_or(i64::MAX);
+        let item_count: usize = self.turns.iter().map(|t| t.items.len()).sum();
+        self.next_item_index = i64::try_from(item_count.saturating_add(1)).unwrap_or(i64::MAX);
+    }
+
     fn finish_current_turn(&mut self) {
         if let Some(turn) = self.current_turn.take() {
             if turn.items.is_empty() {
@@ -213,6 +232,7 @@ mod tests {
     use codex_protocol::protocol::AgentMessageEvent;
     use codex_protocol::protocol::AgentReasoningEvent;
     use codex_protocol::protocol::AgentReasoningRawContentEvent;
+    use codex_protocol::protocol::ThreadRolledBackEvent;
     use codex_protocol::protocol::TurnAbortReason;
     use codex_protocol::protocol::TurnAbortedEvent;
     use codex_protocol::protocol::UserMessageEvent;
@@ -410,4 +430,95 @@ mod tests {
             }
         );
     }
+
+    #[test]
+    fn drops_last_turns_on_thread_rollback() {
+        let events = vec![
+            EventMsg::UserMessage(UserMessageEvent {
+                message: "First".into(),
+                images: None,
+            }),
+            EventMsg::AgentMessage(AgentMessageEvent {
+                message: "A1".into(),
+            }),
+            EventMsg::UserMessage(UserMessageEvent {
+                message: "Second".into(),
+                images: None,
+            }),
+            EventMsg::AgentMessage(AgentMessageEvent {
+                message: "A2".into(),
+            }),
+            EventMsg::ThreadRolledBack(ThreadRolledBackEvent { num_turns: 1 }),
+            EventMsg::UserMessage(UserMessageEvent {
+                message: "Third".into(),
+                images: None,
+            }),
+            EventMsg::AgentMessage(AgentMessageEvent {
+                message: "A3".into(),
+            }),
+        ];
+
+        let turns = build_turns_from_event_msgs(&events);
+        let expected = vec![
+            Turn {
+                id: "turn-1".into(),
+                status: TurnStatus::Completed,
+                error: None,
+                items: vec![
+                    ThreadItem::UserMessage {
+                        id: "item-1".into(),
+                        content: vec![UserInput::Text {
+                            text: "First".into(),
+                        }],
+                    },
+                    ThreadItem::AgentMessage {
+                        id: "item-2".into(),
+                        text: "A1".into(),
+                    },
+                ],
+            },
+            Turn {
+                id: "turn-2".into(),
+                status: TurnStatus::Completed,
+                error: None,
+                items: vec![
+                    ThreadItem::UserMessage {
+                        id: "item-3".into(),
+                        content: vec![UserInput::Text {
+                            text: "Third".into(),
+                        }],
+                    },
+                    ThreadItem::AgentMessage {
+                        id: "item-4".into(),
+                        text: "A3".into(),
+                    },
+                ],
+            },
+        ];
+        assert_eq!(turns, expected);
+    }
+
+    #[test]
+    fn thread_rollback_clears_all_turns_when_num_turns_exceeds_history() {
+        let events = vec![
+            EventMsg::UserMessage(UserMessageEvent {
+                message: "One".into(),
+                images: None,
+            }),
+            EventMsg::AgentMessage(AgentMessageEvent {
+                message: "A1".into(),
+            }),
+            EventMsg::UserMessage(UserMessageEvent {
+                message: "Two".into(),
+                images: None,
+            }),
+            EventMsg::AgentMessage(AgentMessageEvent {
+                message: "A2".into(),
+            }),
+            EventMsg::ThreadRolledBack(ThreadRolledBackEvent { num_turns: 99 }),
+        ];
+
+        let turns = build_turns_from_event_msgs(&events);
+        assert_eq!(turns, Vec::<Turn>::new());
+    }
 }

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

@@ -1,7 +1,7 @@
 use std::collections::HashMap;
 use std::path::PathBuf;
 
-use codex_protocol::ConversationId;
+use codex_protocol::ThreadId;
 use codex_protocol::config_types::ForcedLoginMethod;
 use codex_protocol::config_types::ReasoningSummary;
 use codex_protocol::config_types::SandboxMode;
@@ -68,7 +68,7 @@ pub struct NewConversationParams {
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
 #[serde(rename_all = "camelCase")]
 pub struct NewConversationResponse {
-    pub conversation_id: ConversationId,
+    pub conversation_id: ThreadId,
     pub model: String,
     pub reasoning_effort: Option<ReasoningEffort>,
     pub rollout_path: PathBuf,
@@ -77,7 +77,16 @@ pub struct NewConversationResponse {
 #[derive(Serialize, Deserialize, Debug, Clone, JsonSchema, TS)]
 #[serde(rename_all = "camelCase")]
 pub struct ResumeConversationResponse {
-    pub conversation_id: ConversationId,
+    pub conversation_id: ThreadId,
+    pub model: String,
+    pub initial_messages: Option<Vec<EventMsg>>,
+    pub rollout_path: PathBuf,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct ForkConversationResponse {
+    pub conversation_id: ThreadId,
     pub model: String,
     pub initial_messages: Option<Vec<EventMsg>>,
     pub rollout_path: PathBuf,
@@ -90,9 +99,9 @@ pub enum GetConversationSummaryParams {
         #[serde(rename = "rolloutPath")]
         rollout_path: PathBuf,
     },
-    ConversationId {
+    ThreadId {
         #[serde(rename = "conversationId")]
-        conversation_id: ConversationId,
+        conversation_id: ThreadId,
     },
 }
 
@@ -113,7 +122,7 @@ pub struct ListConversationsParams {
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
 #[serde(rename_all = "camelCase")]
 pub struct ConversationSummary {
-    pub conversation_id: ConversationId,
+    pub conversation_id: ThreadId,
     pub path: PathBuf,
     pub preview: String,
     pub timestamp: Option<String>,
@@ -143,11 +152,19 @@ pub struct ListConversationsResponse {
 #[serde(rename_all = "camelCase")]
 pub struct ResumeConversationParams {
     pub path: Option<PathBuf>,
-    pub conversation_id: Option<ConversationId>,
+    pub conversation_id: Option<ThreadId>,
     pub history: Option<Vec<ResponseItem>>,
     pub overrides: Option<NewConversationParams>,
 }
 
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+pub struct ForkConversationParams {
+    pub path: Option<PathBuf>,
+    pub conversation_id: Option<ThreadId>,
+    pub overrides: Option<NewConversationParams>,
+}
+
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
 #[serde(rename_all = "camelCase")]
 pub struct AddConversationSubscriptionResponse {
@@ -158,7 +175,7 @@ pub struct AddConversationSubscriptionResponse {
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
 #[serde(rename_all = "camelCase")]
 pub struct ArchiveConversationParams {
-    pub conversation_id: ConversationId,
+    pub conversation_id: ThreadId,
     pub rollout_path: PathBuf,
 }
 
@@ -198,7 +215,7 @@ pub struct GitDiffToRemoteResponse {
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
 #[serde(rename_all = "camelCase")]
 pub struct ApplyPatchApprovalParams {
-    pub conversation_id: ConversationId,
+    pub conversation_id: ThreadId,
     /// Use to correlate this with [codex_core::protocol::PatchApplyBeginEvent]
     /// and [codex_core::protocol::PatchApplyEndEvent].
     pub call_id: String,
@@ -219,7 +236,7 @@ pub struct ApplyPatchApprovalResponse {
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
 #[serde(rename_all = "camelCase")]
 pub struct ExecCommandApprovalParams {
-    pub conversation_id: ConversationId,
+    pub conversation_id: ThreadId,
     /// Use to correlate this with [codex_core::protocol::ExecCommandBeginEvent]
     /// and [codex_core::protocol::ExecCommandEndEvent].
     pub call_id: String,
@@ -369,14 +386,14 @@ pub struct SandboxSettings {
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
 #[serde(rename_all = "camelCase")]
 pub struct SendUserMessageParams {
-    pub conversation_id: ConversationId,
+    pub conversation_id: ThreadId,
     pub items: Vec<InputItem>,
 }
 
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
 #[serde(rename_all = "camelCase")]
 pub struct SendUserTurnParams {
-    pub conversation_id: ConversationId,
+    pub conversation_id: ThreadId,
     pub items: Vec<InputItem>,
     pub cwd: PathBuf,
     pub approval_policy: AskForApproval,
@@ -384,6 +401,8 @@ pub struct SendUserTurnParams {
     pub model: String,
     pub effort: Option<ReasoningEffort>,
     pub summary: ReasoningSummary,
+    /// Optional JSON Schema used to constrain the final assistant message for this turn.
+    pub output_schema: Option<serde_json::Value>,
 }
 
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
@@ -393,7 +412,7 @@ pub struct SendUserTurnResponse {}
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
 #[serde(rename_all = "camelCase")]
 pub struct InterruptConversationParams {
-    pub conversation_id: ConversationId,
+    pub conversation_id: ThreadId,
 }
 
 #[derive(Serialize, Deserialize, Debug, Clone, JsonSchema, TS)]
@@ -409,7 +428,7 @@ pub struct SendUserMessageResponse {}
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
 #[serde(rename_all = "camelCase")]
 pub struct AddConversationListenerParams {
-    pub conversation_id: ConversationId,
+    pub conversation_id: ThreadId,
     #[serde(default)]
     pub experimental_raw_events: bool,
 }
@@ -443,7 +462,7 @@ pub struct LoginChatGptCompleteNotification {
 #[derive(Serialize, Deserialize, Debug, Clone, JsonSchema, TS)]
 #[serde(rename_all = "camelCase")]
 pub struct SessionConfiguredNotification {
-    pub session_id: ConversationId,
+    pub session_id: ThreadId,
     pub model: String,
     pub reasoning_effort: Option<ReasoningEffort>,
     pub history_log_id: u64,

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

@@ -18,6 +18,7 @@ use codex_protocol::plan_tool::StepStatus as CorePlanStepStatus;
 use codex_protocol::protocol::AskForApproval as CoreAskForApproval;
 use codex_protocol::protocol::CodexErrorInfo as CoreCodexErrorInfo;
 use codex_protocol::protocol::CreditsSnapshot as CoreCreditsSnapshot;
+use codex_protocol::protocol::NetworkAccess as CoreNetworkAccess;
 use codex_protocol::protocol::RateLimitSnapshot as CoreRateLimitSnapshot;
 use codex_protocol::protocol::RateLimitWindow as CoreRateLimitWindow;
 use codex_protocol::protocol::SessionSource as CoreSessionSource;
@@ -88,6 +89,7 @@ pub enum CodexErrorInfo {
     InternalServerError,
     Unauthorized,
     BadRequest,
+    ThreadRollbackFailed,
     SandboxError,
     /// The response SSE stream disconnected in the middle of a turn before completion.
     ResponseStreamDisconnected {
@@ -118,6 +120,7 @@ impl From<CoreCodexErrorInfo> for CodexErrorInfo {
             CoreCodexErrorInfo::InternalServerError => CodexErrorInfo::InternalServerError,
             CoreCodexErrorInfo::Unauthorized => CodexErrorInfo::Unauthorized,
             CoreCodexErrorInfo::BadRequest => CodexErrorInfo::BadRequest,
+            CoreCodexErrorInfo::ThreadRollbackFailed => CodexErrorInfo::ThreadRollbackFailed,
             CoreCodexErrorInfo::SandboxError => CodexErrorInfo::SandboxError,
             CoreCodexErrorInfo::ResponseStreamDisconnected { http_status_code } => {
                 CodexErrorInfo::ResponseStreamDisconnected { http_status_code }
@@ -226,6 +229,8 @@ pub enum ConfigLayerSource {
     #[serde(rename_all = "camelCase")]
     #[ts(rename_all = "camelCase")]
     System {
+        /// This is the path to the system config.toml file, though it is not
+        /// guaranteed to exist.
         file: AbsolutePathBuf,
     },
 
@@ -236,9 +241,19 @@ pub enum ConfigLayerSource {
     #[serde(rename_all = "camelCase")]
     #[ts(rename_all = "camelCase")]
     User {
+        /// This is the path to the user's config.toml file, though it is not
+        /// guaranteed to exist.
         file: AbsolutePathBuf,
     },
 
+    /// Path to a .codex/ folder within a project. There could be multiple of
+    /// these between `cwd` and the project/repo root.
+    #[serde(rename_all = "camelCase")]
+    #[ts(rename_all = "camelCase")]
+    Project {
+        dot_codex_folder: AbsolutePathBuf,
+    },
+
     /// Session-layer overrides supplied via `-c`/`--config`.
     SessionFlags,
 
@@ -246,6 +261,8 @@ pub enum ConfigLayerSource {
     /// as the last layer on top of everything else. This scheme did not quite
     /// work out as intended, but we keep this variant as a "best effort" while
     /// we phase out `managed_config.toml` in favor of `requirements.toml`.
+    #[serde(rename_all = "camelCase")]
+    #[ts(rename_all = "camelCase")]
     LegacyManagedConfigTomlFromFile {
         file: AbsolutePathBuf,
     },
@@ -261,6 +278,7 @@ impl ConfigLayerSource {
             ConfigLayerSource::Mdm { .. } => 0,
             ConfigLayerSource::System { .. } => 10,
             ConfigLayerSource::User { .. } => 20,
+            ConfigLayerSource::Project { .. } => 25,
             ConfigLayerSource::SessionFlags => 30,
             ConfigLayerSource::LegacyManagedConfigTomlFromFile { .. } => 40,
             ConfigLayerSource::LegacyManagedConfigTomlFromMdm => 50,
@@ -314,6 +332,15 @@ pub struct ProfileV2 {
     pub additional: HashMap<String, JsonValue>,
 }
 
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "snake_case")]
+#[ts(export_to = "v2/")]
+pub struct AnalyticsConfig {
+    pub enabled: Option<bool>,
+    #[serde(default, flatten)]
+    pub additional: HashMap<String, JsonValue>,
+}
+
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
 #[serde(rename_all = "snake_case")]
 #[ts(export_to = "v2/")]
@@ -338,6 +365,7 @@ pub struct Config {
     pub model_reasoning_effort: Option<ReasoningEffort>,
     pub model_reasoning_summary: Option<ReasoningSummary>,
     pub model_verbosity: Option<Verbosity>,
+    pub analytics: Option<AnalyticsConfig>,
     #[serde(default, flatten)]
     pub additional: HashMap<String, JsonValue>,
 }
@@ -425,6 +453,22 @@ pub struct ConfigReadResponse {
     pub layers: Option<Vec<ConfigLayer>>,
 }
 
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+#[ts(export_to = "v2/")]
+pub struct ConfigRequirements {
+    pub allowed_approval_policies: Option<Vec<AskForApproval>>,
+    pub allowed_sandbox_modes: Option<Vec<SandboxMode>>,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+#[ts(export_to = "v2/")]
+pub struct ConfigRequirementsReadResponse {
+    /// Null if no requirements are configured (e.g. no requirements.toml/MDM entries).
+    pub requirements: Option<ConfigRequirements>,
+}
+
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
 #[serde(rename_all = "camelCase")]
 #[ts(export_to = "v2/")]
@@ -459,17 +503,45 @@ pub struct ConfigEdit {
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Eq, JsonSchema, TS)]
 #[serde(rename_all = "camelCase")]
 #[ts(export_to = "v2/")]
-pub enum ApprovalDecision {
+pub enum CommandExecutionApprovalDecision {
+    /// User approved the command.
     Accept,
-    /// Approve and remember the approval for the session.
+    /// User approved the command and future identical commands should run without prompting.
     AcceptForSession,
+    /// User approved the command, and wants to apply the proposed execpolicy amendment so future
+    /// matching commands can run without prompting.
     AcceptWithExecpolicyAmendment {
         execpolicy_amendment: ExecPolicyAmendment,
     },
+    /// User denied the command. The agent will continue the turn.
     Decline,
+    /// User denied the command. The turn will also be immediately interrupted.
     Cancel,
 }
 
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Eq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+#[ts(export_to = "v2/")]
+pub enum FileChangeApprovalDecision {
+    /// User approved the file changes.
+    Accept,
+    /// User approved the file changes and future changes to the same files should run without prompting.
+    AcceptForSession,
+    /// User denied the file changes. The agent will continue the turn.
+    Decline,
+    /// User denied the file changes. The turn will also be immediately interrupted.
+    Cancel,
+}
+
+#[derive(Serialize, Deserialize, Debug, Default, Clone, PartialEq, Eq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+#[ts(export_to = "v2/")]
+pub enum NetworkAccess {
+    #[default]
+    Restricted,
+    Enabled,
+}
+
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Eq, JsonSchema, TS)]
 #[serde(tag = "type", rename_all = "camelCase")]
 #[ts(tag = "type")]
@@ -479,6 +551,12 @@ pub enum SandboxPolicy {
     ReadOnly,
     #[serde(rename_all = "camelCase")]
     #[ts(rename_all = "camelCase")]
+    ExternalSandbox {
+        #[serde(default)]
+        network_access: NetworkAccess,
+    },
+    #[serde(rename_all = "camelCase")]
+    #[ts(rename_all = "camelCase")]
     WorkspaceWrite {
         #[serde(default)]
         writable_roots: Vec<AbsolutePathBuf>,
@@ -498,6 +576,14 @@ impl SandboxPolicy {
                 codex_protocol::protocol::SandboxPolicy::DangerFullAccess
             }
             SandboxPolicy::ReadOnly => codex_protocol::protocol::SandboxPolicy::ReadOnly,
+            SandboxPolicy::ExternalSandbox { network_access } => {
+                codex_protocol::protocol::SandboxPolicy::ExternalSandbox {
+                    network_access: match network_access {
+                        NetworkAccess::Restricted => CoreNetworkAccess::Restricted,
+                        NetworkAccess::Enabled => CoreNetworkAccess::Enabled,
+                    },
+                }
+            }
             SandboxPolicy::WorkspaceWrite {
                 writable_roots,
                 network_access,
@@ -520,6 +606,14 @@ impl From<codex_protocol::protocol::SandboxPolicy> for SandboxPolicy {
                 SandboxPolicy::DangerFullAccess
             }
             codex_protocol::protocol::SandboxPolicy::ReadOnly => SandboxPolicy::ReadOnly,
+            codex_protocol::protocol::SandboxPolicy::ExternalSandbox { network_access } => {
+                SandboxPolicy::ExternalSandbox {
+                    network_access: match network_access {
+                        CoreNetworkAccess::Restricted => NetworkAccess::Restricted,
+                        CoreNetworkAccess::Enabled => NetworkAccess::Enabled,
+                    },
+                }
+            }
             codex_protocol::protocol::SandboxPolicy::WorkspaceWrite {
                 writable_roots,
                 network_access,
@@ -986,6 +1080,47 @@ pub struct ThreadResumeResponse {
     pub reasoning_effort: Option<ReasoningEffort>,
 }
 
+#[derive(Serialize, Deserialize, Debug, Default, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+#[ts(export_to = "v2/")]
+/// There are two ways to fork a thread:
+/// 1. By thread_id: load the thread from disk by thread_id and fork it into a new thread.
+/// 2. By path: load the thread from disk by path and fork it into a new thread.
+///
+/// If using path, the thread_id param will be ignored.
+///
+/// Prefer using thread_id whenever possible.
+pub struct ThreadForkParams {
+    pub thread_id: String,
+
+    /// [UNSTABLE] Specify the rollout path to fork from.
+    /// If specified, the thread_id param will be ignored.
+    pub path: Option<PathBuf>,
+
+    /// Configuration overrides for the forked thread, if any.
+    pub model: Option<String>,
+    pub model_provider: 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>,
+    pub developer_instructions: Option<String>,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+#[ts(export_to = "v2/")]
+pub struct ThreadForkResponse {
+    pub thread: Thread,
+    pub model: String,
+    pub model_provider: String,
+    pub cwd: PathBuf,
+    pub approval_policy: AskForApproval,
+    pub sandbox: SandboxPolicy,
+    pub reasoning_effort: Option<ReasoningEffort>,
+}
+
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
 #[serde(rename_all = "camelCase")]
 #[ts(export_to = "v2/")]
@@ -998,6 +1133,30 @@ pub struct ThreadArchiveParams {
 #[ts(export_to = "v2/")]
 pub struct ThreadArchiveResponse {}
 
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+#[ts(export_to = "v2/")]
+pub struct ThreadRollbackParams {
+    pub thread_id: String,
+    /// The number of turns to drop from the end of the thread. Must be >= 1.
+    ///
+    /// This only modifies the thread's history and does not revert local file changes
+    /// that have been made by the agent. Clients are responsible for reverting these changes.
+    pub num_turns: u32,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+#[ts(export_to = "v2/")]
+pub struct ThreadRollbackResponse {
+    /// The updated thread after applying the rollback, with `turns` populated.
+    ///
+    /// The ThreadItems stored in each Turn are lossy since we explicitly do not
+    /// persist all agent interactions, such as command executions. This is the same
+    /// behavior as `thread/resume`.
+    pub thread: Thread,
+}
+
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
 #[serde(rename_all = "camelCase")]
 #[ts(export_to = "v2/")]
@@ -1021,6 +1180,27 @@ pub struct ThreadListResponse {
     pub next_cursor: Option<String>,
 }
 
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Default, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+#[ts(export_to = "v2/")]
+pub struct ThreadLoadedListParams {
+    /// Opaque pagination cursor returned by a previous call.
+    pub cursor: Option<String>,
+    /// Optional page size; defaults to no limit.
+    pub limit: Option<u32>,
+}
+
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+#[ts(export_to = "v2/")]
+pub struct ThreadLoadedListResponse {
+    /// Thread ids for sessions currently loaded in memory.
+    pub data: Vec<String>,
+    /// 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")]
 #[ts(export_to = "v2/")]
@@ -1049,6 +1229,7 @@ pub enum SkillScope {
     User,
     Repo,
     System,
+    Admin,
 }
 
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
@@ -1057,6 +1238,9 @@ pub enum SkillScope {
 pub struct SkillMetadata {
     pub name: String,
     pub description: String,
+    #[ts(optional)]
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub short_description: Option<String>,
     pub path: PathBuf,
     pub scope: SkillScope,
 }
@@ -1083,6 +1267,7 @@ impl From<CoreSkillMetadata> for SkillMetadata {
         Self {
             name: value.name,
             description: value.description,
+            short_description: value.short_description,
             path: value.path,
             scope: value.scope.into(),
         }
@@ -1095,6 +1280,7 @@ impl From<CoreSkillScope> for SkillScope {
             CoreSkillScope::User => Self::User,
             CoreSkillScope::Repo => Self::Repo,
             CoreSkillScope::System => Self::System,
+            CoreSkillScope::Admin => Self::Admin,
         }
     }
 }
@@ -1130,7 +1316,7 @@ pub struct Thread {
     pub source: SessionSource,
     /// Optional Git metadata captured when the thread was created.
     pub git_info: Option<GitInfo>,
-    /// Only populated on a `thread/resume` response.
+    /// Only populated on `thread/resume`, `thread/rollback`, `thread/fork` responses.
     /// For all other responses and notifications returning a Thread,
     /// the turns field will be an empty list.
     pub turns: Vec<Turn>,
@@ -1158,6 +1344,7 @@ pub struct ThreadTokenUsageUpdatedNotification {
 pub struct ThreadTokenUsage {
     pub total: TokenUsageBreakdown,
     pub last: TokenUsageBreakdown,
+    // TODO(aibrahim): make this not optional
     #[ts(type = "number | null")]
     pub model_context_window: Option<i64>,
 }
@@ -1205,7 +1392,7 @@ impl From<CoreTokenUsage> for TokenUsageBreakdown {
 #[ts(export_to = "v2/")]
 pub struct Turn {
     pub id: String,
-    /// Only populated on a `thread/resume` response.
+    /// Only populated on a `thread/resume` or `thread/fork` response.
     /// For all other responses and notifications returning a Turn,
     /// the items field will be an empty list.
     pub items: Vec<ThreadItem>,
@@ -1221,6 +1408,8 @@ pub struct Turn {
 pub struct TurnError {
     pub message: String,
     pub codex_error_info: Option<CodexErrorInfo>,
+    #[serde(default)]
+    pub additional_details: Option<String>,
 }
 
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
@@ -1264,6 +1453,8 @@ pub struct TurnStartParams {
     pub effort: Option<ReasoningEffort>,
     /// Override the reasoning summary for this turn and subsequent turns.
     pub summary: Option<ReasoningSummary>,
+    /// Optional JSON Schema used to constrain the final assistant message for this turn.
+    pub output_schema: Option<JsonValue>,
 }
 
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
@@ -1347,6 +1538,7 @@ pub enum UserInput {
     Text { text: String },
     Image { url: String },
     LocalImage { path: PathBuf },
+    Skill { name: String, path: PathBuf },
 }
 
 impl UserInput {
@@ -1355,6 +1547,7 @@ impl UserInput {
             UserInput::Text { text } => CoreUserInput::Text { text },
             UserInput::Image { url } => CoreUserInput::Image { image_url: url },
             UserInput::LocalImage { path } => CoreUserInput::LocalImage { path },
+            UserInput::Skill { name, path } => CoreUserInput::Skill { name, path },
         }
     }
 }
@@ -1365,6 +1558,7 @@ impl From<CoreUserInput> for UserInput {
             CoreUserInput::Text { text } => UserInput::Text { text },
             CoreUserInput::Image { image_url } => UserInput::Image { url: image_url },
             CoreUserInput::LocalImage { path } => UserInput::LocalImage { path },
+            CoreUserInput::Skill { name, path } => UserInput::Skill { name, path },
             _ => unreachable!("unsupported user input variant"),
         }
     }
@@ -1791,7 +1985,7 @@ pub struct CommandExecutionRequestApprovalParams {
 #[serde(rename_all = "camelCase")]
 #[ts(export_to = "v2/")]
 pub struct CommandExecutionRequestApprovalResponse {
-    pub decision: ApprovalDecision,
+    pub decision: CommandExecutionApprovalDecision,
 }
 
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
@@ -1811,7 +2005,7 @@ pub struct FileChangeRequestApprovalParams {
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
 #[ts(export_to = "v2/")]
 pub struct FileChangeRequestApprovalResponse {
-    pub decision: ApprovalDecision,
+    pub decision: FileChangeApprovalDecision,
 }
 
 #[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
@@ -1893,6 +2087,16 @@ pub struct AccountLoginCompletedNotification {
     pub error: Option<String>,
 }
 
+#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
+#[serde(rename_all = "camelCase")]
+#[ts(export_to = "v2/")]
+pub struct DeprecationNoticeNotification {
+    /// Concise summary of what is deprecated.
+    pub summary: String,
+    /// Optional extra guidance, such as migration steps or rationale.
+    pub details: Option<String>,
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;
@@ -1902,11 +2106,30 @@ mod tests {
     use codex_protocol::items::TurnItem;
     use codex_protocol::items::UserMessageItem;
     use codex_protocol::items::WebSearchItem;
+    use codex_protocol::protocol::NetworkAccess as CoreNetworkAccess;
     use codex_protocol::user_input::UserInput as CoreUserInput;
     use pretty_assertions::assert_eq;
     use serde_json::json;
     use std::path::PathBuf;
 
+    #[test]
+    fn sandbox_policy_round_trips_external_sandbox_network_access() {
+        let v2_policy = SandboxPolicy::ExternalSandbox {
+            network_access: NetworkAccess::Enabled,
+        };
+
+        let core_policy = v2_policy.to_core();
+        assert_eq!(
+            core_policy,
+            codex_protocol::protocol::SandboxPolicy::ExternalSandbox {
+                network_access: CoreNetworkAccess::Enabled,
+            }
+        );
+
+        let back_to_v2 = SandboxPolicy::from(core_policy);
+        assert_eq!(back_to_v2, v2_policy);
+    }
+
     #[test]
     fn core_turn_item_into_thread_item_converts_supported_variants() {
         let user_item = TurnItem::UserMessage(UserMessageItem {
@@ -1921,6 +2144,10 @@ mod tests {
                 CoreUserInput::LocalImage {
                     path: PathBuf::from("local/image.png"),
                 },
+                CoreUserInput::Skill {
+                    name: "skill-creator".to_string(),
+                    path: PathBuf::from("/repo/.codex/skills/skill-creator/SKILL.md"),
+                },
             ],
         });
 
@@ -1938,6 +2165,10 @@ mod tests {
                     UserInput::LocalImage {
                         path: PathBuf::from("local/image.png"),
                     },
+                    UserInput::Skill {
+                        name: "skill-creator".to_string(),
+                        path: PathBuf::from("/repo/.codex/skills/skill-creator/SKILL.md"),
+                    },
                 ],
             }
         );

codex-rs/app-server-test-client/BUILD.bazel

@@ -0,0 +1,6 @@
+load("//:defs.bzl", "codex_rust_crate")
+
+codex_rust_crate(
+    name = "codex-app-server-test-client",
+    crate_name = "codex_app_server_test_client",
+)

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

@@ -13,16 +13,18 @@ use std::time::Duration;
 use anyhow::Context;
 use anyhow::Result;
 use anyhow::bail;
+use clap::ArgAction;
 use clap::Parser;
 use clap::Subcommand;
 use codex_app_server_protocol::AddConversationListenerParams;
 use codex_app_server_protocol::AddConversationSubscriptionResponse;
-use codex_app_server_protocol::ApprovalDecision;
 use codex_app_server_protocol::AskForApproval;
 use codex_app_server_protocol::ClientInfo;
 use codex_app_server_protocol::ClientRequest;
+use codex_app_server_protocol::CommandExecutionApprovalDecision;
 use codex_app_server_protocol::CommandExecutionRequestApprovalParams;
 use codex_app_server_protocol::CommandExecutionRequestApprovalResponse;
+use codex_app_server_protocol::FileChangeApprovalDecision;
 use codex_app_server_protocol::FileChangeRequestApprovalParams;
 use codex_app_server_protocol::FileChangeRequestApprovalResponse;
 use codex_app_server_protocol::GetAccountRateLimitsResponse;
@@ -35,6 +37,8 @@ use codex_app_server_protocol::JSONRPCRequest;
 use codex_app_server_protocol::JSONRPCResponse;
 use codex_app_server_protocol::LoginChatGptCompleteNotification;
 use codex_app_server_protocol::LoginChatGptResponse;
+use codex_app_server_protocol::ModelListParams;
+use codex_app_server_protocol::ModelListResponse;
 use codex_app_server_protocol::NewConversationParams;
 use codex_app_server_protocol::NewConversationResponse;
 use codex_app_server_protocol::RequestId;
@@ -49,7 +53,7 @@ use codex_app_server_protocol::TurnStartParams;
 use codex_app_server_protocol::TurnStartResponse;
 use codex_app_server_protocol::TurnStatus;
 use codex_app_server_protocol::UserInput as V2UserInput;
-use codex_protocol::ConversationId;
+use codex_protocol::ThreadId;
 use codex_protocol::protocol::Event;
 use codex_protocol::protocol::EventMsg;
 use serde::Serialize;
@@ -65,6 +69,19 @@ struct Cli {
     #[arg(long, env = "CODEX_BIN", default_value = "codex")]
     codex_bin: String,
 
+    /// Forwarded to the `codex` CLI as `--config key=value`. Repeatable.
+    ///
+    /// Example:
+    ///   `--config 'model_providers.mock.base_url="http://localhost:4010/v2"'`
+    #[arg(
+        short = 'c',
+        long = "config",
+        value_name = "key=value",
+        action = ArgAction::Append,
+        global = true
+    )]
+    config_overrides: Vec<String>,
+
     #[command(subcommand)]
     command: CliCommand,
 }
@@ -113,37 +130,54 @@ enum CliCommand {
     TestLogin,
     /// Fetch the current account rate limits from the Codex app-server.
     GetAccountRateLimits,
+    /// List the available models from the Codex app-server.
+    #[command(name = "model-list")]
+    ModelList,
 }
 
 fn main() -> Result<()> {
-    let Cli { codex_bin, command } = Cli::parse();
+    let Cli {
+        codex_bin,
+        config_overrides,
+        command,
+    } = Cli::parse();
 
     match command {
-        CliCommand::SendMessage { user_message } => send_message(codex_bin, user_message),
-        CliCommand::SendMessageV2 { user_message } => send_message_v2(codex_bin, user_message),
+        CliCommand::SendMessage { user_message } => {
+            send_message(&codex_bin, &config_overrides, user_message)
+        }
+        CliCommand::SendMessageV2 { user_message } => {
+            send_message_v2(&codex_bin, &config_overrides, user_message)
+        }
         CliCommand::TriggerCmdApproval { user_message } => {
-            trigger_cmd_approval(codex_bin, user_message)
+            trigger_cmd_approval(&codex_bin, &config_overrides, user_message)
         }
         CliCommand::TriggerPatchApproval { user_message } => {
-            trigger_patch_approval(codex_bin, user_message)
+            trigger_patch_approval(&codex_bin, &config_overrides, user_message)
         }
-        CliCommand::NoTriggerCmdApproval => no_trigger_cmd_approval(codex_bin),
+        CliCommand::NoTriggerCmdApproval => no_trigger_cmd_approval(&codex_bin, &config_overrides),
         CliCommand::SendFollowUpV2 {
             first_message,
             follow_up_message,
-        } => send_follow_up_v2(codex_bin, first_message, follow_up_message),
-        CliCommand::TestLogin => test_login(codex_bin),
-        CliCommand::GetAccountRateLimits => get_account_rate_limits(codex_bin),
+        } => send_follow_up_v2(
+            &codex_bin,
+            &config_overrides,
+            first_message,
+            follow_up_message,
+        ),
+        CliCommand::TestLogin => test_login(&codex_bin, &config_overrides),
+        CliCommand::GetAccountRateLimits => get_account_rate_limits(&codex_bin, &config_overrides),
+        CliCommand::ModelList => model_list(&codex_bin, &config_overrides),
     }
 }
 
-fn send_message(codex_bin: String, user_message: String) -> Result<()> {
-    let mut client = CodexClient::spawn(codex_bin)?;
+fn send_message(codex_bin: &str, config_overrides: &[String], user_message: String) -> Result<()> {
+    let mut client = CodexClient::spawn(codex_bin, config_overrides)?;
 
     let initialize = client.initialize()?;
     println!("< initialize response: {initialize:?}");
 
-    let conversation = client.new_conversation()?;
+    let conversation = client.start_thread()?;
     println!("< newConversation response: {conversation:?}");
 
     let subscription = client.add_conversation_listener(&conversation.conversation_id)?;
@@ -154,51 +188,66 @@ fn send_message(codex_bin: String, user_message: String) -> Result<()> {
 
     client.stream_conversation(&conversation.conversation_id)?;
 
-    client.remove_conversation_listener(subscription.subscription_id)?;
+    client.remove_thread_listener(subscription.subscription_id)?;
 
     Ok(())
 }
 
-fn send_message_v2(codex_bin: String, user_message: String) -> Result<()> {
-    send_message_v2_with_policies(codex_bin, user_message, None, None)
+fn send_message_v2(
+    codex_bin: &str,
+    config_overrides: &[String],
+    user_message: String,
+) -> Result<()> {
+    send_message_v2_with_policies(codex_bin, config_overrides, user_message, None, None)
 }
 
-fn trigger_cmd_approval(codex_bin: String, user_message: Option<String>) -> Result<()> {
+fn trigger_cmd_approval(
+    codex_bin: &str,
+    config_overrides: &[String],
+    user_message: Option<String>,
+) -> Result<()> {
     let default_prompt =
         "Run `touch /tmp/should-trigger-approval` so I can confirm the file exists.";
     let message = user_message.unwrap_or_else(|| default_prompt.to_string());
     send_message_v2_with_policies(
         codex_bin,
+        config_overrides,
         message,
         Some(AskForApproval::OnRequest),
         Some(SandboxPolicy::ReadOnly),
     )
 }
 
-fn trigger_patch_approval(codex_bin: String, user_message: Option<String>) -> Result<()> {
+fn trigger_patch_approval(
+    codex_bin: &str,
+    config_overrides: &[String],
+    user_message: Option<String>,
+) -> Result<()> {
     let default_prompt =
         "Create a file named APPROVAL_DEMO.txt containing a short hello message using apply_patch.";
     let message = user_message.unwrap_or_else(|| default_prompt.to_string());
     send_message_v2_with_policies(
         codex_bin,
+        config_overrides,
         message,
         Some(AskForApproval::OnRequest),
         Some(SandboxPolicy::ReadOnly),
     )
 }
 
-fn no_trigger_cmd_approval(codex_bin: String) -> Result<()> {
+fn no_trigger_cmd_approval(codex_bin: &str, config_overrides: &[String]) -> Result<()> {
     let prompt = "Run `touch should_not_trigger_approval.txt`";
-    send_message_v2_with_policies(codex_bin, prompt.to_string(), None, None)
+    send_message_v2_with_policies(codex_bin, config_overrides, prompt.to_string(), None, None)
 }
 
 fn send_message_v2_with_policies(
-    codex_bin: String,
+    codex_bin: &str,
+    config_overrides: &[String],
     user_message: String,
     approval_policy: Option<AskForApproval>,
     sandbox_policy: Option<SandboxPolicy>,
 ) -> Result<()> {
-    let mut client = CodexClient::spawn(codex_bin)?;
+    let mut client = CodexClient::spawn(codex_bin, config_overrides)?;
 
     let initialize = client.initialize()?;
     println!("< initialize response: {initialize:?}");
@@ -222,11 +271,12 @@ fn send_message_v2_with_policies(
 }
 
 fn send_follow_up_v2(
-    codex_bin: String,
+    codex_bin: &str,
+    config_overrides: &[String],
     first_message: String,
     follow_up_message: String,
 ) -> Result<()> {
-    let mut client = CodexClient::spawn(codex_bin)?;
+    let mut client = CodexClient::spawn(codex_bin, config_overrides)?;
 
     let initialize = client.initialize()?;
     println!("< initialize response: {initialize:?}");
@@ -259,8 +309,8 @@ fn send_follow_up_v2(
     Ok(())
 }
 
-fn test_login(codex_bin: String) -> Result<()> {
-    let mut client = CodexClient::spawn(codex_bin)?;
+fn test_login(codex_bin: &str, config_overrides: &[String]) -> Result<()> {
+    let mut client = CodexClient::spawn(codex_bin, config_overrides)?;
 
     let initialize = client.initialize()?;
     println!("< initialize response: {initialize:?}");
@@ -289,8 +339,8 @@ fn test_login(codex_bin: String) -> Result<()> {
     }
 }
 
-fn get_account_rate_limits(codex_bin: String) -> Result<()> {
-    let mut client = CodexClient::spawn(codex_bin)?;
+fn get_account_rate_limits(codex_bin: &str, config_overrides: &[String]) -> Result<()> {
+    let mut client = CodexClient::spawn(codex_bin, config_overrides)?;
 
     let initialize = client.initialize()?;
     println!("< initialize response: {initialize:?}");
@@ -301,6 +351,18 @@ fn get_account_rate_limits(codex_bin: String) -> Result<()> {
     Ok(())
 }
 
+fn model_list(codex_bin: &str, config_overrides: &[String]) -> Result<()> {
+    let mut client = CodexClient::spawn(codex_bin, config_overrides)?;
+
+    let initialize = client.initialize()?;
+    println!("< initialize response: {initialize:?}");
+
+    let response = client.model_list(ModelListParams::default())?;
+    println!("< model/list response: {response:?}");
+
+    Ok(())
+}
+
 struct CodexClient {
     child: Child,
     stdin: Option<ChildStdin>,
@@ -309,8 +371,12 @@ struct CodexClient {
 }
 
 impl CodexClient {
-    fn spawn(codex_bin: String) -> Result<Self> {
-        let mut codex_app_server = Command::new(&codex_bin)
+    fn spawn(codex_bin: &str, config_overrides: &[String]) -> Result<Self> {
+        let mut cmd = Command::new(codex_bin);
+        for override_kv in config_overrides {
+            cmd.arg("--config").arg(override_kv);
+        }
+        let mut codex_app_server = cmd
             .arg("app-server")
             .stdin(Stdio::piped())
             .stdout(Stdio::piped())
@@ -351,7 +417,7 @@ impl CodexClient {
         self.send_request(request, request_id, "initialize")
     }
 
-    fn new_conversation(&mut self) -> Result<NewConversationResponse> {
+    fn start_thread(&mut self) -> Result<NewConversationResponse> {
         let request_id = self.request_id();
         let request = ClientRequest::NewConversation {
             request_id: request_id.clone(),
@@ -363,7 +429,7 @@ impl CodexClient {
 
     fn add_conversation_listener(
         &mut self,
-        conversation_id: &ConversationId,
+        conversation_id: &ThreadId,
     ) -> Result<AddConversationSubscriptionResponse> {
         let request_id = self.request_id();
         let request = ClientRequest::AddConversationListener {
@@ -377,7 +443,7 @@ impl CodexClient {
         self.send_request(request, request_id, "addConversationListener")
     }
 
-    fn remove_conversation_listener(&mut self, subscription_id: Uuid) -> Result<()> {
+    fn remove_thread_listener(&mut self, subscription_id: Uuid) -> Result<()> {
         let request_id = self.request_id();
         let request = ClientRequest::RemoveConversationListener {
             request_id: request_id.clone(),
@@ -395,7 +461,7 @@ impl CodexClient {
 
     fn send_user_message(
         &mut self,
-        conversation_id: &ConversationId,
+        conversation_id: &ThreadId,
         message: &str,
     ) -> Result<SendUserMessageResponse> {
         let request_id = self.request_id();
@@ -452,7 +518,17 @@ impl CodexClient {
         self.send_request(request, request_id, "account/rateLimits/read")
     }
 
-    fn stream_conversation(&mut self, conversation_id: &ConversationId) -> Result<()> {
+    fn model_list(&mut self, params: ModelListParams) -> Result<ModelListResponse> {
+        let request_id = self.request_id();
+        let request = ClientRequest::ModelList {
+            request_id: request_id.clone(),
+            params,
+        };
+
+        self.send_request(request, request_id, "model/list")
+    }
+
+    fn stream_conversation(&mut self, conversation_id: &ThreadId) -> Result<()> {
         loop {
             let notification = self.next_notification()?;
 
@@ -469,7 +545,7 @@ impl CodexClient {
                         print!("{}", event.delta);
                         std::io::stdout().flush().ok();
                     }
-                    EventMsg::TaskComplete(event) => {
+                    EventMsg::TurnComplete(event) => {
                         println!("\n[task complete: {event:?}]");
                         break;
                     }
@@ -589,7 +665,7 @@ impl CodexClient {
     fn extract_event(
         &self,
         notification: JSONRPCNotification,
-        conversation_id: &ConversationId,
+        conversation_id: &ThreadId,
     ) -> Result<Option<Event>> {
         let params = notification
             .params
@@ -603,7 +679,7 @@ impl CodexClient {
         let conversation_value = map
             .remove("conversationId")
             .context("event missing conversationId")?;
-        let notification_conversation: ConversationId = serde_json::from_value(conversation_value)
+        let notification_conversation: ThreadId = serde_json::from_value(conversation_value)
             .context("conversationId was not a valid UUID")?;
 
         if &notification_conversation != conversation_id {
@@ -770,7 +846,7 @@ impl CodexClient {
         }
 
         let response = CommandExecutionRequestApprovalResponse {
-            decision: ApprovalDecision::Accept,
+            decision: CommandExecutionApprovalDecision::Accept,
         };
         self.send_server_request_response(request_id, &response)?;
         println!("< approved commandExecution request for item {item_id}");
@@ -801,7 +877,7 @@ impl CodexClient {
         }
 
         let response = FileChangeRequestApprovalResponse {
-            decision: ApprovalDecision::Accept,
+            decision: FileChangeApprovalDecision::Accept,
         };
         self.send_server_request_response(request_id, &response)?;
         println!("< approved fileChange request for item {item_id}");

codex-rs/app-server/BUILD.bazel

@@ -0,0 +1,8 @@
+load("//:defs.bzl", "codex_rust_crate")
+
+codex_rust_crate(
+    name = "app-server",
+    crate_name = "codex_app_server",
+    integration_deps_extra = ["//codex-rs/app-server/tests/common:common"],
+    test_tags = ["no-sandbox"],
+)

codex-rs/app-server/Cargo.toml

@@ -48,7 +48,6 @@ 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 }
 mcp-types = { workspace = true }

codex-rs/app-server/README.md

@@ -11,6 +11,8 @@
 - [Initialization](#initialization)
 - [API Overview](#api-overview)
 - [Events](#events)
+- [Approvals](#approvals)
+- [Skills](#skills)
 - [Auth endpoints](#auth-endpoints)
 
 ## Protocol
@@ -39,7 +41,7 @@ Use the thread APIs to create, list, or archive conversations. Drive a conversat
 ## Lifecycle Overview
 
 - Initialize once: Immediately after launching the codex app-server process, send an `initialize` request with your client metadata, then emit an `initialized` notification. Any other request before this handshake gets rejected.
-- Start (or resume) a thread: Call `thread/start` to open a fresh conversation. The response returns the thread object and you’ll also get a `thread/started` notification. If you’re continuing an existing conversation, call `thread/resume` with its ID instead.
+- Start (or resume) a thread: Call `thread/start` to open a fresh conversation. The response returns the thread object and you’ll also get a `thread/started` notification. If you’re continuing an existing conversation, call `thread/resume` with its ID instead. If you want to branch from an existing conversation, call `thread/fork` to create a new thread id with copied history.
 - Begin a turn: To send user input, call `turn/start` with the target `threadId` and the user's input. Optional fields let you override model, cwd, sandbox policy, etc. This immediately returns the new turn object and triggers a `turn/started` notification.
 - Stream events: After `turn/start`, keep reading JSON-RPC notifications on stdout. You’ll see `item/started`, `item/completed`, deltas like `item/agentMessage/delta`, tool progress, etc. These represent streaming model output plus any side effects (commands, tool calls, reasoning notes).
 - Finish the turn: When the model is done (or the turn is interrupted via making the `turn/interrupt` call), the server sends `turn/completed` with the final turn state and token usage.
@@ -50,6 +52,10 @@ Clients must send a single `initialize` request before invoking any other method
 
 Applications building on top of `codex app-server` should identify themselves via the `clientInfo` parameter.
 
+**Important**: `clientInfo.name` is used to identify the client for the OpenAI Compliance Logs Platform. If
+you are developing a new Codex integration that is intended for enterprise use, please contact us to get it
+added to a known clients list. For more context: https://chatgpt.com/admin/api-reference#tag/Logs:-Codex
+
 Example (from OpenAI's official VSCode extension):
 
 ```json
@@ -58,7 +64,7 @@ Example (from OpenAI's official VSCode extension):
   "id": 0,
   "params": {
     "clientInfo": {
-      "name": "codex-vscode",
+      "name": "codex_vscode",
       "title": "Codex VS Code Extension",
       "version": "0.1.0"
     }
@@ -70,8 +76,11 @@ Example (from OpenAI's official VSCode extension):
 
 - `thread/start` — create a new thread; emits `thread/started` and auto-subscribes you to turn/item events for that thread.
 - `thread/resume` — reopen an existing thread by id so subsequent `turn/start` calls append to it.
+- `thread/fork` — fork an existing thread into a new thread id by copying the stored history; emits `thread/started` and auto-subscribes you to turn/item events for the new thread.
 - `thread/list` — page through stored rollouts; supports cursor-based pagination and optional `modelProviders` filtering.
+- `thread/loaded/list` — list the thread ids currently loaded in memory.
 - `thread/archive` — move a thread’s rollout file into the archived directory; returns `{}` on success.
+- `thread/rollback` — drop the last N turns from the agent’s in-memory context and persist a rollback marker in the rollout so future resumes see the pruned history; returns the updated `thread` (with `turns` populated) on success.
 - `turn/start` — add user input to a thread and begin Codex generation; responds with the initial `turn` object and streams `turn/started`, `item/*`, and `turn/completed` notifications.
 - `turn/interrupt` — request cancellation of an in-flight turn by `(thread_id, turn_id)`; success is an empty `{}` response and the turn finishes with `status: "interrupted"`.
 - `review/start` — kick off Codex’s automated reviewer for a thread; responds like `turn/start` and emits `item/started`/`item/completed` notifications with `enteredReviewMode` and `exitedReviewMode` items, plus a final assistant `agentMessage` containing the review.
@@ -85,6 +94,7 @@ Example (from OpenAI's official VSCode extension):
 - `config/read` — fetch the effective config on disk after resolving config layering.
 - `config/value/write` — write a single config key/value to the user's config.toml on disk.
 - `config/batchWrite` — apply multiple config edits atomically to the user's config.toml on disk.
+- `configRequirements/read` — fetch the loaded requirements allow-lists from `requirements.toml` and/or MDM (or `null` if none are configured).
 
 ### Example: Start or resume a thread
 
@@ -117,6 +127,14 @@ To continue a stored session, call `thread/resume` with the `thread.id` you prev
 { "id": 11, "result": { "thread": { "id": "thr_123", … } } }
 ```
 
+To branch from a stored session, call `thread/fork` with the `thread.id`. This creates a new thread id and emits a `thread/started` notification for it:
+
+```json
+{ "method": "thread/fork", "id": 12, "params": { "threadId": "thr_123" } }
+{ "id": 12, "result": { "thread": { "id": "thr_456", … } } }
+{ "method": "thread/started", "params": { "thread": { … } } }
+```
+
 ### Example: List threads (with pagination & filters)
 
 `thread/list` lets you render a history UI. Pass any combination of:
@@ -143,6 +161,17 @@ Example:
 
 When `nextCursor` is `null`, you’ve reached the final page.
 
+### Example: List loaded threads
+
+`thread/loaded/list` returns thread ids currently loaded in memory. This is useful when you want to check which sessions are active without scanning rollouts on disk.
+
+```json
+{ "method": "thread/loaded/list", "id": 21 }
+{ "id": 21, "result": {
+    "data": ["thr_123", "thr_456"]
+} }
+```
+
 ### Example: Archive a thread
 
 Use `thread/archive` to move the persisted rollout (stored as a JSONL file on disk) into the archived sessions directory.
@@ -162,7 +191,7 @@ Turns attach user input (text or images) to a thread and trigger Codex generatio
 - `{"type":"image","url":"https://…png"}`
 - `{"type":"localImage","path":"/tmp/screenshot.png"}`
 
-You can optionally specify config overrides on the new turn. If specified, these settings become the default for subsequent turns on the same thread.
+You can optionally specify config overrides on the new turn. If specified, these settings become the default for subsequent turns on the same thread. `outputSchema` applies only to the current turn.
 
 ```json
 { "method": "turn/start", "id": 30, "params": {
@@ -172,13 +201,20 @@ You can optionally specify config overrides on the new turn. If specified, these
     "cwd": "/Users/me/project",
     "approvalPolicy": "unlessTrusted",
     "sandboxPolicy": {
-        "mode": "workspaceWrite",
+        "type": "workspaceWrite",
         "writableRoots": ["/Users/me/project"],
         "networkAccess": true
     },
     "model": "gpt-5.1-codex",
     "effort": "medium",
-    "summary": "concise"
+    "summary": "concise",
+    // Optional JSON Schema to constrain the final assistant message for this turn.
+    "outputSchema": {
+        "type": "object",
+        "properties": { "answer": { "type": "string" } },
+        "required": ["answer"],
+        "additionalProperties": false
+    }
 } }
 { "id": 30, "result": { "turn": {
     "id": "turn_456",
@@ -188,6 +224,26 @@ You can optionally specify config overrides on the new turn. If specified, these
 } } }
 ```
 
+### Example: Start a turn (invoke a skill)
+
+Invoke a skill explicitly by including `$<skill-name>` in the text input and adding a `skill` input item alongside it.
+
+```json
+{ "method": "turn/start", "id": 33, "params": {
+    "threadId": "thr_123",
+    "input": [
+        { "type": "text", "text": "$skill-creator Add a new skill for triaging flaky CI and include step-by-step usage." },
+        { "type": "skill", "name": "skill-creator", "path": "/Users/me/.codex/skills/skill-creator/SKILL.md" }
+    ]
+} }
+{ "id": 33, "result": { "turn": {
+    "id": "turn_457",
+    "status": "inProgress",
+    "items": [],
+    "error": null
+} } }
+```
+
 ### Example: Interrupt an active turn
 
 You can cancel a running Turn with `turn/interrupt`.
@@ -285,10 +341,12 @@ Run a standalone command (argv vector) in the server’s sandbox without creatin
 { "id": 32, "result": { "exitCode": 0, "stdout": "...", "stderr": "" } }
 ```
 
+- For clients that are already sandboxed externally, set `sandboxPolicy` to `{"type":"externalSandbox","networkAccess":"enabled"}` (or omit `networkAccess` to keep it restricted). Codex will not enforce its own sandbox in this mode; it tells the model it has full file-system access and passes the `networkAccess` state through `environment_context`.
+
 Notes:
 
 - Empty `command` arrays are rejected.
-- `sandboxPolicy` accepts the same shape used by `turn/start` (e.g., `dangerFullAccess`, `readOnly`, `workspaceWrite` with flags).
+- `sandboxPolicy` accepts the same shape used by `turn/start` (e.g., `dangerFullAccess`, `readOnly`, `workspaceWrite` with flags, `externalSandbox` with `networkAccess` `restricted|enabled`).
 - When omitted, `timeoutMs` falls back to the server default.
 
 ## Events
@@ -300,7 +358,7 @@ Event notifications are the server-initiated event stream for thread lifecycles,
 The app-server streams JSON-RPC notifications while a turn is running. Each turn starts with `turn/started` (initial `turn`) and ends with `turn/completed` (final `turn` status). Token usage events stream separately via `thread/tokenUsage/updated`. Clients subscribe to the events they care about, rendering each item incrementally as updates arrive. The per-item lifecycle is always: `item/started` → zero or more item-specific deltas → `item/completed`.
 
 - `turn/started` — `{ turn }` with the turn id, empty `items`, and `status: "inProgress"`.
-- `turn/completed` — `{ turn }` where `turn.status` is `completed`, `interrupted`, or `failed`; failures carry `{ error: { message, codexErrorInfo? } }`.
+- `turn/completed` — `{ turn }` where `turn.status` is `completed`, `interrupted`, or `failed`; failures carry `{ error: { message, codexErrorInfo?, additionalDetails? } }`.
 - `turn/diff/updated` — `{ threadId, turnId, diff }` represents the up-to-date snapshot of the turn-level unified diff, emitted after every FileChange item. `diff` is the latest aggregated unified diff across every file change in the turn. UIs can render this to show the full "what changed" view without stitching individual `fileChange` items.
 - `turn/plan/updated` — `{ turnId, explanation?, plan }` whenever the agent shares or changes its plan; each `plan` entry is `{ step, status }` with `status` in `pending`, `inProgress`, or `completed`.
 
@@ -350,7 +408,7 @@ There are additional item-specific events:
 
 ### Errors
 
-`error` event is emitted whenever the server hits an error mid-turn (for example, upstream model errors or quota limits). Carries the same `{ error: { message, codexErrorInfo? } }` payload as `turn.status: "failed"` and may precede that terminal notification.
+`error` event is emitted whenever the server hits an error mid-turn (for example, upstream model errors or quota limits). Carries the same `{ error: { message, codexErrorInfo?, additionalDetails? } }` payload as `turn.status: "failed"` and may precede that terminal notification.
 
 `codexErrorInfo` maps to the `CodexErrorInfo` enum. Common values:
 
@@ -395,6 +453,46 @@ Order of messages:
 
 UI guidance for IDEs: surface an approval dialog as soon as the request arrives. The turn will proceed after the server receives a response to the approval request. The terminal `item/completed` notification will be sent with the appropriate status.
 
+## Skills
+
+Invoke a skill by including `$<skill-name>` in the text input. Add a `skill` input item (recommended) so the backend injects full skill instructions instead of relying on the model to resolve the name.
+
+```json
+{
+  "method": "turn/start",
+  "id": 101,
+  "params": {
+    "threadId": "thread-1",
+    "input": [
+      { "type": "text", "text": "$skill-creator Add a new skill for triaging flaky CI." },
+      { "type": "skill", "name": "skill-creator", "path": "/Users/me/.codex/skills/skill-creator/SKILL.md" }
+    ]
+  }
+}
+```
+
+If you omit the `skill` item, the model will still parse the `$<skill-name>` marker and try to locate the skill, which can add latency.
+
+Example:
+
+```
+$skill-creator Add a new skill for triaging flaky CI and include step-by-step usage.
+```
+
+Use `skills/list` to fetch the available skills (optionally scoped by `cwd` and/or with `forceReload`).
+
+```json
+{ "method": "skills/list", "id": 25, "params": {
+    "cwd": "/Users/me/project",
+    "forceReload": false
+} }
+{ "id": 25, "result": {
+    "skills": [
+        { "name": "skill-creator", "description": "Create or update a Codex skill" }
+    ]
+} }
+```
+
 ## Auth endpoints
 
 The JSON-RPC auth/account surface exposes request/response methods plus server-initiated notifications (no `id`). Use these to determine auth state, start or cancel logins, logout, and inspect ChatGPT rate limits.

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

@@ -1,24 +1,32 @@
 use crate::codex_message_processor::ApiVersion;
 use crate::codex_message_processor::PendingInterrupts;
+use crate::codex_message_processor::PendingRollbacks;
 use crate::codex_message_processor::TurnSummary;
 use crate::codex_message_processor::TurnSummaryStore;
+use crate::codex_message_processor::read_event_msgs_from_rollout;
+use crate::codex_message_processor::read_summary_from_rollout;
+use crate::codex_message_processor::summary_to_thread;
+use crate::error_code::INTERNAL_ERROR_CODE;
+use crate::error_code::INVALID_REQUEST_ERROR_CODE;
 use crate::outgoing_message::OutgoingMessageSender;
 use codex_app_server_protocol::AccountRateLimitsUpdatedNotification;
 use codex_app_server_protocol::AgentMessageDeltaNotification;
 use codex_app_server_protocol::ApplyPatchApprovalParams;
 use codex_app_server_protocol::ApplyPatchApprovalResponse;
-use codex_app_server_protocol::ApprovalDecision;
 use codex_app_server_protocol::CodexErrorInfo as V2CodexErrorInfo;
 use codex_app_server_protocol::CommandAction as V2ParsedCommand;
+use codex_app_server_protocol::CommandExecutionApprovalDecision;
 use codex_app_server_protocol::CommandExecutionOutputDeltaNotification;
 use codex_app_server_protocol::CommandExecutionRequestApprovalParams;
 use codex_app_server_protocol::CommandExecutionRequestApprovalResponse;
 use codex_app_server_protocol::CommandExecutionStatus;
 use codex_app_server_protocol::ContextCompactedNotification;
+use codex_app_server_protocol::DeprecationNoticeNotification;
 use codex_app_server_protocol::ErrorNotification;
 use codex_app_server_protocol::ExecCommandApprovalParams;
 use codex_app_server_protocol::ExecCommandApprovalResponse;
 use codex_app_server_protocol::ExecPolicyAmendment as V2ExecPolicyAmendment;
+use codex_app_server_protocol::FileChangeApprovalDecision;
 use codex_app_server_protocol::FileChangeOutputDeltaNotification;
 use codex_app_server_protocol::FileChangeRequestApprovalParams;
 use codex_app_server_protocol::FileChangeRequestApprovalResponse;
@@ -26,6 +34,7 @@ use codex_app_server_protocol::FileUpdateChange;
 use codex_app_server_protocol::InterruptConversationResponse;
 use codex_app_server_protocol::ItemCompletedNotification;
 use codex_app_server_protocol::ItemStartedNotification;
+use codex_app_server_protocol::JSONRPCErrorError;
 use codex_app_server_protocol::McpToolCallError;
 use codex_app_server_protocol::McpToolCallResult;
 use codex_app_server_protocol::McpToolCallStatus;
@@ -39,6 +48,7 @@ use codex_app_server_protocol::ServerNotification;
 use codex_app_server_protocol::ServerRequestPayload;
 use codex_app_server_protocol::TerminalInteractionNotification;
 use codex_app_server_protocol::ThreadItem;
+use codex_app_server_protocol::ThreadRollbackResponse;
 use codex_app_server_protocol::ThreadTokenUsage;
 use codex_app_server_protocol::ThreadTokenUsageUpdatedNotification;
 use codex_app_server_protocol::Turn;
@@ -49,9 +59,11 @@ use codex_app_server_protocol::TurnInterruptResponse;
 use codex_app_server_protocol::TurnPlanStep;
 use codex_app_server_protocol::TurnPlanUpdatedNotification;
 use codex_app_server_protocol::TurnStatus;
-use codex_core::CodexConversation;
+use codex_app_server_protocol::build_turns_from_event_msgs;
+use codex_core::CodexThread;
 use codex_core::parse_command::shlex_join;
 use codex_core::protocol::ApplyPatchApprovalRequestEvent;
+use codex_core::protocol::CodexErrorInfo as CoreCodexErrorInfo;
 use codex_core::protocol::Event;
 use codex_core::protocol::EventMsg;
 use codex_core::protocol::ExecApprovalRequestEvent;
@@ -65,7 +77,7 @@ use codex_core::protocol::TokenCountEvent;
 use codex_core::protocol::TurnDiffEvent;
 use codex_core::review_format::format_review_findings_block;
 use codex_core::review_prompts;
-use codex_protocol::ConversationId;
+use codex_protocol::ThreadId;
 use codex_protocol::plan_tool::UpdatePlanArgs;
 use codex_protocol::protocol::ReviewOutputEvent;
 use std::collections::HashMap;
@@ -77,21 +89,24 @@ use tracing::error;
 
 type JsonValue = serde_json::Value;
 
+#[allow(clippy::too_many_arguments)]
 pub(crate) async fn apply_bespoke_event_handling(
     event: Event,
-    conversation_id: ConversationId,
-    conversation: Arc<CodexConversation>,
+    conversation_id: ThreadId,
+    conversation: Arc<CodexThread>,
     outgoing: Arc<OutgoingMessageSender>,
     pending_interrupts: PendingInterrupts,
+    pending_rollbacks: PendingRollbacks,
     turn_summary_store: TurnSummaryStore,
     api_version: ApiVersion,
+    fallback_model_provider: String,
 ) {
     let Event {
         id: event_turn_id,
         msg,
     } = event;
     match msg {
-        EventMsg::TaskComplete(_ev) => {
+        EventMsg::TurnComplete(_ev) => {
             handle_turn_complete(
                 conversation_id,
                 event_turn_id,
@@ -283,6 +298,15 @@ pub(crate) async fn apply_bespoke_event_handling(
                 .send_server_notification(ServerNotification::ContextCompacted(notification))
                 .await;
         }
+        EventMsg::DeprecationNotice(event) => {
+            let notification = DeprecationNoticeNotification {
+                summary: event.summary,
+                details: event.details,
+            };
+            outgoing
+                .send_server_notification(ServerNotification::DeprecationNotice(notification))
+                .await;
+        }
         EventMsg::ReasoningContentDelta(event) => {
             let notification = ReasoningSummaryTextDeltaNotification {
                 thread_id: conversation_id.to_string(),
@@ -327,14 +351,35 @@ pub(crate) async fn apply_bespoke_event_handling(
                 .await;
         }
         EventMsg::Error(ev) => {
+            let message = ev.message.clone();
+            let codex_error_info = ev.codex_error_info.clone();
+
+            // If this error belongs to an in-flight `thread/rollback` request, fail that request
+            // (and clear pending state) so subsequent rollbacks are unblocked.
+            //
+            // Don't send a notification for this error.
+            if matches!(
+                codex_error_info,
+                Some(CoreCodexErrorInfo::ThreadRollbackFailed)
+            ) {
+                return handle_thread_rollback_failed(
+                    conversation_id,
+                    message,
+                    &pending_rollbacks,
+                    &outgoing,
+                )
+                .await;
+            };
+
             let turn_error = TurnError {
                 message: ev.message,
                 codex_error_info: ev.codex_error_info.map(V2CodexErrorInfo::from),
+                additional_details: None,
             };
             handle_error(conversation_id, turn_error.clone(), &turn_summary_store).await;
             outgoing
                 .send_server_notification(ServerNotification::Error(ErrorNotification {
-                    error: turn_error,
+                    error: turn_error.clone(),
                     will_retry: false,
                     thread_id: conversation_id.to_string(),
                     turn_id: event_turn_id.clone(),
@@ -347,6 +392,7 @@ pub(crate) async fn apply_bespoke_event_handling(
             let turn_error = TurnError {
                 message: ev.message,
                 codex_error_info: ev.codex_error_info.map(V2CodexErrorInfo::from),
+                additional_details: ev.additional_details,
             };
             outgoing
                 .send_server_notification(ServerNotification::Error(ErrorNotification {
@@ -678,6 +724,58 @@ pub(crate) async fn apply_bespoke_event_handling(
             )
             .await;
         }
+        EventMsg::ThreadRolledBack(_rollback_event) => {
+            let pending = {
+                let mut map = pending_rollbacks.lock().await;
+                map.remove(&conversation_id)
+            };
+
+            if let Some(request_id) = pending {
+                let rollout_path = conversation.rollout_path();
+                let response = match read_summary_from_rollout(
+                    rollout_path.as_path(),
+                    fallback_model_provider.as_str(),
+                )
+                .await
+                {
+                    Ok(summary) => {
+                        let mut thread = summary_to_thread(summary);
+                        match read_event_msgs_from_rollout(rollout_path.as_path()).await {
+                            Ok(events) => {
+                                thread.turns = build_turns_from_event_msgs(&events);
+                                ThreadRollbackResponse { thread }
+                            }
+                            Err(err) => {
+                                let error = JSONRPCErrorError {
+                                    code: INTERNAL_ERROR_CODE,
+                                    message: format!(
+                                        "failed to load rollout `{}`: {err}",
+                                        rollout_path.display()
+                                    ),
+                                    data: None,
+                                };
+                                outgoing.send_error(request_id, error).await;
+                                return;
+                            }
+                        }
+                    }
+                    Err(err) => {
+                        let error = JSONRPCErrorError {
+                            code: INTERNAL_ERROR_CODE,
+                            message: format!(
+                                "failed to load rollout `{}`: {err}",
+                                rollout_path.display()
+                            ),
+                            data: None,
+                        };
+                        outgoing.send_error(request_id, error).await;
+                        return;
+                    }
+                };
+
+                outgoing.send_response(request_id, response).await;
+            }
+        }
         EventMsg::TurnDiff(turn_diff_event) => {
             handle_turn_diff(
                 conversation_id,
@@ -704,7 +802,7 @@ pub(crate) async fn apply_bespoke_event_handling(
 }
 
 async fn handle_turn_diff(
-    conversation_id: ConversationId,
+    conversation_id: ThreadId,
     event_turn_id: &str,
     turn_diff_event: TurnDiffEvent,
     api_version: ApiVersion,
@@ -723,7 +821,7 @@ async fn handle_turn_diff(
 }
 
 async fn handle_turn_plan_update(
-    conversation_id: ConversationId,
+    conversation_id: ThreadId,
     event_turn_id: &str,
     plan_update_event: UpdatePlanArgs,
     api_version: ApiVersion,
@@ -747,7 +845,7 @@ async fn handle_turn_plan_update(
 }
 
 async fn emit_turn_completed_with_status(
-    conversation_id: ConversationId,
+    conversation_id: ThreadId,
     event_turn_id: String,
     status: TurnStatus,
     error: Option<TurnError>,
@@ -768,7 +866,7 @@ async fn emit_turn_completed_with_status(
 }
 
 async fn complete_file_change_item(
-    conversation_id: ConversationId,
+    conversation_id: ThreadId,
     item_id: String,
     changes: Vec<FileUpdateChange>,
     status: PatchApplyStatus,
@@ -800,7 +898,7 @@ async fn complete_file_change_item(
 
 #[allow(clippy::too_many_arguments)]
 async fn complete_command_execution_item(
-    conversation_id: ConversationId,
+    conversation_id: ThreadId,
     turn_id: String,
     item_id: String,
     command: String,
@@ -833,7 +931,7 @@ async fn complete_command_execution_item(
 
 async fn maybe_emit_raw_response_item_completed(
     api_version: ApiVersion,
-    conversation_id: ConversationId,
+    conversation_id: ThreadId,
     turn_id: &str,
     item: codex_protocol::models::ResponseItem,
     outgoing: &OutgoingMessageSender,
@@ -853,7 +951,7 @@ async fn maybe_emit_raw_response_item_completed(
 }
 
 async fn find_and_remove_turn_summary(
-    conversation_id: ConversationId,
+    conversation_id: ThreadId,
     turn_summary_store: &TurnSummaryStore,
 ) -> TurnSummary {
     let mut map = turn_summary_store.lock().await;
@@ -861,7 +959,7 @@ async fn find_and_remove_turn_summary(
 }
 
 async fn handle_turn_complete(
-    conversation_id: ConversationId,
+    conversation_id: ThreadId,
     event_turn_id: String,
     outgoing: &OutgoingMessageSender,
     turn_summary_store: &TurnSummaryStore,
@@ -877,7 +975,7 @@ async fn handle_turn_complete(
 }
 
 async fn handle_turn_interrupted(
-    conversation_id: ConversationId,
+    conversation_id: ThreadId,
     event_turn_id: String,
     outgoing: &OutgoingMessageSender,
     turn_summary_store: &TurnSummaryStore,
@@ -894,8 +992,33 @@ async fn handle_turn_interrupted(
     .await;
 }
 
+async fn handle_thread_rollback_failed(
+    conversation_id: ThreadId,
+    message: String,
+    pending_rollbacks: &PendingRollbacks,
+    outgoing: &OutgoingMessageSender,
+) {
+    let pending_rollback = {
+        let mut map = pending_rollbacks.lock().await;
+        map.remove(&conversation_id)
+    };
+
+    if let Some(request_id) = pending_rollback {
+        outgoing
+            .send_error(
+                request_id,
+                JSONRPCErrorError {
+                    code: INVALID_REQUEST_ERROR_CODE,
+                    message: message.clone(),
+                    data: None,
+                },
+            )
+            .await;
+    }
+}
+
 async fn handle_token_count_event(
-    conversation_id: ConversationId,
+    conversation_id: ThreadId,
     turn_id: String,
     token_count_event: TokenCountEvent,
     outgoing: &OutgoingMessageSender,
@@ -923,7 +1046,7 @@ async fn handle_token_count_event(
 }
 
 async fn handle_error(
-    conversation_id: ConversationId,
+    conversation_id: ThreadId,
     error: TurnError,
     turn_summary_store: &TurnSummaryStore,
 ) {
@@ -934,7 +1057,7 @@ async fn handle_error(
 async fn on_patch_approval_response(
     event_turn_id: String,
     receiver: oneshot::Receiver<JsonValue>,
-    codex: Arc<CodexConversation>,
+    codex: Arc<CodexThread>,
 ) {
     let response = receiver.await;
     let value = match response {
@@ -976,7 +1099,7 @@ async fn on_patch_approval_response(
 async fn on_exec_approval_response(
     event_turn_id: String,
     receiver: oneshot::Receiver<JsonValue>,
-    conversation: Arc<CodexConversation>,
+    conversation: Arc<CodexThread>,
 ) {
     let response = receiver.await;
     let value = match response {
@@ -1071,14 +1194,29 @@ fn format_file_change_diff(change: &CoreFileChange) -> String {
     }
 }
 
+fn map_file_change_approval_decision(
+    decision: FileChangeApprovalDecision,
+) -> (ReviewDecision, Option<PatchApplyStatus>) {
+    match decision {
+        FileChangeApprovalDecision::Accept => (ReviewDecision::Approved, None),
+        FileChangeApprovalDecision::AcceptForSession => (ReviewDecision::ApprovedForSession, None),
+        FileChangeApprovalDecision::Decline => {
+            (ReviewDecision::Denied, Some(PatchApplyStatus::Declined))
+        }
+        FileChangeApprovalDecision::Cancel => {
+            (ReviewDecision::Abort, Some(PatchApplyStatus::Declined))
+        }
+    }
+}
+
 #[allow(clippy::too_many_arguments)]
 async fn on_file_change_request_approval_response(
     event_turn_id: String,
-    conversation_id: ConversationId,
+    conversation_id: ThreadId,
     item_id: String,
     changes: Vec<FileUpdateChange>,
     receiver: oneshot::Receiver<JsonValue>,
-    codex: Arc<CodexConversation>,
+    codex: Arc<CodexThread>,
     outgoing: Arc<OutgoingMessageSender>,
     turn_summary_store: TurnSummaryStore,
 ) {
@@ -1089,23 +1227,12 @@ async fn on_file_change_request_approval_response(
                 .unwrap_or_else(|err| {
                     error!("failed to deserialize FileChangeRequestApprovalResponse: {err}");
                     FileChangeRequestApprovalResponse {
-                        decision: ApprovalDecision::Decline,
+                        decision: FileChangeApprovalDecision::Decline,
                     }
                 });
 
-            let (decision, completion_status) = match response.decision {
-                ApprovalDecision::Accept
-                | ApprovalDecision::AcceptForSession
-                | ApprovalDecision::AcceptWithExecpolicyAmendment { .. } => {
-                    (ReviewDecision::Approved, None)
-                }
-                ApprovalDecision::Decline => {
-                    (ReviewDecision::Denied, Some(PatchApplyStatus::Declined))
-                }
-                ApprovalDecision::Cancel => {
-                    (ReviewDecision::Abort, Some(PatchApplyStatus::Declined))
-                }
-            };
+            let (decision, completion_status) =
+                map_file_change_approval_decision(response.decision);
             // Allow EventMsg::PatchApplyEnd to emit ItemCompleted for accepted patches.
             // Only short-circuit on declines/cancels/failures.
             (decision, completion_status)
@@ -1143,13 +1270,13 @@ async fn on_file_change_request_approval_response(
 #[allow(clippy::too_many_arguments)]
 async fn on_command_execution_request_approval_response(
     event_turn_id: String,
-    conversation_id: ConversationId,
+    conversation_id: ThreadId,
     item_id: String,
     command: String,
     cwd: PathBuf,
     command_actions: Vec<V2ParsedCommand>,
     receiver: oneshot::Receiver<JsonValue>,
-    conversation: Arc<CodexConversation>,
+    conversation: Arc<CodexThread>,
     outgoing: Arc<OutgoingMessageSender>,
 ) {
     let response = receiver.await;
@@ -1159,16 +1286,18 @@ async fn on_command_execution_request_approval_response(
                 .unwrap_or_else(|err| {
                     error!("failed to deserialize CommandExecutionRequestApprovalResponse: {err}");
                     CommandExecutionRequestApprovalResponse {
-                        decision: ApprovalDecision::Decline,
+                        decision: CommandExecutionApprovalDecision::Decline,
                     }
                 });
 
             let decision = response.decision;
 
             let (decision, completion_status) = match decision {
-                ApprovalDecision::Accept => (ReviewDecision::Approved, None),
-                ApprovalDecision::AcceptForSession => (ReviewDecision::ApprovedForSession, None),
-                ApprovalDecision::AcceptWithExecpolicyAmendment {
+                CommandExecutionApprovalDecision::Accept => (ReviewDecision::Approved, None),
+                CommandExecutionApprovalDecision::AcceptForSession => {
+                    (ReviewDecision::ApprovedForSession, None)
+                }
+                CommandExecutionApprovalDecision::AcceptWithExecpolicyAmendment {
                     execpolicy_amendment,
                 } => (
                     ReviewDecision::ApprovedExecpolicyAmendment {
@@ -1176,11 +1305,11 @@ async fn on_command_execution_request_approval_response(
                     },
                     None,
                 ),
-                ApprovalDecision::Decline => (
+                CommandExecutionApprovalDecision::Decline => (
                     ReviewDecision::Denied,
                     Some(CommandExecutionStatus::Declined),
                 ),
-                ApprovalDecision::Cancel => (
+                CommandExecutionApprovalDecision::Cancel => (
                     ReviewDecision::Abort,
                     Some(CommandExecutionStatus::Declined),
                 ),
@@ -1320,9 +1449,17 @@ mod tests {
         Arc::new(Mutex::new(HashMap::new()))
     }
 
+    #[test]
+    fn file_change_accept_for_session_maps_to_approved_for_session() {
+        let (decision, completion_status) =
+            map_file_change_approval_decision(FileChangeApprovalDecision::AcceptForSession);
+        assert_eq!(decision, ReviewDecision::ApprovedForSession);
+        assert_eq!(completion_status, None);
+    }
+
     #[tokio::test]
     async fn test_handle_error_records_message() -> Result<()> {
-        let conversation_id = ConversationId::new();
+        let conversation_id = ThreadId::new();
         let turn_summary_store = new_turn_summary_store();
 
         handle_error(
@@ -1330,6 +1467,7 @@ mod tests {
             TurnError {
                 message: "boom".to_string(),
                 codex_error_info: Some(V2CodexErrorInfo::InternalServerError),
+                additional_details: None,
             },
             &turn_summary_store,
         )
@@ -1341,6 +1479,7 @@ mod tests {
             Some(TurnError {
                 message: "boom".to_string(),
                 codex_error_info: Some(V2CodexErrorInfo::InternalServerError),
+                additional_details: None,
             })
         );
         Ok(())
@@ -1348,7 +1487,7 @@ mod tests {
 
     #[tokio::test]
     async fn test_handle_turn_complete_emits_completed_without_error() -> Result<()> {
-        let conversation_id = ConversationId::new();
+        let conversation_id = ThreadId::new();
         let event_turn_id = "complete1".to_string();
         let (tx, mut rx) = mpsc::channel(CHANNEL_CAPACITY);
         let outgoing = Arc::new(OutgoingMessageSender::new(tx));
@@ -1380,7 +1519,7 @@ mod tests {
 
     #[tokio::test]
     async fn test_handle_turn_interrupted_emits_interrupted_with_error() -> Result<()> {
-        let conversation_id = ConversationId::new();
+        let conversation_id = ThreadId::new();
         let event_turn_id = "interrupt1".to_string();
         let turn_summary_store = new_turn_summary_store();
         handle_error(
@@ -1388,6 +1527,7 @@ mod tests {
             TurnError {
                 message: "oops".to_string(),
                 codex_error_info: None,
+                additional_details: None,
             },
             &turn_summary_store,
         )
@@ -1421,7 +1561,7 @@ mod tests {
 
     #[tokio::test]
     async fn test_handle_turn_complete_emits_failed_with_error() -> Result<()> {
-        let conversation_id = ConversationId::new();
+        let conversation_id = ThreadId::new();
         let event_turn_id = "complete_err1".to_string();
         let turn_summary_store = new_turn_summary_store();
         handle_error(
@@ -1429,6 +1569,7 @@ mod tests {
             TurnError {
                 message: "bad".to_string(),
                 codex_error_info: Some(V2CodexErrorInfo::Other),
+                additional_details: None,
             },
             &turn_summary_store,
         )
@@ -1457,6 +1598,7 @@ mod tests {
                     Some(TurnError {
                         message: "bad".to_string(),
                         codex_error_info: Some(V2CodexErrorInfo::Other),
+                        additional_details: None,
                     })
                 );
             }
@@ -1484,7 +1626,7 @@ mod tests {
             ],
         };
 
-        let conversation_id = ConversationId::new();
+        let conversation_id = ThreadId::new();
 
         handle_turn_plan_update(
             conversation_id,
@@ -1518,7 +1660,7 @@ mod tests {
 
     #[tokio::test]
     async fn test_handle_token_count_event_emits_usage_and_rate_limits() -> Result<()> {
-        let conversation_id = ConversationId::new();
+        let conversation_id = ThreadId::new();
         let turn_id = "turn-123".to_string();
         let (tx, mut rx) = mpsc::channel(CHANNEL_CAPACITY);
         let outgoing = Arc::new(OutgoingMessageSender::new(tx));
@@ -1603,7 +1745,7 @@ mod tests {
 
     #[tokio::test]
     async fn test_handle_token_count_event_without_usage_info() -> Result<()> {
-        let conversation_id = ConversationId::new();
+        let conversation_id = ThreadId::new();
         let turn_id = "turn-456".to_string();
         let (tx, mut rx) = mpsc::channel(CHANNEL_CAPACITY);
         let outgoing = Arc::new(OutgoingMessageSender::new(tx));
@@ -1637,7 +1779,7 @@ mod tests {
             },
         };
 
-        let thread_id = ConversationId::new().to_string();
+        let thread_id = ThreadId::new().to_string();
         let turn_id = "turn_1".to_string();
         let notification = construct_mcp_tool_call_notification(
             begin_event.clone(),
@@ -1667,8 +1809,8 @@ mod tests {
     #[tokio::test]
     async fn test_handle_turn_complete_emits_error_multiple_turns() -> Result<()> {
         // Conversation A will have two turns; Conversation B will have one turn.
-        let conversation_a = ConversationId::new();
-        let conversation_b = ConversationId::new();
+        let conversation_a = ThreadId::new();
+        let conversation_b = ThreadId::new();
         let turn_summary_store = new_turn_summary_store();
 
         let (tx, mut rx) = mpsc::channel(CHANNEL_CAPACITY);
@@ -1681,6 +1823,7 @@ mod tests {
             TurnError {
                 message: "a1".to_string(),
                 codex_error_info: Some(V2CodexErrorInfo::BadRequest),
+                additional_details: None,
             },
             &turn_summary_store,
         )
@@ -1700,6 +1843,7 @@ mod tests {
             TurnError {
                 message: "b1".to_string(),
                 codex_error_info: None,
+                additional_details: None,
             },
             &turn_summary_store,
         )
@@ -1736,6 +1880,7 @@ mod tests {
                     Some(TurnError {
                         message: "a1".to_string(),
                         codex_error_info: Some(V2CodexErrorInfo::BadRequest),
+                        additional_details: None,
                     })
                 );
             }
@@ -1756,6 +1901,7 @@ mod tests {
                     Some(TurnError {
                         message: "b1".to_string(),
                         codex_error_info: None,
+                        additional_details: None,
                     })
                 );
             }
@@ -1791,7 +1937,7 @@ mod tests {
             },
         };
 
-        let thread_id = ConversationId::new().to_string();
+        let thread_id = ThreadId::new().to_string();
         let turn_id = "turn_2".to_string();
         let notification = construct_mcp_tool_call_notification(
             begin_event.clone(),
@@ -1842,7 +1988,7 @@ mod tests {
             result: Ok(result),
         };
 
-        let thread_id = ConversationId::new().to_string();
+        let thread_id = ThreadId::new().to_string();
         let turn_id = "turn_3".to_string();
         let notification = construct_mcp_tool_call_end_notification(
             end_event.clone(),
@@ -1885,7 +2031,7 @@ mod tests {
             result: Err("boom".to_string()),
         };
 
-        let thread_id = ConversationId::new().to_string();
+        let thread_id = ThreadId::new().to_string();
         let turn_id = "turn_4".to_string();
         let notification = construct_mcp_tool_call_end_notification(
             end_event.clone(),
@@ -1919,7 +2065,7 @@ mod tests {
         let (tx, mut rx) = mpsc::channel(CHANNEL_CAPACITY);
         let outgoing = OutgoingMessageSender::new(tx);
         let unified_diff = "--- a\n+++ b\n".to_string();
-        let conversation_id = ConversationId::new();
+        let conversation_id = ThreadId::new();
 
         handle_turn_diff(
             conversation_id,
@@ -1954,7 +2100,7 @@ mod tests {
     async fn test_handle_turn_diff_is_noop_for_v1() -> Result<()> {
         let (tx, mut rx) = mpsc::channel(CHANNEL_CAPACITY);
         let outgoing = OutgoingMessageSender::new(tx);
-        let conversation_id = ConversationId::new();
+        let conversation_id = ThreadId::new();
 
         handle_turn_diff(
             conversation_id,

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

@@ -3,12 +3,18 @@ use crate::error_code::INVALID_REQUEST_ERROR_CODE;
 use codex_app_server_protocol::ConfigBatchWriteParams;
 use codex_app_server_protocol::ConfigReadParams;
 use codex_app_server_protocol::ConfigReadResponse;
+use codex_app_server_protocol::ConfigRequirements;
+use codex_app_server_protocol::ConfigRequirementsReadResponse;
 use codex_app_server_protocol::ConfigValueWriteParams;
 use codex_app_server_protocol::ConfigWriteErrorCode;
 use codex_app_server_protocol::ConfigWriteResponse;
 use codex_app_server_protocol::JSONRPCErrorError;
+use codex_app_server_protocol::SandboxMode;
 use codex_core::config::ConfigService;
 use codex_core::config::ConfigServiceError;
+use codex_core::config_loader::ConfigRequirementsToml;
+use codex_core::config_loader::LoaderOverrides;
+use codex_core::config_loader::SandboxModeRequirement as CoreSandboxModeRequirement;
 use serde_json::json;
 use std::path::PathBuf;
 use toml::Value as TomlValue;
@@ -19,9 +25,13 @@ pub(crate) struct ConfigApi {
 }
 
 impl ConfigApi {
-    pub(crate) fn new(codex_home: PathBuf, cli_overrides: Vec<(String, TomlValue)>) -> Self {
+    pub(crate) fn new(
+        codex_home: PathBuf,
+        cli_overrides: Vec<(String, TomlValue)>,
+        loader_overrides: LoaderOverrides,
+    ) -> Self {
         Self {
-            service: ConfigService::new(codex_home, cli_overrides),
+            service: ConfigService::new(codex_home, cli_overrides, loader_overrides),
         }
     }
 
@@ -32,6 +42,19 @@ impl ConfigApi {
         self.service.read(params).await.map_err(map_error)
     }
 
+    pub(crate) async fn config_requirements_read(
+        &self,
+    ) -> Result<ConfigRequirementsReadResponse, JSONRPCErrorError> {
+        let requirements = self
+            .service
+            .read_requirements()
+            .await
+            .map_err(map_error)?
+            .map(map_requirements_toml_to_api);
+
+        Ok(ConfigRequirementsReadResponse { requirements })
+    }
+
     pub(crate) async fn write_value(
         &self,
         params: ConfigValueWriteParams,
@@ -47,6 +70,32 @@ impl ConfigApi {
     }
 }
 
+fn map_requirements_toml_to_api(requirements: ConfigRequirementsToml) -> ConfigRequirements {
+    ConfigRequirements {
+        allowed_approval_policies: requirements.allowed_approval_policies.map(|policies| {
+            policies
+                .into_iter()
+                .map(codex_app_server_protocol::AskForApproval::from)
+                .collect()
+        }),
+        allowed_sandbox_modes: requirements.allowed_sandbox_modes.map(|modes| {
+            modes
+                .into_iter()
+                .filter_map(map_sandbox_mode_requirement_to_api)
+                .collect()
+        }),
+    }
+}
+
+fn map_sandbox_mode_requirement_to_api(mode: CoreSandboxModeRequirement) -> Option<SandboxMode> {
+    match mode {
+        CoreSandboxModeRequirement::ReadOnly => Some(SandboxMode::ReadOnly),
+        CoreSandboxModeRequirement::WorkspaceWrite => Some(SandboxMode::WorkspaceWrite),
+        CoreSandboxModeRequirement::DangerFullAccess => Some(SandboxMode::DangerFullAccess),
+        CoreSandboxModeRequirement::ExternalSandbox => None,
+    }
+}
+
 fn map_error(err: ConfigServiceError) -> JSONRPCErrorError {
     if let Some(code) = err.write_error_code() {
         return config_write_error(code, err.to_string());
@@ -68,3 +117,38 @@ fn config_write_error(code: ConfigWriteErrorCode, message: impl Into<String>) ->
         })),
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use codex_protocol::protocol::AskForApproval as CoreAskForApproval;
+    use pretty_assertions::assert_eq;
+
+    #[test]
+    fn map_requirements_toml_to_api_converts_core_enums() {
+        let requirements = ConfigRequirementsToml {
+            allowed_approval_policies: Some(vec![
+                CoreAskForApproval::Never,
+                CoreAskForApproval::OnRequest,
+            ]),
+            allowed_sandbox_modes: Some(vec![
+                CoreSandboxModeRequirement::ReadOnly,
+                CoreSandboxModeRequirement::ExternalSandbox,
+            ]),
+        };
+
+        let mapped = map_requirements_toml_to_api(requirements);
+
+        assert_eq!(
+            mapped.allowed_approval_policies,
+            Some(vec![
+                codex_app_server_protocol::AskForApproval::Never,
+                codex_app_server_protocol::AskForApproval::OnRequest,
+            ])
+        );
+        assert_eq!(
+            mapped.allowed_sandbox_modes,
+            Some(vec![SandboxMode::ReadOnly]),
+        );
+    }
+}

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

@@ -1,6 +1,5 @@
 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;
@@ -63,11 +62,7 @@ pub(crate) async fn run_fuzzy_file_search(
             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 file_name = file_search::file_name_from_path(&path);
                     let result = FuzzyFileSearchResult {
                         root: root.clone(),
                         path,

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

@@ -1,7 +1,8 @@
 #![deny(clippy::print_stdout, clippy::print_stderr)]
 
 use codex_common::CliConfigOverrides;
-use codex_core::config::Config;
+use codex_core::config::ConfigBuilder;
+use codex_core::config_loader::LoaderOverrides;
 use std::io::ErrorKind;
 use std::io::Result as IoResult;
 use std::path::PathBuf;
@@ -17,13 +18,11 @@ use tokio::io::BufReader;
 use tokio::io::{self};
 use tokio::sync::mpsc;
 use toml::Value as TomlValue;
-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;
 
@@ -44,6 +43,7 @@ const CHANNEL_CAPACITY: usize = 128;
 pub async fn run_main(
     codex_linux_sandbox_exe: Option<PathBuf>,
     cli_config_overrides: CliConfigOverrides,
+    loader_overrides: LoaderOverrides,
 ) -> IoResult<()> {
     // Set up channels.
     let (incoming_tx, mut incoming_rx) = mpsc::channel::<JSONRPCMessage>(CHANNEL_CAPACITY);
@@ -80,7 +80,11 @@ pub async fn run_main(
             format!("error parsing -c overrides: {e}"),
         )
     })?;
-    let config = Config::load_with_cli_overrides(cli_kv_overrides.clone())
+    let loader_overrides_for_config_api = loader_overrides.clone();
+    let config = ConfigBuilder::default()
+        .cli_overrides(cli_kv_overrides.clone())
+        .loader_overrides(loader_overrides)
+        .build()
         .await
         .map_err(|e| {
             std::io::Error::new(ErrorKind::InvalidData, format!("error loading config: {e}"))
@@ -88,13 +92,18 @@ pub async fn run_main(
 
     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}"),
-            )
-        })?;
+    let otel = codex_core::otel_init::build_provider(
+        &config,
+        env!("CARGO_PKG_VERSION"),
+        Some("codex_app_server"),
+        false,
+    )
+    .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`.
@@ -103,11 +112,8 @@ pub async fn run_main(
         .with_span_events(tracing_subscriber::fmt::format::FmtSpan::FULL)
         .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 feedback_layer = feedback.logger_layer();
+    let feedback_metadata_layer = feedback.metadata_layer();
 
     let otel_logger_layer = otel.as_ref().and_then(|o| o.logger_layer());
 
@@ -116,6 +122,7 @@ pub async fn run_main(
     let _ = tracing_subscriber::registry()
         .with(stderr_fmt)
         .with(feedback_layer)
+        .with(feedback_metadata_layer)
         .with(otel_logger_layer)
         .with(otel_tracing_layer)
         .try_init();
@@ -124,11 +131,13 @@ pub async fn run_main(
     let processor_handle = tokio::spawn({
         let outgoing_message_sender = OutgoingMessageSender::new(outgoing_tx);
         let cli_overrides: Vec<(String, TomlValue)> = cli_kv_overrides.clone();
+        let loader_overrides = loader_overrides_for_config_api;
         let mut processor = MessageProcessor::new(
             outgoing_message_sender,
             codex_linux_sandbox_exe,
             std::sync::Arc::new(config),
             cli_overrides,
+            loader_overrides,
             feedback.clone(),
         );
         async move {

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

@@ -1,10 +1,42 @@
 use codex_app_server::run_main;
 use codex_arg0::arg0_dispatch_or_else;
 use codex_common::CliConfigOverrides;
+use codex_core::config_loader::LoaderOverrides;
+use std::path::PathBuf;
+
+// Debug-only test hook: lets integration tests point the server at a temporary
+// managed config file without writing to /etc.
+const MANAGED_CONFIG_PATH_ENV_VAR: &str = "CODEX_APP_SERVER_MANAGED_CONFIG_PATH";
 
 fn main() -> anyhow::Result<()> {
     arg0_dispatch_or_else(|codex_linux_sandbox_exe| async move {
-        run_main(codex_linux_sandbox_exe, CliConfigOverrides::default()).await?;
+        let managed_config_path = managed_config_path_from_debug_env();
+        let loader_overrides = LoaderOverrides {
+            managed_config_path,
+            ..Default::default()
+        };
+
+        run_main(
+            codex_linux_sandbox_exe,
+            CliConfigOverrides::default(),
+            loader_overrides,
+        )
+        .await?;
         Ok(())
     })
 }
+
+fn managed_config_path_from_debug_env() -> Option<PathBuf> {
+    #[cfg(debug_assertions)]
+    {
+        if let Ok(value) = std::env::var(MANAGED_CONFIG_PATH_ENV_VAR) {
+            return if value.is_empty() {
+                None
+            } else {
+                Some(PathBuf::from(value))
+            };
+        }
+    }
+
+    None
+}

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

@@ -18,10 +18,13 @@ use codex_app_server_protocol::JSONRPCRequest;
 use codex_app_server_protocol::JSONRPCResponse;
 use codex_app_server_protocol::RequestId;
 use codex_core::AuthManager;
-use codex_core::ConversationManager;
+use codex_core::ThreadManager;
 use codex_core::config::Config;
+use codex_core::config_loader::LoaderOverrides;
+use codex_core::default_client::SetOriginatorError;
 use codex_core::default_client::USER_AGENT_SUFFIX;
 use codex_core::default_client::get_codex_user_agent;
+use codex_core::default_client::set_default_originator;
 use codex_feedback::CodexFeedback;
 use codex_protocol::protocol::SessionSource;
 use toml::Value as TomlValue;
@@ -41,6 +44,7 @@ impl MessageProcessor {
         codex_linux_sandbox_exe: Option<PathBuf>,
         config: Arc<Config>,
         cli_overrides: Vec<(String, TomlValue)>,
+        loader_overrides: LoaderOverrides,
         feedback: CodexFeedback,
     ) -> Self {
         let outgoing = Arc::new(outgoing);
@@ -49,20 +53,21 @@ impl MessageProcessor {
             false,
             config.cli_auth_credentials_store_mode,
         );
-        let conversation_manager = Arc::new(ConversationManager::new(
+        let thread_manager = Arc::new(ThreadManager::new(
+            config.codex_home.clone(),
             auth_manager.clone(),
             SessionSource::VSCode,
         ));
         let codex_message_processor = CodexMessageProcessor::new(
             auth_manager,
-            conversation_manager,
+            thread_manager,
             outgoing.clone(),
             codex_linux_sandbox_exe,
             Arc::clone(&config),
             cli_overrides.clone(),
             feedback,
         );
-        let config_api = ConfigApi::new(config.codex_home.clone(), cli_overrides);
+        let config_api = ConfigApi::new(config.codex_home.clone(), cli_overrides, loader_overrides);
 
         Self {
             outgoing,
@@ -118,6 +123,27 @@ impl MessageProcessor {
                         title: _title,
                         version,
                     } = params.client_info;
+                    if let Err(error) = set_default_originator(name.clone()) {
+                        match error {
+                            SetOriginatorError::InvalidHeaderValue => {
+                                let error = JSONRPCErrorError {
+                                    code: INVALID_REQUEST_ERROR_CODE,
+                                    message: format!(
+                                        "Invalid clientInfo.name: '{name}'. Must be a valid HTTP header value."
+                                    ),
+                                    data: None,
+                                };
+                                self.outgoing.send_error(request_id, error).await;
+                                return;
+                            }
+                            SetOriginatorError::AlreadyInitialized => {
+                                // No-op. This is expected to happen if the originator is already set via env var.
+                                // TODO(owen): Once we remove support for CODEX_INTERNAL_ORIGINATOR_OVERRIDE,
+                                // this will be an unexpected state and we can return a JSON-RPC error indicating
+                                // internal server error.
+                            }
+                        }
+                    }
                     let user_agent_suffix = format!("{name}; {version}");
                     if let Ok(mut suffix) = USER_AGENT_SUFFIX.lock() {
                         *suffix = Some(user_agent_suffix);
@@ -155,6 +181,12 @@ impl MessageProcessor {
             ClientRequest::ConfigBatchWrite { request_id, params } => {
                 self.handle_config_batch_write(request_id, params).await;
             }
+            ClientRequest::ConfigRequirementsRead {
+                request_id,
+                params: _,
+            } => {
+                self.handle_config_requirements_read(request_id).await;
+            }
             other => {
                 self.codex_message_processor.process_request(other).await;
             }
@@ -207,4 +239,11 @@ impl MessageProcessor {
             Err(error) => self.outgoing.send_error(request_id, error).await,
         }
     }
+
+    async fn handle_config_requirements_read(&self, request_id: RequestId) {
+        match self.config_api.config_requirements_read().await {
+            Ok(response) => self.outgoing.send_response(request_id, response).await,
+            Err(error) => self.outgoing.send_error(request_id, error).await,
+        }
+    }
 }

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

@@ -2,19 +2,17 @@ use std::sync::Arc;
 
 use codex_app_server_protocol::Model;
 use codex_app_server_protocol::ReasoningEffortOption;
-use codex_core::ConversationManager;
+use codex_core::ThreadManager;
 use codex_core::config::Config;
 use codex_protocol::openai_models::ModelPreset;
 use codex_protocol::openai_models::ReasoningEffortPreset;
 
-pub async fn supported_models(
-    conversation_manager: Arc<ConversationManager>,
-    config: &Config,
-) -> Vec<Model> {
-    conversation_manager
+pub async fn supported_models(thread_manager: Arc<ThreadManager>, config: &Config) -> Vec<Model> {
+    thread_manager
         .list_models(config)
         .await
         .into_iter()
+        .filter(|preset| preset.show_in_picker)
         .map(model_from_preset)
         .collect()
 }

codex-rs/app-server/tests/common/BUILD.bazel

@@ -0,0 +1,7 @@
+load("//:defs.bzl", "codex_rust_crate")
+
+codex_rust_crate(
+    name = "common",
+    crate_name = "app_test_support",
+    crate_srcs = glob(["*.rs"]),
+)

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

@@ -9,12 +9,12 @@ 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, features = ["test-support"] }
 codex-protocol = { workspace = true }
+codex-utils-cargo-bin = { workspace = true }
 serde = { workspace = true }
 serde_json = { workspace = true }
 tokio = { workspace = true, features = [

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

@@ -17,9 +17,11 @@ pub use core_test_support::format_with_current_shell_non_login;
 pub use core_test_support::test_path_buf_with_windows;
 pub use core_test_support::test_tmp_path;
 pub use core_test_support::test_tmp_path_buf;
+pub use mcp_process::DEFAULT_CLIENT_NAME;
 pub use mcp_process::McpProcess;
-pub use mock_model_server::create_mock_chat_completions_server;
-pub use mock_model_server::create_mock_chat_completions_server_unchecked;
+pub use mock_model_server::create_mock_responses_server_repeating_assistant;
+pub use mock_model_server::create_mock_responses_server_sequence;
+pub use mock_model_server::create_mock_responses_server_sequence_unchecked;
 pub use models_cache::write_models_cache;
 pub use models_cache::write_models_cache_with_models;
 pub use responses::create_apply_patch_sse_response;

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

@@ -11,7 +11,6 @@ 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::CancelLoginAccountParams;
@@ -22,6 +21,7 @@ use codex_app_server_protocol::ConfigBatchWriteParams;
 use codex_app_server_protocol::ConfigReadParams;
 use codex_app_server_protocol::ConfigValueWriteParams;
 use codex_app_server_protocol::FeedbackUploadParams;
+use codex_app_server_protocol::ForkConversationParams;
 use codex_app_server_protocol::GetAccountParams;
 use codex_app_server_protocol::GetAuthStatusParams;
 use codex_app_server_protocol::InitializeParams;
@@ -44,12 +44,14 @@ use codex_app_server_protocol::SendUserTurnParams;
 use codex_app_server_protocol::ServerRequest;
 use codex_app_server_protocol::SetDefaultModelParams;
 use codex_app_server_protocol::ThreadArchiveParams;
+use codex_app_server_protocol::ThreadForkParams;
 use codex_app_server_protocol::ThreadListParams;
+use codex_app_server_protocol::ThreadLoadedListParams;
 use codex_app_server_protocol::ThreadResumeParams;
+use codex_app_server_protocol::ThreadRollbackParams;
 use codex_app_server_protocol::ThreadStartParams;
 use codex_app_server_protocol::TurnInterruptParams;
 use codex_app_server_protocol::TurnStartParams;
-use std::process::Command as StdCommand;
 use tokio::process::Command;
 
 pub struct McpProcess {
@@ -61,9 +63,11 @@ pub struct McpProcess {
     process: Child,
     stdin: ChildStdin,
     stdout: BufReader<ChildStdout>,
-    pending_user_messages: VecDeque<JSONRPCNotification>,
+    pending_messages: VecDeque<JSONRPCMessage>,
 }
 
+pub const DEFAULT_CLIENT_NAME: &str = "codex-app-server-tests";
+
 impl McpProcess {
     pub async fn new(codex_home: &Path) -> anyhow::Result<Self> {
         Self::new_with_env(codex_home, &[]).await
@@ -78,12 +82,8 @@ impl McpProcess {
         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 program = codex_utils_cargo_bin::cargo_bin("codex-app-server")
+            .context("should find binary for codex-app-server")?;
         let mut cmd = Command::new(program);
 
         cmd.stdin(Stdio::piped());
@@ -132,39 +132,68 @@ impl McpProcess {
             process,
             stdin,
             stdout,
-            pending_user_messages: VecDeque::new(),
+            pending_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(),
+        let initialized = self
+            .initialize_with_client_info(ClientInfo {
+                name: DEFAULT_CLIENT_NAME.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 {
+            })
+            .await?;
+        let JSONRPCMessage::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(())
     }
 
+    /// Sends initialize with the provided client info and returns the response/error message.
+    pub async fn initialize_with_client_info(
+        &mut self,
+        client_info: ClientInfo,
+    ) -> anyhow::Result<JSONRPCMessage> {
+        let params = Some(serde_json::to_value(InitializeParams { client_info })?);
+        let request_id = self.send_request("initialize", params).await?;
+        let message = self.read_jsonrpc_message().await?;
+        match message {
+            JSONRPCMessage::Response(response) => {
+                if response.id != RequestId::Integer(request_id) {
+                    anyhow::bail!(
+                        "initialize response id mismatch: expected {}, got {:?}",
+                        request_id,
+                        response.id
+                    );
+                }
+
+                // Send notifications/initialized to ack the response.
+                self.send_notification(ClientNotification::Initialized)
+                    .await?;
+
+                Ok(JSONRPCMessage::Response(response))
+            }
+            JSONRPCMessage::Error(error) => {
+                if error.id != RequestId::Integer(request_id) {
+                    anyhow::bail!(
+                        "initialize error id mismatch: expected {}, got {:?}",
+                        request_id,
+                        error.id
+                    );
+                }
+                Ok(JSONRPCMessage::Error(error))
+            }
+            JSONRPCMessage::Notification(notification) => {
+                anyhow::bail!("unexpected JSONRPCMessage::Notification: {notification:?}");
+            }
+            JSONRPCMessage::Request(request) => {
+                anyhow::bail!("unexpected JSONRPCMessage::Request: {request:?}");
+            }
+        }
+    }
+
     /// Send a `newConversation` JSON-RPC request.
     pub async fn send_new_conversation_request(
         &mut self,
@@ -203,7 +232,7 @@ impl McpProcess {
     }
 
     /// Send a `removeConversationListener` JSON-RPC request.
-    pub async fn send_remove_conversation_listener_request(
+    pub async fn send_remove_thread_listener_request(
         &mut self,
         params: RemoveConversationListenerParams,
     ) -> anyhow::Result<i64> {
@@ -313,6 +342,15 @@ impl McpProcess {
         self.send_request("thread/resume", params).await
     }
 
+    /// Send a `thread/fork` JSON-RPC request.
+    pub async fn send_thread_fork_request(
+        &mut self,
+        params: ThreadForkParams,
+    ) -> anyhow::Result<i64> {
+        let params = Some(serde_json::to_value(params)?);
+        self.send_request("thread/fork", params).await
+    }
+
     /// Send a `thread/archive` JSON-RPC request.
     pub async fn send_thread_archive_request(
         &mut self,
@@ -322,6 +360,15 @@ impl McpProcess {
         self.send_request("thread/archive", params).await
     }
 
+    /// Send a `thread/rollback` JSON-RPC request.
+    pub async fn send_thread_rollback_request(
+        &mut self,
+        params: ThreadRollbackParams,
+    ) -> anyhow::Result<i64> {
+        let params = Some(serde_json::to_value(params)?);
+        self.send_request("thread/rollback", params).await
+    }
+
     /// Send a `thread/list` JSON-RPC request.
     pub async fn send_thread_list_request(
         &mut self,
@@ -331,6 +378,15 @@ impl McpProcess {
         self.send_request("thread/list", params).await
     }
 
+    /// Send a `thread/loaded/list` JSON-RPC request.
+    pub async fn send_thread_loaded_list_request(
+        &mut self,
+        params: ThreadLoadedListParams,
+    ) -> anyhow::Result<i64> {
+        let params = Some(serde_json::to_value(params)?);
+        self.send_request("thread/loaded/list", params).await
+    }
+
     /// Send a `model/list` JSON-RPC request.
     pub async fn send_list_models_request(
         &mut self,
@@ -349,6 +405,15 @@ impl McpProcess {
         self.send_request("resumeConversation", params).await
     }
 
+    /// Send a `forkConversation` JSON-RPC request.
+    pub async fn send_fork_conversation_request(
+        &mut self,
+        params: ForkConversationParams,
+    ) -> anyhow::Result<i64> {
+        let params = Some(serde_json::to_value(params)?);
+        self.send_request("forkConversation", params).await
+    }
+
     /// Send a `loginApiKey` JSON-RPC request.
     pub async fn send_login_api_key_request(
         &mut self,
@@ -540,27 +605,16 @@ impl McpProcess {
     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?;
+        let message = self
+            .read_stream_until_message(|message| matches!(message, JSONRPCMessage::Request(_)))
+            .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:?}");
-                }
-            }
-        }
+        let JSONRPCMessage::Request(jsonrpc_request) = message else {
+            unreachable!("expected JSONRPCMessage::Request, got {message:?}");
+        };
+        jsonrpc_request
+            .try_into()
+            .with_context(|| "failed to deserialize ServerRequest from JSONRPCRequest")
     }
 
     pub async fn read_stream_until_response_message(
@@ -569,52 +623,32 @@ impl McpProcess {
     ) -> 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);
-                    }
-                }
-            }
-        }
+        let message = self
+            .read_stream_until_message(|message| {
+                Self::message_request_id(message) == Some(&request_id)
+            })
+            .await?;
+
+        let JSONRPCMessage::Response(response) = message else {
+            unreachable!("expected JSONRPCMessage::Response, got {message:?}");
+        };
+        Ok(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);
-                    }
-                }
-            }
-        }
+        let message = self
+            .read_stream_until_message(|message| {
+                Self::message_request_id(message) == Some(&request_id)
+            })
+            .await?;
+
+        let JSONRPCMessage::Error(err) = message else {
+            unreachable!("expected JSONRPCMessage::Error, got {message:?}");
+        };
+        Ok(err)
     }
 
     pub async fn read_stream_until_notification_message(
@@ -623,46 +657,64 @@ impl McpProcess {
     ) -> 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);
+        let message = self
+            .read_stream_until_message(|message| {
+                matches!(
+                    message,
+                    JSONRPCMessage::Notification(notification) if notification.method == method
+                )
+            })
+            .await?;
+
+        let JSONRPCMessage::Notification(notification) = message else {
+            unreachable!("expected JSONRPCMessage::Notification, got {message:?}");
+        };
+        Ok(notification)
+    }
+
+    /// Clears any buffered messages so future reads only consider new stream items.
+    ///
+    /// We call this when e.g. we want to validate against the next turn and no longer care about
+    /// messages buffered from the prior turn.
+    pub fn clear_message_buffer(&mut self) {
+        self.pending_messages.clear();
+    }
+
+    /// Reads the stream until a message matches `predicate`, buffering any non-matching messages
+    /// for later reads.
+    async fn read_stream_until_message<F>(&mut self, predicate: F) -> anyhow::Result<JSONRPCMessage>
+    where
+        F: Fn(&JSONRPCMessage) -> bool,
+    {
+        if let Some(message) = self.take_pending_message(&predicate) {
+            return Ok(message);
         }
 
         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:?}");
-                }
+            if predicate(&message) {
+                return Ok(message);
             }
+            self.pending_messages.push_back(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);
+    fn take_pending_message<F>(&mut self, predicate: &F) -> Option<JSONRPCMessage>
+    where
+        F: Fn(&JSONRPCMessage) -> bool,
+    {
+        if let Some(pos) = self.pending_messages.iter().position(predicate) {
+            return self.pending_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);
+    fn message_request_id(message: &JSONRPCMessage) -> Option<&RequestId> {
+        match message {
+            JSONRPCMessage::Request(request) => Some(&request.id),
+            JSONRPCMessage::Response(response) => Some(&response.id),
+            JSONRPCMessage::Error(err) => Some(&err.id),
+            JSONRPCMessage::Notification(_) => None,
         }
     }
 }

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

@@ -1,17 +1,18 @@
 use std::sync::atomic::AtomicUsize;
 use std::sync::atomic::Ordering;
 
+use core_test_support::responses;
 use wiremock::Mock;
 use wiremock::MockServer;
 use wiremock::Respond;
 use wiremock::ResponseTemplate;
 use wiremock::matchers::method;
-use wiremock::matchers::path;
+use wiremock::matchers::path_regex;
 
 /// 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;
+/// requests to the `/v1/responses` endpoint.
+pub async fn create_mock_responses_server_sequence(responses: Vec<String>) -> MockServer {
+    let server = responses::start_mock_server().await;
 
     let num_calls = responses.len();
     let seq_responder = SeqResponder {
@@ -20,7 +21,7 @@ pub async fn create_mock_chat_completions_server(responses: Vec<String>) -> Mock
     };
 
     Mock::given(method("POST"))
-        .and(path("/v1/chat/completions"))
+        .and(path_regex(".*/responses$"))
         .respond_with(seq_responder)
         .expect(num_calls as u64)
         .mount(&server)
@@ -29,10 +30,10 @@ pub async fn create_mock_chat_completions_server(responses: Vec<String>) -> Mock
     server
 }
 
-/// Same as `create_mock_chat_completions_server` but does not enforce an
+/// Same as `create_mock_responses_server_sequence` but does not enforce an
 /// expectation on the number of calls.
-pub async fn create_mock_chat_completions_server_unchecked(responses: Vec<String>) -> MockServer {
-    let server = MockServer::start().await;
+pub async fn create_mock_responses_server_sequence_unchecked(responses: Vec<String>) -> MockServer {
+    let server = responses::start_mock_server().await;
 
     let seq_responder = SeqResponder {
         num_calls: AtomicUsize::new(0),
@@ -40,7 +41,7 @@ pub async fn create_mock_chat_completions_server_unchecked(responses: Vec<String
     };
 
     Mock::given(method("POST"))
-        .and(path("/v1/chat/completions"))
+        .and(path_regex(".*/responses$"))
         .respond_with(seq_responder)
         .mount(&server)
         .await;
@@ -57,10 +58,24 @@ impl Respond for SeqResponder {
     fn respond(&self, _: &wiremock::Request) -> ResponseTemplate {
         let call_num = self.num_calls.fetch_add(1, Ordering::SeqCst);
         match self.responses.get(call_num) {
-            Some(response) => ResponseTemplate::new(200)
-                .insert_header("content-type", "text/event-stream")
-                .set_body_raw(response.clone(), "text/event-stream"),
+            Some(response) => responses::sse_response(response.clone()),
             None => panic!("no response for {call_num}"),
         }
     }
 }
+
+/// Create a mock responses API server that returns the same assistant message for every request.
+pub async fn create_mock_responses_server_repeating_assistant(message: &str) -> MockServer {
+    let server = responses::start_mock_server().await;
+    let body = responses::sse(vec![
+        responses::ev_response_created("resp-1"),
+        responses::ev_assistant_message("msg-1", message),
+        responses::ev_completed("resp-1"),
+    ]);
+    Mock::given(method("POST"))
+        .and(path_regex(".*/responses$"))
+        .respond_with(responses::sse_response(body))
+        .mount(&server)
+        .await;
+    server
+}

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

@@ -1,12 +1,10 @@
 use chrono::DateTime;
 use chrono::Utc;
-use codex_core::openai_models::model_presets::all_model_presets;
-use codex_protocol::openai_models::ClientVersion;
+use codex_core::models_manager::model_presets::all_model_presets;
 use codex_protocol::openai_models::ConfigShellToolType;
 use codex_protocol::openai_models::ModelInfo;
 use codex_protocol::openai_models::ModelPreset;
 use codex_protocol::openai_models::ModelVisibility;
-use codex_protocol::openai_models::ReasoningSummaryFormat;
 use codex_protocol::openai_models::TruncationPolicyConfig;
 use serde_json::json;
 use std::path::Path;
@@ -17,7 +15,7 @@ fn preset_to_info(preset: &ModelPreset, priority: i32) -> ModelInfo {
         slug: preset.id.clone(),
         display_name: preset.display_name.clone(),
         description: Some(preset.description.clone()),
-        default_reasoning_level: preset.default_reasoning_effort,
+        default_reasoning_level: Some(preset.default_reasoning_effort),
         supported_reasoning_levels: preset.supported_reasoning_efforts.clone(),
         shell_type: ConfigShellToolType::ShellCommand,
         visibility: if preset.show_in_picker {
@@ -25,24 +23,23 @@ fn preset_to_info(preset: &ModelPreset, priority: i32) -> ModelInfo {
         } else {
             ModelVisibility::Hide
         },
-        minimal_client_version: ClientVersion(0, 1, 0),
         supported_in_api: true,
         priority,
         upgrade: preset.upgrade.as_ref().map(|u| u.id.clone()),
-        base_instructions: None,
+        base_instructions: "base instructions".to_string(),
         supports_reasoning_summaries: false,
         support_verbosity: false,
         default_verbosity: None,
         apply_patch_tool_type: None,
         truncation_policy: TruncationPolicyConfig::bytes(10_000),
         supports_parallel_tool_calls: false,
-        context_window: None,
-        reasoning_summary_format: ReasoningSummaryFormat::None,
+        context_window: Some(272_000),
+        auto_compact_token_limit: None,
+        effective_context_window_percent: 95,
         experimental_supported_tools: Vec::new(),
     }
 }
 
-// todo(aibrahim): fix the priorities to be the opposite here.
 /// Write a models_cache.json file to the codex home directory.
 /// This prevents ModelsManager from making network requests to refresh models.
 /// The cache will be treated as fresh (within TTL) and used instead of fetching from the network.
@@ -53,14 +50,14 @@ pub fn write_models_cache(codex_home: &Path) -> std::io::Result<()> {
         .iter()
         .filter(|preset| preset.show_in_picker)
         .collect();
-    // Convert presets to ModelInfo, assigning priorities (higher = earlier in list)
-    // Priority is used for sorting, so first model gets highest priority
+    // Convert presets to ModelInfo, assigning priorities (lower = earlier in list).
+    // Priority is used for sorting, so the first model gets the lowest priority.
     let models: Vec<ModelInfo> = presets
         .iter()
         .enumerate()
         .map(|(idx, preset)| {
-            // Higher priority = earlier in list, so reverse the index
-            let priority = (presets.len() - idx) as i32;
+            // Lower priority = earlier in list.
+            let priority = idx as i32;
             preset_to_info(preset, priority)
         })
         .collect();

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

@@ -1,3 +1,4 @@
+use core_test_support::responses;
 use serde_json::json;
 use std::path::Path;
 
@@ -14,85 +15,30 @@ pub fn create_shell_command_sse_response(
         "workdir": workdir.map(|w| w.to_string_lossy()),
         "timeout_ms": timeout_ms
     }))?;
-    let tool_call = json!({
-        "choices": [
-            {
-                "delta": {
-                    "tool_calls": [
-                        {
-                            "id": call_id,
-                            "function": {
-                                "name": "shell_command",
-                                "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)
+    Ok(responses::sse(vec![
+        responses::ev_response_created("resp-1"),
+        responses::ev_function_call(call_id, "shell_command", &tool_call_arguments),
+        responses::ev_completed("resp-1"),
+    ]))
 }
 
 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)
+    Ok(responses::sse(vec![
+        responses::ev_response_created("resp-1"),
+        responses::ev_assistant_message("msg-1", message),
+        responses::ev_completed("resp-1"),
+    ]))
 }
 
 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 command = format!("apply_patch <<'EOF'\n{patch_content}\nEOF");
-    let tool_call_arguments = serde_json::to_string(&json!({
-        "command": command
-    }))?;
-
-    let tool_call = json!({
-        "choices": [
-            {
-                "delta": {
-                    "tool_calls": [
-                        {
-                            "id": call_id,
-                            "function": {
-                                "name": "shell_command",
-                                "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)
+    Ok(responses::sse(vec![
+        responses::ev_response_created("resp-1"),
+        responses::ev_apply_patch_shell_command_call_via_heredoc(call_id, patch_content),
+        responses::ev_completed("resp-1"),
+    ]))
 }
 
 pub fn create_exec_command_sse_response(call_id: &str) -> anyhow::Result<String> {
@@ -108,28 +54,9 @@ pub fn create_exec_command_sse_response(call_id: &str) -> anyhow::Result<String>
         "cmd": command.join(" "),
         "yield_time_ms": 500
     }))?;
-    let tool_call = json!({
-        "choices": [
-            {
-                "delta": {
-                    "tool_calls": [
-                        {
-                            "id": call_id,
-                            "function": {
-                                "name": "exec_command",
-                                "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)
+    Ok(responses::sse(vec![
+        responses::ev_response_created("resp-1"),
+        responses::ev_function_call(call_id, "exec_command", &tool_call_arguments),
+        responses::ev_completed("resp-1"),
+    ]))
 }

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

@@ -1,5 +1,5 @@
 use anyhow::Result;
-use codex_protocol::ConversationId;
+use codex_protocol::ThreadId;
 use codex_protocol::protocol::GitInfo;
 use codex_protocol::protocol::SessionMeta;
 use codex_protocol::protocol::SessionMetaLine;
@@ -28,7 +28,7 @@ pub fn create_fake_rollout(
 ) -> Result<String> {
     let uuid = Uuid::new_v4();
     let uuid_str = uuid.to_string();
-    let conversation_id = ConversationId::from_string(&uuid_str)?;
+    let conversation_id = ThreadId::from_string(&uuid_str)?;
 
     // sessions/YYYY/MM/DD derived from filename_ts (YYYY-MM-DDThh-mm-ss)
     let year = &filename_ts[0..4];

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

@@ -37,7 +37,7 @@ 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"
+wire_api = "responses"
 request_max_retries = 0
 stream_max_retries = 0
 {requires_line}

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

@@ -1,7 +1,7 @@
 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_mock_responses_server_sequence;
 use app_test_support::create_shell_command_sse_response;
 use app_test_support::format_with_current_shell;
 use app_test_support::to_response;
@@ -65,7 +65,7 @@ async fn test_codex_jsonrpc_conversation_flow() -> Result<()> {
         )?,
         create_final_assistant_message_sse_response("Enjoy your new git repo!")?,
     ];
-    let server = create_mock_chat_completions_server(responses).await;
+    let server = create_mock_responses_server_sequence(responses).await;
     create_config_toml(&codex_home, &server.uri())?;
 
     // Start MCP server and initialize.
@@ -145,9 +145,7 @@ async fn test_codex_jsonrpc_conversation_flow() -> Result<()> {
 
     // 4) removeConversationListener
     let remove_listener_id = mcp
-        .send_remove_conversation_listener_request(RemoveConversationListenerParams {
-            subscription_id,
-        })
+        .send_remove_thread_listener_request(RemoveConversationListenerParams { subscription_id })
         .await?;
     let remove_listener_resp: JSONRPCResponse = timeout(
         DEFAULT_READ_TIMEOUT,
@@ -199,7 +197,7 @@ async fn test_send_user_turn_changes_approval_policy_behavior() -> Result<()> {
         )?,
         create_final_assistant_message_sse_response("done 2")?,
     ];
-    let server = create_mock_chat_completions_server(responses).await;
+    let server = create_mock_responses_server_sequence(responses).await;
     create_config_toml(&codex_home, &server.uri())?;
 
     // Start MCP server and initialize.
@@ -285,7 +283,7 @@ async fn test_send_user_turn_changes_approval_policy_behavior() -> Result<()> {
     )
     .await?;
 
-    // Wait for first TaskComplete
+    // Wait for first TurnComplete
     let _ = timeout(
         DEFAULT_READ_TIMEOUT,
         mcp.read_stream_until_notification_message("codex/event/task_complete"),
@@ -305,6 +303,7 @@ async fn test_send_user_turn_changes_approval_policy_behavior() -> Result<()> {
             model: "mock-model".to_string(),
             effort: Some(ReasoningEffort::Medium),
             summary: ReasoningSummary::Auto,
+            output_schema: None,
         })
         .await?;
     // Acknowledge sendUserTurn
@@ -364,7 +363,7 @@ async fn test_send_user_turn_updates_sandbox_and_cwd_between_turns() -> Result<(
         )?,
         create_final_assistant_message_sse_response("done second")?,
     ];
-    let server = create_mock_chat_completions_server(responses).await;
+    let server = create_mock_responses_server_sequence(responses).await;
     create_config_toml(&codex_home, &server.uri())?;
 
     let mut mcp = McpProcess::new(&codex_home).await?;
@@ -418,6 +417,7 @@ async fn test_send_user_turn_updates_sandbox_and_cwd_between_turns() -> Result<(
             model: model.clone(),
             effort: Some(ReasoningEffort::Medium),
             summary: ReasoningSummary::Auto,
+            output_schema: None,
         })
         .await?;
     timeout(
@@ -430,6 +430,7 @@ async fn test_send_user_turn_updates_sandbox_and_cwd_between_turns() -> Result<(
         mcp.read_stream_until_notification_message("codex/event/task_complete"),
     )
     .await??;
+    mcp.clear_message_buffer();
 
     let second_turn_id = mcp
         .send_send_user_turn_request(SendUserTurnParams {
@@ -443,6 +444,7 @@ async fn test_send_user_turn_updates_sandbox_and_cwd_between_turns() -> Result<(
             model: model.clone(),
             effort: Some(ReasoningEffort::Medium),
             summary: ReasoningSummary::Auto,
+            output_schema: None,
         })
         .await?;
     timeout(
@@ -498,7 +500,7 @@ model_provider = "mock_provider"
 [model_providers.mock_provider]
 name = "Mock provider for test"
 base_url = "{server_uri}/v1"
-wire_api = "chat"
+wire_api = "responses"
 request_max_retries = 0
 stream_max_retries = 0
 "#

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

@@ -1,7 +1,6 @@
 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;
@@ -12,6 +11,7 @@ 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 core_test_support::responses;
 use pretty_assertions::assert_eq;
 use serde_json::json;
 use std::path::Path;
@@ -23,8 +23,9 @@ const DEFAULT_READ_TIMEOUT: std::time::Duration = std::time::Duration::from_secs
 #[tokio::test(flavor = "multi_thread", worker_threads = 2)]
 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")?];
-    let server = create_mock_chat_completions_server(responses).await;
+    let response_body = create_final_assistant_message_sse_response("Done")?;
+    let server = responses::start_mock_server().await;
+    let response_mock = responses::mount_sse_sequence(&server, vec![response_body]).await;
 
     // Temporary Codex home with config pointing at the mock server.
     let codex_home = TempDir::new()?;
@@ -86,32 +87,30 @@ async fn test_conversation_create_and_send_message_ok() -> Result<()> {
     .await??;
     let _ok: SendUserMessageResponse = to_response::<SendUserMessageResponse>(send_resp)?;
 
-    // avoid race condition by waiting for the mock server to receive the chat.completions request
+    // Avoid race condition by waiting for the mock server to receive the responses request.
     let deadline = std::time::Instant::now() + DEFAULT_READ_TIMEOUT;
     let requests = loop {
-        let requests = server.received_requests().await.unwrap_or_default();
+        let requests = response_mock.requests();
         if !requests.is_empty() {
             break requests;
         }
         if std::time::Instant::now() >= deadline {
-            panic!("mock server did not receive the chat.completions request in time");
+            panic!("mock server did not receive the responses request in time");
         }
         tokio::time::sleep(std::time::Duration::from_millis(10)).await;
     };
 
-    // Verify the outbound request body matches expectations for Chat Completions.
+    // Verify the outbound request body matches expectations for Responses.
     let request = requests
         .first()
         .expect("mock server should have received at least one request");
-    let body = request.body_json::<serde_json::Value>()?;
+    let body = request.body_json();
     assert_eq!(body["model"], json!("o3"));
-    assert!(body["stream"].as_bool().unwrap_or(false));
-    let messages = body["messages"]
-        .as_array()
-        .expect("messages should be array");
-    let last = messages.last().expect("at least one message");
-    assert_eq!(last["role"], json!("user"));
-    assert_eq!(last["content"], json!("Hello"));
+    let user_texts = request.message_input_texts("user");
+    assert!(
+        user_texts.iter().any(|text| text == "Hello"),
+        "expected user input to include Hello, got {user_texts:?}"
+    );
 
     drop(server);
     Ok(())
@@ -133,7 +132,7 @@ model_provider = "mock_provider"
 [model_providers.mock_provider]
 name = "Mock provider for test"
 base_url = "{server_uri}/v1"
-wire_api = "chat"
+wire_api = "responses"
 request_max_retries = 0
 stream_max_retries = 0
 "#

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

@@ -0,0 +1,140 @@
+use anyhow::Result;
+use app_test_support::McpProcess;
+use app_test_support::create_fake_rollout;
+use app_test_support::to_response;
+use codex_app_server_protocol::ForkConversationParams;
+use codex_app_server_protocol::ForkConversationResponse;
+use codex_app_server_protocol::JSONRPCNotification;
+use codex_app_server_protocol::JSONRPCResponse;
+use codex_app_server_protocol::NewConversationParams; // reused for overrides shape
+use codex_app_server_protocol::RequestId;
+use codex_app_server_protocol::ServerNotification;
+use codex_app_server_protocol::SessionConfiguredNotification;
+use codex_core::protocol::EventMsg;
+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 fork_conversation_creates_new_rollout() -> Result<()> {
+    let codex_home = TempDir::new()?;
+
+    let preview = "Hello A";
+    let conversation_id = create_fake_rollout(
+        codex_home.path(),
+        "2025-01-02T12-00-00",
+        "2025-01-02T12:00:00Z",
+        preview,
+        Some("openai"),
+        None,
+    )?;
+
+    let original_path = codex_home
+        .path()
+        .join("sessions")
+        .join("2025")
+        .join("01")
+        .join("02")
+        .join(format!(
+            "rollout-2025-01-02T12-00-00-{conversation_id}.jsonl"
+        ));
+    assert!(
+        original_path.exists(),
+        "expected original rollout to exist at {}",
+        original_path.display()
+    );
+    let original_contents = std::fs::read_to_string(&original_path)?;
+
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
+
+    let fork_req_id = mcp
+        .send_fork_conversation_request(ForkConversationParams {
+            path: Some(original_path.clone()),
+            conversation_id: None,
+            overrides: Some(NewConversationParams {
+                model: Some("o3".to_string()),
+                ..Default::default()
+            }),
+        })
+        .await?;
+
+    // Expect a sessionConfigured notification for the forked session.
+    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,
+        session_id,
+        rollout_path,
+        initial_messages: session_initial_messages,
+        ..
+    }) = session_configured
+    else {
+        unreachable!("expected sessionConfigured notification");
+    };
+
+    assert_eq!(model, "o3");
+    assert_ne!(
+        session_id.to_string(),
+        conversation_id,
+        "expected a new conversation id when forking"
+    );
+    assert_ne!(
+        rollout_path, original_path,
+        "expected a new rollout path when forking"
+    );
+    assert!(
+        rollout_path.exists(),
+        "expected forked rollout to exist at {}",
+        rollout_path.display()
+    );
+
+    let session_initial_messages =
+        session_initial_messages.expect("expected initial messages when forking from rollout");
+    match session_initial_messages.as_slice() {
+        [EventMsg::UserMessage(message)] => {
+            assert_eq!(message.message, preview);
+        }
+        other => panic!("unexpected initial messages from rollout fork: {other:#?}"),
+    }
+
+    // Then the response for forkConversation.
+    let fork_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(fork_req_id)),
+    )
+    .await??;
+    let ForkConversationResponse {
+        conversation_id: forked_id,
+        model: forked_model,
+        initial_messages: response_initial_messages,
+        rollout_path: response_rollout_path,
+    } = to_response::<ForkConversationResponse>(fork_resp)?;
+
+    assert_eq!(forked_model, "o3");
+    assert_eq!(response_rollout_path, rollout_path);
+    assert_ne!(forked_id.to_string(), conversation_id);
+
+    let response_initial_messages =
+        response_initial_messages.expect("expected initial messages in fork response");
+    match response_initial_messages.as_slice() {
+        [EventMsg::UserMessage(message)] => {
+            assert_eq!(message.message, preview);
+        }
+        other => panic!("unexpected initial messages in fork response: {other:#?}"),
+    }
+
+    let after_contents = std::fs::read_to_string(&original_path)?;
+    assert_eq!(
+        after_contents, original_contents,
+        "fork should not mutate the original rollout file"
+    );
+
+    Ok(())
+}

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

@@ -18,7 +18,7 @@ use tempfile::TempDir;
 use tokio::time::timeout;
 
 use app_test_support::McpProcess;
-use app_test_support::create_mock_chat_completions_server;
+use app_test_support::create_mock_responses_server_sequence;
 use app_test_support::create_shell_command_sse_response;
 use app_test_support::to_response;
 
@@ -56,7 +56,7 @@ async fn shell_command_interruption() -> anyhow::Result<()> {
     std::fs::create_dir(&working_directory)?;
 
     // Create mock server with a single SSE response: the long sleep command
-    let server = create_mock_chat_completions_server(vec![create_shell_command_sse_response(
+    let server = create_mock_responses_server_sequence(vec![create_shell_command_sse_response(
         shell_command.clone(),
         Some(&working_directory),
         Some(10_000), // 10 seconds timeout in ms
@@ -153,7 +153,7 @@ model_provider = "mock_provider"
 [model_providers.mock_provider]
 name = "Mock provider for test"
 base_url = "{server_uri}/v1"
-wire_api = "chat"
+wire_api = "responses"
 request_max_retries = 0
 stream_max_retries = 0
 "#

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

@@ -6,7 +6,7 @@ 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::NewConversationParams;
 use codex_app_server_protocol::RequestId;
 use codex_app_server_protocol::ResumeConversationParams;
 use codex_app_server_protocol::ResumeConversationResponse;

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

@@ -32,7 +32,7 @@ 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"
+wire_api = "responses"
 request_max_retries = 0
 stream_max_retries = 0
 "#,

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

@@ -1,12 +1,14 @@
-mod archive_conversation;
+mod archive_thread;
 mod auth;
 mod codex_message_processor_flow;
 mod config;
-mod create_conversation;
+mod create_thread;
+mod fork_thread;
 mod fuzzy_file_search;
 mod interrupt;
 mod list_resume;
 mod login;
+mod output_schema;
 mod send_message;
 mod set_default_model;
 mod user_agent;

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

@@ -0,0 +1,282 @@
+use anyhow::Result;
+use app_test_support::McpProcess;
+use app_test_support::to_response;
+use codex_app_server_protocol::AddConversationListenerParams;
+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::SendUserTurnParams;
+use codex_app_server_protocol::SendUserTurnResponse;
+use codex_core::protocol::AskForApproval;
+use codex_core::protocol::SandboxPolicy;
+use codex_protocol::config_types::ReasoningSummary;
+use codex_protocol::openai_models::ReasoningEffort;
+use core_test_support::responses;
+use core_test_support::skip_if_no_network;
+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 send_user_turn_accepts_output_schema_v1() -> Result<()> {
+    skip_if_no_network!(Ok(()));
+
+    let server = responses::start_mock_server().await;
+    let body = responses::sse(vec![
+        responses::ev_response_created("resp-1"),
+        responses::ev_assistant_message("msg-1", "Done"),
+        responses::ev_completed("resp-1"),
+    ]);
+    let response_mock = responses::mount_sse_once(&server, body).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 {
+            ..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)?;
+
+    let 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(listener_id)),
+    )
+    .await??;
+
+    let output_schema = serde_json::json!({
+        "type": "object",
+        "properties": {
+            "answer": { "type": "string" }
+        },
+        "required": ["answer"],
+        "additionalProperties": false
+    });
+
+    let send_turn_id = mcp
+        .send_send_user_turn_request(SendUserTurnParams {
+            conversation_id,
+            items: vec![InputItem::Text {
+                text: "Hello".to_string(),
+            }],
+            cwd: codex_home.path().to_path_buf(),
+            approval_policy: AskForApproval::Never,
+            sandbox_policy: SandboxPolicy::new_read_only_policy(),
+            model: "mock-model".to_string(),
+            effort: Some(ReasoningEffort::Medium),
+            summary: ReasoningSummary::Auto,
+            output_schema: Some(output_schema.clone()),
+        })
+        .await?;
+    let _send_turn_resp: SendUserTurnResponse = to_response::<SendUserTurnResponse>(
+        timeout(
+            DEFAULT_READ_TIMEOUT,
+            mcp.read_stream_until_response_message(RequestId::Integer(send_turn_id)),
+        )
+        .await??,
+    )?;
+
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("codex/event/task_complete"),
+    )
+    .await??;
+
+    let request = response_mock.single_request();
+    let payload = request.body_json();
+    let text = payload.get("text").expect("request missing text field");
+    let format = text
+        .get("format")
+        .expect("request missing text.format field");
+    assert_eq!(
+        format,
+        &serde_json::json!({
+            "name": "codex_output_schema",
+            "type": "json_schema",
+            "strict": true,
+            "schema": output_schema,
+        })
+    );
+
+    Ok(())
+}
+
+#[tokio::test]
+async fn send_user_turn_output_schema_is_per_turn_v1() -> Result<()> {
+    skip_if_no_network!(Ok(()));
+
+    let server = responses::start_mock_server().await;
+    let body1 = responses::sse(vec![
+        responses::ev_response_created("resp-1"),
+        responses::ev_assistant_message("msg-1", "Done"),
+        responses::ev_completed("resp-1"),
+    ]);
+    let response_mock1 = responses::mount_sse_once(&server, body1).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 {
+            ..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)?;
+
+    let 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(listener_id)),
+    )
+    .await??;
+
+    let output_schema = serde_json::json!({
+        "type": "object",
+        "properties": {
+            "answer": { "type": "string" }
+        },
+        "required": ["answer"],
+        "additionalProperties": false
+    });
+
+    let send_turn_id = mcp
+        .send_send_user_turn_request(SendUserTurnParams {
+            conversation_id,
+            items: vec![InputItem::Text {
+                text: "Hello".to_string(),
+            }],
+            cwd: codex_home.path().to_path_buf(),
+            approval_policy: AskForApproval::Never,
+            sandbox_policy: SandboxPolicy::new_read_only_policy(),
+            model: "mock-model".to_string(),
+            effort: Some(ReasoningEffort::Medium),
+            summary: ReasoningSummary::Auto,
+            output_schema: Some(output_schema.clone()),
+        })
+        .await?;
+    let _send_turn_resp: SendUserTurnResponse = to_response::<SendUserTurnResponse>(
+        timeout(
+            DEFAULT_READ_TIMEOUT,
+            mcp.read_stream_until_response_message(RequestId::Integer(send_turn_id)),
+        )
+        .await??,
+    )?;
+
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("codex/event/task_complete"),
+    )
+    .await??;
+
+    let payload1 = response_mock1.single_request().body_json();
+    assert_eq!(
+        payload1.pointer("/text/format"),
+        Some(&serde_json::json!({
+            "name": "codex_output_schema",
+            "type": "json_schema",
+            "strict": true,
+            "schema": output_schema,
+        }))
+    );
+
+    let body2 = responses::sse(vec![
+        responses::ev_response_created("resp-2"),
+        responses::ev_assistant_message("msg-2", "Done"),
+        responses::ev_completed("resp-2"),
+    ]);
+    let response_mock2 = responses::mount_sse_once(&server, body2).await;
+
+    let send_turn_id_2 = mcp
+        .send_send_user_turn_request(SendUserTurnParams {
+            conversation_id,
+            items: vec![InputItem::Text {
+                text: "Hello again".to_string(),
+            }],
+            cwd: codex_home.path().to_path_buf(),
+            approval_policy: AskForApproval::Never,
+            sandbox_policy: SandboxPolicy::new_read_only_policy(),
+            model: "mock-model".to_string(),
+            effort: Some(ReasoningEffort::Medium),
+            summary: ReasoningSummary::Auto,
+            output_schema: None,
+        })
+        .await?;
+    let _send_turn_resp_2: SendUserTurnResponse = to_response::<SendUserTurnResponse>(
+        timeout(
+            DEFAULT_READ_TIMEOUT,
+            mcp.read_stream_until_response_message(RequestId::Integer(send_turn_id_2)),
+        )
+        .await??,
+    )?;
+
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("codex/event/task_complete"),
+    )
+    .await??;
+
+    let payload2 = response_mock2.single_request().body_json();
+    assert_eq!(payload2.pointer("/text/format"), None);
+
+    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 = "never"
+sandbox_mode = "read-only"
+
+model_provider = "mock_provider"
+
+[model_providers.mock_provider]
+name = "Mock provider for test"
+base_url = "{server_uri}/v1"
+wire_api = "responses"
+request_max_retries = 0
+stream_max_retries = 0
+"#
+        ),
+    )
+}

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

@@ -1,7 +1,5 @@
 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;
@@ -13,10 +11,11 @@ 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::ThreadId;
 use codex_protocol::models::ContentItem;
 use codex_protocol::models::ResponseItem;
 use codex_protocol::protocol::RawResponseItemEvent;
+use core_test_support::responses;
 use pretty_assertions::assert_eq;
 use std::path::Path;
 use tempfile::TempDir;
@@ -26,13 +25,21 @@ const DEFAULT_READ_TIMEOUT: std::time::Duration = std::time::Duration::from_secs
 
 #[tokio::test]
 async fn test_send_message_success() -> Result<()> {
-    // Spin up a mock completions server that immediately ends the Codex turn.
+    // Spin up a mock responses 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;
+    let server = responses::start_mock_server().await;
+    let body1 = responses::sse(vec![
+        responses::ev_response_created("resp-1"),
+        responses::ev_assistant_message("msg-1", "Done"),
+        responses::ev_completed("resp-1"),
+    ]);
+    let body2 = responses::sse(vec![
+        responses::ev_response_created("resp-2"),
+        responses::ev_assistant_message("msg-2", "Done"),
+        responses::ev_completed("resp-2"),
+    ]);
+    let _response_mock1 = responses::mount_sse_once(&server, body1).await;
+    let _response_mock2 = responses::mount_sse_once(&server, body2).await;
 
     // Create a temporary Codex home with config pointing at the mock server.
     let codex_home = TempDir::new()?;
@@ -81,7 +88,7 @@ async fn test_send_message_success() -> Result<()> {
 #[expect(clippy::expect_used)]
 async fn send_message(
     message: &str,
-    conversation_id: ConversationId,
+    conversation_id: ThreadId,
     mcp: &mut McpProcess,
 ) -> Result<()> {
     // Now exercise sendUserMessage.
@@ -135,8 +142,13 @@ async fn send_message(
 
 #[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 server = responses::start_mock_server().await;
+    let body = responses::sse(vec![
+        responses::ev_response_created("resp-1"),
+        responses::ev_assistant_message("msg-1", "Done"),
+        responses::ev_completed("resp-1"),
+    ]);
+    let _response_mock = responses::mount_sse_once(&server, body).await;
 
     let codex_home = TempDir::new()?;
     create_config_toml(codex_home.path(), &server.uri())?;
@@ -220,7 +232,7 @@ async fn test_send_message_session_not_found() -> Result<()> {
     let mut mcp = McpProcess::new(codex_home.path()).await?;
     timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
 
-    let unknown = ConversationId::new();
+    let unknown = ThreadId::new();
     let req_id = mcp
         .send_send_user_message_request(SendUserMessageParams {
             conversation_id: unknown,
@@ -259,7 +271,7 @@ model_provider = "mock_provider"
 [model_providers.mock_provider]
 name = "Mock provider for test"
 base_url = "{server_uri}/v1"
-wire_api = "chat"
+wire_api = "responses"
 request_max_retries = 0
 stream_max_retries = 0
 "#
@@ -268,10 +280,8 @@ stream_max_retries = 0
 }
 
 #[expect(clippy::expect_used)]
-async fn read_raw_response_item(
-    mcp: &mut McpProcess,
-    conversation_id: ConversationId,
-) -> ResponseItem {
+async fn read_raw_response_item(mcp: &mut McpProcess, conversation_id: ThreadId) -> ResponseItem {
+    // TODO: Switch to rawResponseItem/completed once we migrate to app server v2 in codex web.
     loop {
         let raw_notification: JSONRPCNotification = timeout(
             DEFAULT_READ_TIMEOUT,

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

@@ -1,4 +1,5 @@
 use anyhow::Result;
+use app_test_support::DEFAULT_CLIENT_NAME;
 use app_test_support::McpProcess;
 use app_test_support::to_response;
 use codex_app_server_protocol::GetUserAgentResponse;
@@ -25,13 +26,13 @@ async fn get_user_agent_returns_current_codex_user_agent() -> Result<()> {
     .await??;
 
     let os_info = os_info::get();
-    let originator = codex_core::default_client::originator().value.as_str();
+    let originator = DEFAULT_CLIENT_NAME;
     let os_type = os_info.os_type();
     let os_version = os_info.version();
     let architecture = os_info.architecture().unwrap_or("unknown");
     let terminal_ua = codex_core::terminal::user_agent();
     let user_agent = format!(
-        "{originator}/0.0.0 ({os_type} {os_version}; {architecture}) {terminal_ua} (codex-app-server-tests; 0.1.0)"
+        "{originator}/0.0.0 ({os_type} {os_version}; {architecture}) {terminal_ua} ({DEFAULT_CLIENT_NAME}; 0.1.0)"
     );
 
     let received: GetUserAgentResponse = to_response(response)?;

codex-rs/app-server/tests/suite/v2/account.rs

@@ -67,7 +67,7 @@ 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"
+wire_api = "responses"
 request_max_retries = 0
 stream_max_retries = 0
 {requires_line}

codex-rs/app-server/tests/suite/v2/config_rpc.rs

@@ -18,6 +18,7 @@ use codex_app_server_protocol::RequestId;
 use codex_app_server_protocol::SandboxMode;
 use codex_app_server_protocol::ToolsV2;
 use codex_app_server_protocol::WriteStatus;
+use codex_core::config_loader::SYSTEM_CONFIG_TOML_FILE_UNIX;
 use codex_utils_absolute_path::AbsolutePathBuf;
 use pretty_assertions::assert_eq;
 use serde_json::json;
@@ -73,8 +74,7 @@ sandbox_mode = "workspace-write"
         }
     );
     let layers = layers.expect("layers present");
-    assert_eq!(layers.len(), 1);
-    assert_eq!(layers[0].name, ConfigLayerSource::User { file: user_file });
+    assert_layers_user_then_optional_system(&layers, user_file)?;
 
     Ok(())
 }
@@ -136,8 +136,7 @@ view_image = false
     );
 
     let layers = layers.expect("layers present");
-    assert_eq!(layers.len(), 1);
-    assert_eq!(layers[0].name, ConfigLayerSource::User { file: user_file });
+    assert_layers_user_then_optional_system(&layers, user_file)?;
 
     Ok(())
 }
@@ -185,7 +184,10 @@ writable_roots = [{}]
 
     let mut mcp = McpProcess::new_with_env(
         codex_home.path(),
-        &[("CODEX_MANAGED_CONFIG_PATH", Some(&managed_path_str))],
+        &[(
+            "CODEX_APP_SERVER_MANAGED_CONFIG_PATH",
+            Some(&managed_path_str),
+        )],
     )
     .await?;
     timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
@@ -257,12 +259,7 @@ writable_roots = [{}]
     );
 
     let layers = layers.expect("layers present");
-    assert_eq!(layers.len(), 2);
-    assert_eq!(
-        layers[0].name,
-        ConfigLayerSource::LegacyManagedConfigTomlFromFile { file: managed_file }
-    );
-    assert_eq!(layers[1].name, ConfigLayerSource::User { file: user_file });
+    assert_layers_managed_user_then_optional_system(&layers, managed_file, user_file)?;
 
     Ok(())
 }
@@ -433,3 +430,50 @@ async fn config_batch_write_applies_multiple_edits() -> Result<()> {
 
     Ok(())
 }
+
+fn assert_layers_user_then_optional_system(
+    layers: &[codex_app_server_protocol::ConfigLayer],
+    user_file: AbsolutePathBuf,
+) -> Result<()> {
+    if cfg!(unix) {
+        let system_file = AbsolutePathBuf::from_absolute_path(SYSTEM_CONFIG_TOML_FILE_UNIX)?;
+        assert_eq!(layers.len(), 2);
+        assert_eq!(layers[0].name, ConfigLayerSource::User { file: user_file });
+        assert_eq!(
+            layers[1].name,
+            ConfigLayerSource::System { file: system_file }
+        );
+    } else {
+        assert_eq!(layers.len(), 1);
+        assert_eq!(layers[0].name, ConfigLayerSource::User { file: user_file });
+    }
+    Ok(())
+}
+
+fn assert_layers_managed_user_then_optional_system(
+    layers: &[codex_app_server_protocol::ConfigLayer],
+    managed_file: AbsolutePathBuf,
+    user_file: AbsolutePathBuf,
+) -> Result<()> {
+    if cfg!(unix) {
+        let system_file = AbsolutePathBuf::from_absolute_path(SYSTEM_CONFIG_TOML_FILE_UNIX)?;
+        assert_eq!(layers.len(), 3);
+        assert_eq!(
+            layers[0].name,
+            ConfigLayerSource::LegacyManagedConfigTomlFromFile { file: managed_file }
+        );
+        assert_eq!(layers[1].name, ConfigLayerSource::User { file: user_file });
+        assert_eq!(
+            layers[2].name,
+            ConfigLayerSource::System { file: system_file }
+        );
+    } else {
+        assert_eq!(layers.len(), 2);
+        assert_eq!(
+            layers[0].name,
+            ConfigLayerSource::LegacyManagedConfigTomlFromFile { file: managed_file }
+        );
+        assert_eq!(layers[1].name, ConfigLayerSource::User { file: user_file });
+    }
+    Ok(())
+}

codex-rs/app-server/tests/suite/v2/initialize.rs

@@ -0,0 +1,137 @@
+use anyhow::Result;
+use app_test_support::McpProcess;
+use app_test_support::create_mock_responses_server_sequence_unchecked;
+use app_test_support::to_response;
+use codex_app_server_protocol::ClientInfo;
+use codex_app_server_protocol::InitializeResponse;
+use codex_app_server_protocol::JSONRPCMessage;
+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 initialize_uses_client_info_name_as_originator() -> Result<()> {
+    let responses = Vec::new();
+    let server = create_mock_responses_server_sequence_unchecked(responses).await;
+    let codex_home = TempDir::new()?;
+    create_config_toml(codex_home.path(), &server.uri(), "never")?;
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+
+    let message = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.initialize_with_client_info(ClientInfo {
+            name: "codex_vscode".to_string(),
+            title: Some("Codex VS Code Extension".to_string()),
+            version: "0.1.0".to_string(),
+        }),
+    )
+    .await??;
+
+    let JSONRPCMessage::Response(response) = message else {
+        anyhow::bail!("expected initialize response, got {message:?}");
+    };
+    let InitializeResponse { user_agent } = to_response::<InitializeResponse>(response)?;
+
+    assert!(user_agent.starts_with("codex_vscode/"));
+    Ok(())
+}
+
+#[tokio::test]
+async fn initialize_respects_originator_override_env_var() -> Result<()> {
+    let responses = Vec::new();
+    let server = create_mock_responses_server_sequence_unchecked(responses).await;
+    let codex_home = TempDir::new()?;
+    create_config_toml(codex_home.path(), &server.uri(), "never")?;
+    let mut mcp = McpProcess::new_with_env(
+        codex_home.path(),
+        &[(
+            "CODEX_INTERNAL_ORIGINATOR_OVERRIDE",
+            Some("codex_originator_via_env_var"),
+        )],
+    )
+    .await?;
+
+    let message = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.initialize_with_client_info(ClientInfo {
+            name: "codex_vscode".to_string(),
+            title: Some("Codex VS Code Extension".to_string()),
+            version: "0.1.0".to_string(),
+        }),
+    )
+    .await??;
+
+    let JSONRPCMessage::Response(response) = message else {
+        anyhow::bail!("expected initialize response, got {message:?}");
+    };
+    let InitializeResponse { user_agent } = to_response::<InitializeResponse>(response)?;
+
+    assert!(user_agent.starts_with("codex_originator_via_env_var/"));
+    Ok(())
+}
+
+#[tokio::test]
+async fn initialize_rejects_invalid_client_name() -> Result<()> {
+    let responses = Vec::new();
+    let server = create_mock_responses_server_sequence_unchecked(responses).await;
+    let codex_home = TempDir::new()?;
+    create_config_toml(codex_home.path(), &server.uri(), "never")?;
+    let mut mcp = McpProcess::new_with_env(
+        codex_home.path(),
+        &[("CODEX_INTERNAL_ORIGINATOR_OVERRIDE", None)],
+    )
+    .await?;
+
+    let message = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.initialize_with_client_info(ClientInfo {
+            name: "bad\rname".to_string(),
+            title: Some("Bad Client".to_string()),
+            version: "0.1.0".to_string(),
+        }),
+    )
+    .await??;
+
+    let JSONRPCMessage::Error(error) = message else {
+        anyhow::bail!("expected initialize error, got {message:?}");
+    };
+
+    assert_eq!(error.error.code, -32600);
+    assert_eq!(
+        error.error.message,
+        "Invalid clientInfo.name: 'bad\rname'. Must be a valid HTTP header value."
+    );
+    assert_eq!(error.error.data, None);
+    Ok(())
+}
+
+// Helper to create a config.toml pointing at the mock model server.
+fn create_config_toml(
+    codex_home: &Path,
+    server_uri: &str,
+    approval_policy: &str,
+) -> std::io::Result<()> {
+    let config_toml = codex_home.join("config.toml");
+    std::fs::write(
+        config_toml,
+        format!(
+            r#"
+model = "mock-model"
+approval_policy = "{approval_policy}"
+sandbox_mode = "read-only"
+
+model_provider = "mock_provider"
+
+[model_providers.mock_provider]
+name = "Mock provider for test"
+base_url = "{server_uri}/v1"
+wire_api = "responses"
+request_max_retries = 0
+stream_max_retries = 0
+"#
+        ),
+    )
+}

codex-rs/app-server/tests/suite/v2/mod.rs

@@ -1,11 +1,16 @@
 mod account;
 mod config_rpc;
+mod initialize;
 mod model_list;
+mod output_schema;
 mod rate_limits;
 mod review;
 mod thread_archive;
+mod thread_fork;
 mod thread_list;
+mod thread_loaded_list;
 mod thread_resume;
+mod thread_rollback;
 mod thread_start;
 mod turn_interrupt;
 mod turn_start;

codex-rs/app-server/tests/suite/v2/model_list.rs

@@ -48,57 +48,32 @@ async fn list_models_returns_all_models_with_large_limit() -> Result<()> {
 
     let expected_models = vec![
         Model {
-            id: "gpt-5.2".to_string(),
-            model: "gpt-5.2".to_string(),
-            display_name: "gpt-5.2".to_string(),
-            description:
-                "Latest frontier model with improvements across knowledge, reasoning and coding"
-                    .to_string(),
+            id: "gpt-5.2-codex".to_string(),
+            model: "gpt-5.2-codex".to_string(),
+            display_name: "gpt-5.2-codex".to_string(),
+            description: "Latest frontier agentic coding model.".to_string(),
             supported_reasoning_efforts: vec![
                 ReasoningEffortOption {
                     reasoning_effort: ReasoningEffort::Low,
-                    description: "Balances speed with some reasoning; useful for straightforward \
-                                   queries and short explanations"
-                        .to_string(),
+                    description: "Fast responses with lighter reasoning".to_string(),
                 },
                 ReasoningEffortOption {
                     reasoning_effort: ReasoningEffort::Medium,
-                    description: "Provides a solid balance of reasoning depth and latency for \
-                         general-purpose tasks"
+                    description: "Balances speed and reasoning depth for everyday tasks"
                         .to_string(),
                 },
                 ReasoningEffortOption {
                     reasoning_effort: ReasoningEffort::High,
-                    description: "Maximizes reasoning depth for complex or ambiguous problems"
-                        .to_string(),
+                    description: "Greater reasoning depth for complex problems".to_string(),
                 },
                 ReasoningEffortOption {
                     reasoning_effort: ReasoningEffort::XHigh,
-                    description: "Extra high reasoning for complex problems".to_string(),
+                    description: "Extra high reasoning depth for complex problems".to_string(),
                 },
             ],
             default_reasoning_effort: ReasoningEffort::Medium,
             is_default: true,
         },
-        Model {
-            id: "gpt-5.1-codex-mini".to_string(),
-            model: "gpt-5.1-codex-mini".to_string(),
-            display_name: "gpt-5.1-codex-mini".to_string(),
-            description: "Optimized for codex. Cheaper, faster, but less capable.".to_string(),
-            supported_reasoning_efforts: vec![
-                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: false,
-        },
         Model {
             id: "gpt-5.1-codex-max".to_string(),
             model: "gpt-5.1-codex-max".to_string(),
@@ -127,23 +102,48 @@ async fn list_models_returns_all_models_with_large_limit() -> Result<()> {
             is_default: false,
         },
         Model {
-            id: "gpt-5.2-codex".to_string(),
-            model: "gpt-5.2-codex".to_string(),
-            display_name: "gpt-5.2-codex".to_string(),
-            description: "Latest frontier agentic coding model.".to_string(),
+            id: "gpt-5.1-codex-mini".to_string(),
+            model: "gpt-5.1-codex-mini".to_string(),
+            display_name: "gpt-5.1-codex-mini".to_string(),
+            description: "Optimized for codex. Cheaper, faster, but less capable.".to_string(),
+            supported_reasoning_efforts: vec![
+                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: false,
+        },
+        Model {
+            id: "gpt-5.2".to_string(),
+            model: "gpt-5.2".to_string(),
+            display_name: "gpt-5.2".to_string(),
+            description:
+                "Latest frontier model with improvements across knowledge, reasoning and coding"
+                    .to_string(),
             supported_reasoning_efforts: vec![
                 ReasoningEffortOption {
                     reasoning_effort: ReasoningEffort::Low,
-                    description: "Fast responses with lighter reasoning".to_string(),
+                    description: "Balances speed with some reasoning; useful for straightforward \
+                                   queries and short explanations"
+                        .to_string(),
                 },
                 ReasoningEffortOption {
                     reasoning_effort: ReasoningEffort::Medium,
-                    description: "Balances speed and reasoning depth for everyday tasks"
+                    description: "Provides a solid balance of reasoning depth and latency for \
+                         general-purpose tasks"
                         .to_string(),
                 },
                 ReasoningEffortOption {
                     reasoning_effort: ReasoningEffort::High,
-                    description: "Greater reasoning depth for complex problems".to_string(),
+                    description: "Maximizes reasoning depth for complex or ambiguous problems"
+                        .to_string(),
                 },
                 ReasoningEffortOption {
                     reasoning_effort: ReasoningEffort::XHigh,
@@ -187,7 +187,7 @@ async fn list_models_pagination_works() -> Result<()> {
     } = to_response::<ModelListResponse>(first_response)?;
 
     assert_eq!(first_items.len(), 1);
-    assert_eq!(first_items[0].id, "gpt-5.2");
+    assert_eq!(first_items[0].id, "gpt-5.2-codex");
     let next_cursor = first_cursor.ok_or_else(|| anyhow!("cursor for second page"))?;
 
     let second_request = mcp
@@ -209,7 +209,7 @@ async fn list_models_pagination_works() -> Result<()> {
     } = to_response::<ModelListResponse>(second_response)?;
 
     assert_eq!(second_items.len(), 1);
-    assert_eq!(second_items[0].id, "gpt-5.1-codex-mini");
+    assert_eq!(second_items[0].id, "gpt-5.1-codex-max");
     let third_cursor = second_cursor.ok_or_else(|| anyhow!("cursor for third page"))?;
 
     let third_request = mcp
@@ -231,7 +231,7 @@ async fn list_models_pagination_works() -> Result<()> {
     } = to_response::<ModelListResponse>(third_response)?;
 
     assert_eq!(third_items.len(), 1);
-    assert_eq!(third_items[0].id, "gpt-5.1-codex-max");
+    assert_eq!(third_items[0].id, "gpt-5.1-codex-mini");
     let fourth_cursor = third_cursor.ok_or_else(|| anyhow!("cursor for fourth page"))?;
 
     let fourth_request = mcp
@@ -253,7 +253,7 @@ async fn list_models_pagination_works() -> Result<()> {
     } = to_response::<ModelListResponse>(fourth_response)?;
 
     assert_eq!(fourth_items.len(), 1);
-    assert_eq!(fourth_items[0].id, "gpt-5.2-codex");
+    assert_eq!(fourth_items[0].id, "gpt-5.2");
     assert!(fourth_cursor.is_none());
     Ok(())
 }

codex-rs/app-server/tests/suite/v2/output_schema.rs

@@ -0,0 +1,231 @@
+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::ThreadStartParams;
+use codex_app_server_protocol::ThreadStartResponse;
+use codex_app_server_protocol::TurnStartParams;
+use codex_app_server_protocol::TurnStartResponse;
+use codex_app_server_protocol::UserInput as V2UserInput;
+use core_test_support::responses;
+use core_test_support::skip_if_no_network;
+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 turn_start_accepts_output_schema_v2() -> Result<()> {
+    skip_if_no_network!(Ok(()));
+
+    let server = responses::start_mock_server().await;
+    let body = responses::sse(vec![
+        responses::ev_response_created("resp-1"),
+        responses::ev_assistant_message("msg-1", "Done"),
+        responses::ev_completed("resp-1"),
+    ]);
+    let response_mock = responses::mount_sse_once(&server, body).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 thread_req = mcp
+        .send_thread_start_request(ThreadStartParams {
+            ..Default::default()
+        })
+        .await?;
+    let thread_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(thread_req)),
+    )
+    .await??;
+    let ThreadStartResponse { thread, .. } = to_response::<ThreadStartResponse>(thread_resp)?;
+
+    let output_schema = serde_json::json!({
+        "type": "object",
+        "properties": {
+            "answer": { "type": "string" }
+        },
+        "required": ["answer"],
+        "additionalProperties": false
+    });
+
+    let turn_req = mcp
+        .send_turn_start_request(TurnStartParams {
+            thread_id: thread.id.clone(),
+            input: vec![V2UserInput::Text {
+                text: "Hello".to_string(),
+            }],
+            output_schema: Some(output_schema.clone()),
+            ..Default::default()
+        })
+        .await?;
+    let turn_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(turn_req)),
+    )
+    .await??;
+    let _turn: TurnStartResponse = to_response::<TurnStartResponse>(turn_resp)?;
+
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("turn/completed"),
+    )
+    .await??;
+
+    let request = response_mock.single_request();
+    let payload = request.body_json();
+    let text = payload.get("text").expect("request missing text field");
+    let format = text
+        .get("format")
+        .expect("request missing text.format field");
+    assert_eq!(
+        format,
+        &serde_json::json!({
+            "name": "codex_output_schema",
+            "type": "json_schema",
+            "strict": true,
+            "schema": output_schema,
+        })
+    );
+
+    Ok(())
+}
+
+#[tokio::test]
+async fn turn_start_output_schema_is_per_turn_v2() -> Result<()> {
+    skip_if_no_network!(Ok(()));
+
+    let server = responses::start_mock_server().await;
+    let body1 = responses::sse(vec![
+        responses::ev_response_created("resp-1"),
+        responses::ev_assistant_message("msg-1", "Done"),
+        responses::ev_completed("resp-1"),
+    ]);
+    let response_mock1 = responses::mount_sse_once(&server, body1).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 thread_req = mcp
+        .send_thread_start_request(ThreadStartParams {
+            ..Default::default()
+        })
+        .await?;
+    let thread_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(thread_req)),
+    )
+    .await??;
+    let ThreadStartResponse { thread, .. } = to_response::<ThreadStartResponse>(thread_resp)?;
+
+    let output_schema = serde_json::json!({
+        "type": "object",
+        "properties": {
+            "answer": { "type": "string" }
+        },
+        "required": ["answer"],
+        "additionalProperties": false
+    });
+
+    let turn_req_1 = mcp
+        .send_turn_start_request(TurnStartParams {
+            thread_id: thread.id.clone(),
+            input: vec![V2UserInput::Text {
+                text: "Hello".to_string(),
+            }],
+            output_schema: Some(output_schema.clone()),
+            ..Default::default()
+        })
+        .await?;
+    let turn_resp_1: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(turn_req_1)),
+    )
+    .await??;
+    let _turn: TurnStartResponse = to_response::<TurnStartResponse>(turn_resp_1)?;
+
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("turn/completed"),
+    )
+    .await??;
+
+    let payload1 = response_mock1.single_request().body_json();
+    assert_eq!(
+        payload1.pointer("/text/format"),
+        Some(&serde_json::json!({
+            "name": "codex_output_schema",
+            "type": "json_schema",
+            "strict": true,
+            "schema": output_schema,
+        }))
+    );
+
+    let body2 = responses::sse(vec![
+        responses::ev_response_created("resp-2"),
+        responses::ev_assistant_message("msg-2", "Done"),
+        responses::ev_completed("resp-2"),
+    ]);
+    let response_mock2 = responses::mount_sse_once(&server, body2).await;
+
+    let turn_req_2 = mcp
+        .send_turn_start_request(TurnStartParams {
+            thread_id: thread.id.clone(),
+            input: vec![V2UserInput::Text {
+                text: "Hello again".to_string(),
+            }],
+            output_schema: None,
+            ..Default::default()
+        })
+        .await?;
+    let turn_resp_2: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(turn_req_2)),
+    )
+    .await??;
+    let _turn: TurnStartResponse = to_response::<TurnStartResponse>(turn_resp_2)?;
+
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("turn/completed"),
+    )
+    .await??;
+
+    let payload2 = response_mock2.single_request().body_json();
+    assert_eq!(payload2.pointer("/text/format"), None);
+
+    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 = "never"
+sandbox_mode = "read-only"
+
+model_provider = "mock_provider"
+
+[model_providers.mock_provider]
+name = "Mock provider for test"
+base_url = "{server_uri}/v1"
+wire_api = "responses"
+request_max_retries = 0
+stream_max_retries = 0
+"#
+        ),
+    )
+}

codex-rs/app-server/tests/suite/v2/review.rs

@@ -1,7 +1,6 @@
 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_unchecked;
+use app_test_support::create_mock_responses_server_repeating_assistant;
 use app_test_support::to_response;
 use codex_app_server_protocol::ItemCompletedNotification;
 use codex_app_server_protocol::ItemStartedNotification;
@@ -44,10 +43,7 @@ async fn review_start_runs_review_turn_and_emits_code_review_item() -> Result<()
         "overall_confidence_score": 0.75
     })
     .to_string();
-    let responses = vec![create_final_assistant_message_sse_response(
-        &review_payload,
-    )?];
-    let server = create_mock_chat_completions_server_unchecked(responses).await;
+    let server = create_mock_responses_server_repeating_assistant(&review_payload).await;
 
     let codex_home = TempDir::new()?;
     create_config_toml(codex_home.path(), &server.uri())?;
@@ -135,7 +131,7 @@ async fn review_start_runs_review_turn_and_emits_code_review_item() -> Result<()
 
 #[tokio::test]
 async fn review_start_rejects_empty_base_branch() -> Result<()> {
-    let server = create_mock_chat_completions_server_unchecked(vec![]).await;
+    let server = create_mock_responses_server_repeating_assistant("Done").await;
     let codex_home = TempDir::new()?;
     create_config_toml(codex_home.path(), &server.uri())?;
 
@@ -176,10 +172,7 @@ async fn review_start_with_detached_delivery_returns_new_thread_id() -> Result<(
         "overall_confidence_score": 0.5
     })
     .to_string();
-    let responses = vec![create_final_assistant_message_sse_response(
-        &review_payload,
-    )?];
-    let server = create_mock_chat_completions_server_unchecked(responses).await;
+    let server = create_mock_responses_server_repeating_assistant(&review_payload).await;
 
     let codex_home = TempDir::new()?;
     create_config_toml(codex_home.path(), &server.uri())?;
@@ -219,7 +212,7 @@ async fn review_start_with_detached_delivery_returns_new_thread_id() -> Result<(
 
 #[tokio::test]
 async fn review_start_rejects_empty_commit_sha() -> Result<()> {
-    let server = create_mock_chat_completions_server_unchecked(vec![]).await;
+    let server = create_mock_responses_server_repeating_assistant("Done").await;
     let codex_home = TempDir::new()?;
     create_config_toml(codex_home.path(), &server.uri())?;
 
@@ -254,7 +247,7 @@ async fn review_start_rejects_empty_commit_sha() -> Result<()> {
 
 #[tokio::test]
 async fn review_start_rejects_empty_custom_instructions() -> Result<()> {
-    let server = create_mock_chat_completions_server_unchecked(vec![]).await;
+    let server = create_mock_responses_server_repeating_assistant("Done").await;
     let codex_home = TempDir::new()?;
     create_config_toml(codex_home.path(), &server.uri())?;
 
@@ -320,7 +313,7 @@ model_provider = "mock_provider"
 [model_providers.mock_provider]
 name = "Mock provider"
 base_url = "{server_uri}/v1"
-wire_api = "chat"
+wire_api = "responses"
 request_max_retries = 0
 stream_max_retries = 0
 "#

codex-rs/app-server/tests/suite/v2/thread_archive.rs

@@ -8,7 +8,7 @@ use codex_app_server_protocol::ThreadArchiveResponse;
 use codex_app_server_protocol::ThreadStartParams;
 use codex_app_server_protocol::ThreadStartResponse;
 use codex_core::ARCHIVED_SESSIONS_SUBDIR;
-use codex_core::find_conversation_path_by_id_str;
+use codex_core::find_thread_path_by_id_str;
 use std::path::Path;
 use tempfile::TempDir;
 use tokio::time::timeout;
@@ -39,7 +39,7 @@ async fn thread_archive_moves_rollout_into_archived_directory() -> Result<()> {
     assert!(!thread.id.is_empty());
 
     // Locate the rollout path recorded for this thread id.
-    let rollout_path = find_conversation_path_by_id_str(codex_home.path(), &thread.id)
+    let rollout_path = find_thread_path_by_id_str(codex_home.path(), &thread.id)
         .await?
         .expect("expected rollout path for thread id to exist");
     assert!(

codex-rs/app-server/tests/suite/v2/thread_fork.rs

@@ -0,0 +1,140 @@
+use anyhow::Result;
+use app_test_support::McpProcess;
+use app_test_support::create_fake_rollout;
+use app_test_support::create_mock_responses_server_repeating_assistant;
+use app_test_support::to_response;
+use codex_app_server_protocol::JSONRPCNotification;
+use codex_app_server_protocol::JSONRPCResponse;
+use codex_app_server_protocol::RequestId;
+use codex_app_server_protocol::SessionSource;
+use codex_app_server_protocol::ThreadForkParams;
+use codex_app_server_protocol::ThreadForkResponse;
+use codex_app_server_protocol::ThreadItem;
+use codex_app_server_protocol::ThreadStartedNotification;
+use codex_app_server_protocol::TurnStatus;
+use codex_app_server_protocol::UserInput;
+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 thread_fork_creates_new_thread_and_emits_started() -> Result<()> {
+    let server = create_mock_responses_server_repeating_assistant("Done").await;
+    let codex_home = TempDir::new()?;
+    create_config_toml(codex_home.path(), &server.uri())?;
+
+    let preview = "Saved user message";
+    let conversation_id = create_fake_rollout(
+        codex_home.path(),
+        "2025-01-05T12-00-00",
+        "2025-01-05T12:00:00Z",
+        preview,
+        Some("mock_provider"),
+        None,
+    )?;
+
+    let original_path = codex_home
+        .path()
+        .join("sessions")
+        .join("2025")
+        .join("01")
+        .join("05")
+        .join(format!(
+            "rollout-2025-01-05T12-00-00-{conversation_id}.jsonl"
+        ));
+    assert!(
+        original_path.exists(),
+        "expected original rollout to exist at {}",
+        original_path.display()
+    );
+    let original_contents = std::fs::read_to_string(&original_path)?;
+
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
+
+    let fork_id = mcp
+        .send_thread_fork_request(ThreadForkParams {
+            thread_id: conversation_id.clone(),
+            ..Default::default()
+        })
+        .await?;
+    let fork_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(fork_id)),
+    )
+    .await??;
+    let ThreadForkResponse { thread, .. } = to_response::<ThreadForkResponse>(fork_resp)?;
+
+    let after_contents = std::fs::read_to_string(&original_path)?;
+    assert_eq!(
+        after_contents, original_contents,
+        "fork should not mutate the original rollout file"
+    );
+
+    assert_ne!(thread.id, conversation_id);
+    assert_eq!(thread.preview, preview);
+    assert_eq!(thread.model_provider, "mock_provider");
+    assert!(thread.path.is_absolute());
+    assert_ne!(thread.path, original_path);
+    assert!(thread.cwd.is_absolute());
+    assert_eq!(thread.source, SessionSource::VsCode);
+
+    assert_eq!(
+        thread.turns.len(),
+        1,
+        "expected forked thread to include one turn"
+    );
+    let turn = &thread.turns[0];
+    assert_eq!(turn.status, TurnStatus::Completed);
+    assert_eq!(turn.items.len(), 1, "expected user message item");
+    match &turn.items[0] {
+        ThreadItem::UserMessage { content, .. } => {
+            assert_eq!(
+                content,
+                &vec![UserInput::Text {
+                    text: preview.to_string()
+                }]
+            );
+        }
+        other => panic!("expected user message item, got {other:?}"),
+    }
+
+    // A corresponding thread/started notification should arrive.
+    let notif: JSONRPCNotification = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("thread/started"),
+    )
+    .await??;
+    let started: ThreadStartedNotification =
+        serde_json::from_value(notif.params.expect("params must be present"))?;
+    assert_eq!(started.thread, thread);
+
+    Ok(())
+}
+
+// Helper to create a config.toml pointing at the mock model server.
+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 = "read-only"
+
+model_provider = "mock_provider"
+
+[model_providers.mock_provider]
+name = "Mock provider for test"
+base_url = "{server_uri}/v1"
+wire_api = "responses"
+request_max_retries = 0
+stream_max_retries = 0
+"#
+        ),
+    )
+}

codex-rs/app-server/tests/suite/v2/thread_loaded_list.rs

@@ -0,0 +1,139 @@
+use anyhow::Result;
+use app_test_support::McpProcess;
+use app_test_support::create_mock_responses_server_repeating_assistant;
+use app_test_support::to_response;
+use codex_app_server_protocol::JSONRPCResponse;
+use codex_app_server_protocol::RequestId;
+use codex_app_server_protocol::ThreadLoadedListParams;
+use codex_app_server_protocol::ThreadLoadedListResponse;
+use codex_app_server_protocol::ThreadStartParams;
+use codex_app_server_protocol::ThreadStartResponse;
+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 thread_loaded_list_returns_loaded_thread_ids() -> Result<()> {
+    let server = create_mock_responses_server_repeating_assistant("Done").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 thread_id = start_thread(&mut mcp).await?;
+
+    let list_id = mcp
+        .send_thread_loaded_list_request(ThreadLoadedListParams::default())
+        .await?;
+    let resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(list_id)),
+    )
+    .await??;
+    let ThreadLoadedListResponse {
+        mut data,
+        next_cursor,
+    } = to_response::<ThreadLoadedListResponse>(resp)?;
+    data.sort();
+    assert_eq!(data, vec![thread_id]);
+    assert_eq!(next_cursor, None);
+
+    Ok(())
+}
+
+#[tokio::test]
+async fn thread_loaded_list_paginates() -> Result<()> {
+    let server = create_mock_responses_server_repeating_assistant("Done").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 first = start_thread(&mut mcp).await?;
+    let second = start_thread(&mut mcp).await?;
+
+    let mut expected = [first, second];
+    expected.sort();
+
+    let list_id = mcp
+        .send_thread_loaded_list_request(ThreadLoadedListParams {
+            cursor: None,
+            limit: Some(1),
+        })
+        .await?;
+    let resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(list_id)),
+    )
+    .await??;
+    let ThreadLoadedListResponse {
+        data: first_page,
+        next_cursor,
+    } = to_response::<ThreadLoadedListResponse>(resp)?;
+    assert_eq!(first_page, vec![expected[0].clone()]);
+    assert_eq!(next_cursor, Some(expected[0].clone()));
+
+    let list_id = mcp
+        .send_thread_loaded_list_request(ThreadLoadedListParams {
+            cursor: next_cursor,
+            limit: Some(1),
+        })
+        .await?;
+    let resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(list_id)),
+    )
+    .await??;
+    let ThreadLoadedListResponse {
+        data: second_page,
+        next_cursor,
+    } = to_response::<ThreadLoadedListResponse>(resp)?;
+    assert_eq!(second_page, vec![expected[1].clone()]);
+    assert_eq!(next_cursor, None);
+
+    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 = "never"
+sandbox_mode = "read-only"
+
+model_provider = "mock_provider"
+
+[model_providers.mock_provider]
+name = "Mock provider for test"
+base_url = "{server_uri}/v1"
+wire_api = "responses"
+request_max_retries = 0
+stream_max_retries = 0
+"#
+        ),
+    )
+}
+
+async fn start_thread(mcp: &mut McpProcess) -> Result<String> {
+    let req_id = mcp
+        .send_thread_start_request(ThreadStartParams {
+            model: Some("gpt-5.1".to_string()),
+            ..Default::default()
+        })
+        .await?;
+    let resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(req_id)),
+    )
+    .await??;
+    let ThreadStartResponse { thread, .. } = to_response::<ThreadStartResponse>(resp)?;
+    Ok(thread.id)
+}

codex-rs/app-server/tests/suite/v2/thread_resume.rs

@@ -1,7 +1,7 @@
 use anyhow::Result;
 use app_test_support::McpProcess;
 use app_test_support::create_fake_rollout;
-use app_test_support::create_mock_chat_completions_server;
+use app_test_support::create_mock_responses_server_repeating_assistant;
 use app_test_support::to_response;
 use codex_app_server_protocol::JSONRPCResponse;
 use codex_app_server_protocol::RequestId;
@@ -23,7 +23,7 @@ const DEFAULT_READ_TIMEOUT: std::time::Duration = std::time::Duration::from_secs
 
 #[tokio::test]
 async fn thread_resume_returns_original_thread() -> Result<()> {
-    let server = create_mock_chat_completions_server(vec![]).await;
+    let server = create_mock_responses_server_repeating_assistant("Done").await;
     let codex_home = TempDir::new()?;
     create_config_toml(codex_home.path(), &server.uri())?;
 
@@ -66,7 +66,7 @@ async fn thread_resume_returns_original_thread() -> Result<()> {
 
 #[tokio::test]
 async fn thread_resume_returns_rollout_history() -> Result<()> {
-    let server = create_mock_chat_completions_server(vec![]).await;
+    let server = create_mock_responses_server_repeating_assistant("Done").await;
     let codex_home = TempDir::new()?;
     create_config_toml(codex_home.path(), &server.uri())?;
 
@@ -130,7 +130,7 @@ async fn thread_resume_returns_rollout_history() -> Result<()> {
 
 #[tokio::test]
 async fn thread_resume_prefers_path_over_thread_id() -> Result<()> {
-    let server = create_mock_chat_completions_server(vec![]).await;
+    let server = create_mock_responses_server_repeating_assistant("Done").await;
     let codex_home = TempDir::new()?;
     create_config_toml(codex_home.path(), &server.uri())?;
 
@@ -174,7 +174,7 @@ async fn thread_resume_prefers_path_over_thread_id() -> Result<()> {
 
 #[tokio::test]
 async fn thread_resume_supports_history_and_overrides() -> Result<()> {
-    let server = create_mock_chat_completions_server(vec![]).await;
+    let server = create_mock_responses_server_repeating_assistant("Done").await;
     let codex_home = TempDir::new()?;
     create_config_toml(codex_home.path(), &server.uri())?;
 
@@ -247,7 +247,7 @@ model_provider = "mock_provider"
 [model_providers.mock_provider]
 name = "Mock provider for test"
 base_url = "{server_uri}/v1"
-wire_api = "chat"
+wire_api = "responses"
 request_max_retries = 0
 stream_max_retries = 0
 "#

codex-rs/app-server/tests/suite/v2/thread_rollback.rs

@@ -0,0 +1,177 @@
+use anyhow::Result;
+use app_test_support::McpProcess;
+use app_test_support::create_final_assistant_message_sse_response;
+use app_test_support::create_mock_responses_server_sequence_unchecked;
+use app_test_support::to_response;
+use codex_app_server_protocol::JSONRPCResponse;
+use codex_app_server_protocol::RequestId;
+use codex_app_server_protocol::ThreadItem;
+use codex_app_server_protocol::ThreadResumeParams;
+use codex_app_server_protocol::ThreadResumeResponse;
+use codex_app_server_protocol::ThreadRollbackParams;
+use codex_app_server_protocol::ThreadRollbackResponse;
+use codex_app_server_protocol::ThreadStartParams;
+use codex_app_server_protocol::ThreadStartResponse;
+use codex_app_server_protocol::TurnStartParams;
+use codex_app_server_protocol::UserInput as V2UserInput;
+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]
+async fn thread_rollback_drops_last_turns_and_persists_to_rollout() -> Result<()> {
+    // Three Codex turns hit the mock model (session start + two turn/start calls).
+    let responses = vec![
+        create_final_assistant_message_sse_response("Done")?,
+        create_final_assistant_message_sse_response("Done")?,
+        create_final_assistant_message_sse_response("Done")?,
+    ];
+    let server = create_mock_responses_server_sequence_unchecked(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??;
+
+    // Start a thread.
+    let start_id = mcp
+        .send_thread_start_request(ThreadStartParams {
+            model: Some("mock-model".to_string()),
+            ..Default::default()
+        })
+        .await?;
+    let start_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(start_id)),
+    )
+    .await??;
+    let ThreadStartResponse { thread, .. } = to_response::<ThreadStartResponse>(start_resp)?;
+
+    // Two turns.
+    let first_text = "First";
+    let turn1_id = mcp
+        .send_turn_start_request(TurnStartParams {
+            thread_id: thread.id.clone(),
+            input: vec![V2UserInput::Text {
+                text: first_text.to_string(),
+            }],
+            ..Default::default()
+        })
+        .await?;
+    let _turn1_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(turn1_id)),
+    )
+    .await??;
+    let _completed1 = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("turn/completed"),
+    )
+    .await??;
+
+    let turn2_id = mcp
+        .send_turn_start_request(TurnStartParams {
+            thread_id: thread.id.clone(),
+            input: vec![V2UserInput::Text {
+                text: "Second".to_string(),
+            }],
+            ..Default::default()
+        })
+        .await?;
+    let _turn2_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(turn2_id)),
+    )
+    .await??;
+    let _completed2 = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("turn/completed"),
+    )
+    .await??;
+
+    // Roll back the last turn.
+    let rollback_id = mcp
+        .send_thread_rollback_request(ThreadRollbackParams {
+            thread_id: thread.id.clone(),
+            num_turns: 1,
+        })
+        .await?;
+    let rollback_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(rollback_id)),
+    )
+    .await??;
+    let ThreadRollbackResponse {
+        thread: rolled_back_thread,
+    } = to_response::<ThreadRollbackResponse>(rollback_resp)?;
+
+    assert_eq!(rolled_back_thread.turns.len(), 1);
+    assert_eq!(rolled_back_thread.turns[0].items.len(), 2);
+    match &rolled_back_thread.turns[0].items[0] {
+        ThreadItem::UserMessage { content, .. } => {
+            assert_eq!(
+                content,
+                &vec![V2UserInput::Text {
+                    text: first_text.to_string()
+                }]
+            );
+        }
+        other => panic!("expected user message item, got {other:?}"),
+    }
+
+    // Resume and confirm the history is pruned.
+    let resume_id = mcp
+        .send_thread_resume_request(ThreadResumeParams {
+            thread_id: thread.id,
+            ..Default::default()
+        })
+        .await?;
+    let resume_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(resume_id)),
+    )
+    .await??;
+    let ThreadResumeResponse { thread, .. } = to_response::<ThreadResumeResponse>(resume_resp)?;
+
+    assert_eq!(thread.turns.len(), 1);
+    assert_eq!(thread.turns[0].items.len(), 2);
+    match &thread.turns[0].items[0] {
+        ThreadItem::UserMessage { content, .. } => {
+            assert_eq!(
+                content,
+                &vec![V2UserInput::Text {
+                    text: first_text.to_string()
+                }]
+            );
+        }
+        other => panic!("expected user message item, got {other:?}"),
+    }
+
+    Ok(())
+}
+
+fn create_config_toml(codex_home: &std::path::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 = "read-only"
+
+model_provider = "mock_provider"
+
+[model_providers.mock_provider]
+name = "Mock provider for test"
+base_url = "{server_uri}/v1"
+wire_api = "responses"
+request_max_retries = 0
+stream_max_retries = 0
+"#
+        ),
+    )
+}

codex-rs/app-server/tests/suite/v2/thread_start.rs

@@ -1,6 +1,6 @@
 use anyhow::Result;
 use app_test_support::McpProcess;
-use app_test_support::create_mock_chat_completions_server;
+use app_test_support::create_mock_responses_server_repeating_assistant;
 use app_test_support::to_response;
 use codex_app_server_protocol::JSONRPCNotification;
 use codex_app_server_protocol::JSONRPCResponse;
@@ -17,7 +17,7 @@ const DEFAULT_READ_TIMEOUT: std::time::Duration = std::time::Duration::from_secs
 #[tokio::test]
 async fn thread_start_creates_thread_and_emits_started() -> Result<()> {
     // Provide a mock server and config so model wiring is valid.
-    let server = create_mock_chat_completions_server(vec![]).await;
+    let server = create_mock_responses_server_repeating_assistant("Done").await;
 
     let codex_home = TempDir::new()?;
     create_config_toml(codex_home.path(), &server.uri())?;
@@ -85,7 +85,7 @@ model_provider = "mock_provider"
 [model_providers.mock_provider]
 name = "Mock provider for test"
 base_url = "{server_uri}/v1"
-wire_api = "chat"
+wire_api = "responses"
 request_max_retries = 0
 stream_max_retries = 0
 "#

codex-rs/app-server/tests/suite/v2/turn_interrupt.rs

@@ -2,7 +2,7 @@
 
 use anyhow::Result;
 use app_test_support::McpProcess;
-use app_test_support::create_mock_chat_completions_server;
+use app_test_support::create_mock_responses_server_sequence;
 use app_test_support::create_shell_command_sse_response;
 use app_test_support::to_response;
 use codex_app_server_protocol::JSONRPCNotification;
@@ -41,7 +41,7 @@ async fn turn_interrupt_aborts_running_turn() -> Result<()> {
     std::fs::create_dir(&working_directory)?;
 
     // Mock server: long-running shell command then (after abort) nothing else needed.
-    let server = create_mock_chat_completions_server(vec![create_shell_command_sse_response(
+    let server = create_mock_responses_server_sequence(vec![create_shell_command_sse_response(
         shell_command.clone(),
         Some(&working_directory),
         Some(10_000),
@@ -135,7 +135,7 @@ model_provider = "mock_provider"
 [model_providers.mock_provider]
 name = "Mock provider for test"
 base_url = "{server_uri}/v1"
-wire_api = "chat"
+wire_api = "responses"
 request_max_retries = 0
 stream_max_retries = 0
 "#

codex-rs/app-server/tests/suite/v2/turn_start.rs

@@ -3,14 +3,16 @@ use app_test_support::McpProcess;
 use app_test_support::create_apply_patch_sse_response;
 use app_test_support::create_exec_command_sse_response;
 use app_test_support::create_final_assistant_message_sse_response;
-use app_test_support::create_mock_chat_completions_server;
-use app_test_support::create_mock_chat_completions_server_unchecked;
+use app_test_support::create_mock_responses_server_sequence;
+use app_test_support::create_mock_responses_server_sequence_unchecked;
 use app_test_support::create_shell_command_sse_response;
 use app_test_support::format_with_current_shell_display;
 use app_test_support::to_response;
-use codex_app_server_protocol::ApprovalDecision;
+use codex_app_server_protocol::ClientInfo;
+use codex_app_server_protocol::CommandExecutionApprovalDecision;
 use codex_app_server_protocol::CommandExecutionRequestApprovalResponse;
 use codex_app_server_protocol::CommandExecutionStatus;
+use codex_app_server_protocol::FileChangeApprovalDecision;
 use codex_app_server_protocol::FileChangeOutputDeltaNotification;
 use codex_app_server_protocol::FileChangeRequestApprovalResponse;
 use codex_app_server_protocol::ItemCompletedNotification;
@@ -39,6 +41,76 @@ use tempfile::TempDir;
 use tokio::time::timeout;
 
 const DEFAULT_READ_TIMEOUT: std::time::Duration = std::time::Duration::from_secs(10);
+const TEST_ORIGINATOR: &str = "codex_vscode";
+
+#[tokio::test]
+async fn turn_start_sends_originator_header() -> Result<()> {
+    let responses = vec![create_final_assistant_message_sse_response("Done")?];
+    let server = create_mock_responses_server_sequence_unchecked(responses).await;
+
+    let codex_home = TempDir::new()?;
+    create_config_toml(codex_home.path(), &server.uri(), "never")?;
+
+    let mut mcp = McpProcess::new(codex_home.path()).await?;
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.initialize_with_client_info(ClientInfo {
+            name: TEST_ORIGINATOR.to_string(),
+            title: Some("Codex VS Code Extension".to_string()),
+            version: "0.1.0".to_string(),
+        }),
+    )
+    .await??;
+
+    let thread_req = mcp
+        .send_thread_start_request(ThreadStartParams {
+            model: Some("mock-model".to_string()),
+            ..Default::default()
+        })
+        .await?;
+    let thread_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(thread_req)),
+    )
+    .await??;
+    let ThreadStartResponse { thread, .. } = to_response::<ThreadStartResponse>(thread_resp)?;
+
+    let turn_req = mcp
+        .send_turn_start_request(TurnStartParams {
+            thread_id: thread.id.clone(),
+            input: vec![V2UserInput::Text {
+                text: "Hello".to_string(),
+            }],
+            ..Default::default()
+        })
+        .await?;
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(turn_req)),
+    )
+    .await??;
+
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("turn/completed"),
+    )
+    .await??;
+
+    let requests = server
+        .received_requests()
+        .await
+        .expect("failed to fetch received requests");
+    assert!(!requests.is_empty());
+    for request in requests {
+        let originator = request
+            .headers
+            .get("originator")
+            .expect("originator header missing");
+        assert_eq!(originator.to_str()?, TEST_ORIGINATOR);
+    }
+
+    Ok(())
+}
 
 #[tokio::test]
 async fn turn_start_emits_notifications_and_accepts_model_override() -> Result<()> {
@@ -49,7 +121,7 @@ async fn turn_start_emits_notifications_and_accepts_model_override() -> Result<(
         create_final_assistant_message_sse_response("Done")?,
         create_final_assistant_message_sse_response("Done")?,
     ];
-    let server = create_mock_chat_completions_server_unchecked(responses).await;
+    let server = create_mock_responses_server_sequence_unchecked(responses).await;
 
     let codex_home = TempDir::new()?;
     create_config_toml(codex_home.path(), &server.uri(), "never")?;
@@ -156,7 +228,7 @@ async fn turn_start_accepts_local_image_input() -> Result<()> {
     ];
     // Use the unchecked variant because the request payload includes a LocalImage
     // which the strict matcher does not currently cover.
-    let server = create_mock_chat_completions_server_unchecked(responses).await;
+    let server = create_mock_responses_server_sequence_unchecked(responses).await;
 
     let codex_home = TempDir::new()?;
     create_config_toml(codex_home.path(), &server.uri(), "never")?;
@@ -232,7 +304,7 @@ async fn turn_start_exec_approval_toggle_v2() -> Result<()> {
         )?,
         create_final_assistant_message_sse_response("done 2")?,
     ];
-    let server = create_mock_chat_completions_server(responses).await;
+    let server = create_mock_responses_server_sequence(responses).await;
     // Default approval is untrusted to force elicitation on first turn.
     create_config_toml(codex_home.as_path(), &server.uri(), "untrusted")?;
 
@@ -356,7 +428,7 @@ async fn turn_start_exec_approval_decline_v2() -> Result<()> {
         )?,
         create_final_assistant_message_sse_response("done")?,
     ];
-    let server = create_mock_chat_completions_server(responses).await;
+    let server = create_mock_responses_server_sequence(responses).await;
     create_config_toml(codex_home.as_path(), &server.uri(), "untrusted")?;
 
     let mut mcp = McpProcess::new(codex_home.as_path()).await?;
@@ -426,7 +498,7 @@ async fn turn_start_exec_approval_decline_v2() -> Result<()> {
     mcp.send_response(
         request_id,
         serde_json::to_value(CommandExecutionRequestApprovalResponse {
-            decision: ApprovalDecision::Decline,
+            decision: CommandExecutionApprovalDecision::Decline,
         })?,
     )
     .await?;
@@ -502,7 +574,7 @@ async fn turn_start_updates_sandbox_and_cwd_between_turns_v2() -> Result<()> {
         )?,
         create_final_assistant_message_sse_response("done second")?,
     ];
-    let server = create_mock_chat_completions_server(responses).await;
+    let server = create_mock_responses_server_sequence(responses).await;
     create_config_toml(&codex_home, &server.uri(), "untrusted")?;
 
     let mut mcp = McpProcess::new(&codex_home).await?;
@@ -540,6 +612,7 @@ async fn turn_start_updates_sandbox_and_cwd_between_turns_v2() -> Result<()> {
             model: Some("mock-model".to_string()),
             effort: Some(ReasoningEffort::Medium),
             summary: Some(ReasoningSummary::Auto),
+            output_schema: None,
         })
         .await?;
     timeout(
@@ -552,6 +625,7 @@ async fn turn_start_updates_sandbox_and_cwd_between_turns_v2() -> Result<()> {
         mcp.read_stream_until_notification_message("codex/event/task_complete"),
     )
     .await??;
+    mcp.clear_message_buffer();
 
     // second turn with workspace-write and second_cwd, ensure exec begins in second_cwd
     let second_turn = mcp
@@ -566,6 +640,7 @@ async fn turn_start_updates_sandbox_and_cwd_between_turns_v2() -> Result<()> {
             model: Some("mock-model".to_string()),
             effort: Some(ReasoningEffort::Medium),
             summary: Some(ReasoningSummary::Auto),
+            output_schema: None,
         })
         .await?;
     timeout(
@@ -633,7 +708,7 @@ async fn turn_start_file_change_approval_v2() -> Result<()> {
         create_apply_patch_sse_response(patch, "patch-call")?,
         create_final_assistant_message_sse_response("patch applied")?,
     ];
-    let server = create_mock_chat_completions_server(responses).await;
+    let server = create_mock_responses_server_sequence(responses).await;
     create_config_toml(&codex_home, &server.uri(), "untrusted")?;
 
     let mut mcp = McpProcess::new(&codex_home).await?;
@@ -720,7 +795,7 @@ async fn turn_start_file_change_approval_v2() -> Result<()> {
     mcp.send_response(
         request_id,
         serde_json::to_value(FileChangeRequestApprovalResponse {
-            decision: ApprovalDecision::Accept,
+            decision: FileChangeApprovalDecision::Accept,
         })?,
     )
     .await?;
@@ -780,6 +855,190 @@ async fn turn_start_file_change_approval_v2() -> Result<()> {
     Ok(())
 }
 
+#[tokio::test]
+async fn turn_start_file_change_approval_accept_for_session_persists_v2() -> Result<()> {
+    skip_if_no_network!(Ok(()));
+
+    let tmp = TempDir::new()?;
+    let codex_home = tmp.path().join("codex_home");
+    std::fs::create_dir(&codex_home)?;
+    let workspace = tmp.path().join("workspace");
+    std::fs::create_dir(&workspace)?;
+
+    let patch_1 = r#"*** Begin Patch
+*** Add File: README.md
++new line
+*** End Patch
+"#;
+    let patch_2 = r#"*** Begin Patch
+*** Update File: README.md
+@@
+-new line
++updated line
+*** End Patch
+"#;
+
+    let responses = vec![
+        create_apply_patch_sse_response(patch_1, "patch-call-1")?,
+        create_final_assistant_message_sse_response("patch 1 applied")?,
+        create_apply_patch_sse_response(patch_2, "patch-call-2")?,
+        create_final_assistant_message_sse_response("patch 2 applied")?,
+    ];
+    let server = create_mock_responses_server_sequence(responses).await;
+    create_config_toml(&codex_home, &server.uri(), "untrusted")?;
+
+    let mut mcp = McpProcess::new(&codex_home).await?;
+    timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
+
+    let start_req = mcp
+        .send_thread_start_request(ThreadStartParams {
+            model: Some("mock-model".to_string()),
+            cwd: Some(workspace.to_string_lossy().into_owned()),
+            ..Default::default()
+        })
+        .await?;
+    let start_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(start_req)),
+    )
+    .await??;
+    let ThreadStartResponse { thread, .. } = to_response::<ThreadStartResponse>(start_resp)?;
+
+    // First turn: expect FileChangeRequestApproval, respond with AcceptForSession, and verify the file exists.
+    let turn_1_req = mcp
+        .send_turn_start_request(TurnStartParams {
+            thread_id: thread.id.clone(),
+            input: vec![V2UserInput::Text {
+                text: "apply patch 1".into(),
+            }],
+            cwd: Some(workspace.clone()),
+            ..Default::default()
+        })
+        .await?;
+    let turn_1_resp: JSONRPCResponse = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(turn_1_req)),
+    )
+    .await??;
+    let TurnStartResponse { turn: turn_1 } = to_response::<TurnStartResponse>(turn_1_resp)?;
+
+    let started_file_change_1 = timeout(DEFAULT_READ_TIMEOUT, async {
+        loop {
+            let started_notif = mcp
+                .read_stream_until_notification_message("item/started")
+                .await?;
+            let started: ItemStartedNotification =
+                serde_json::from_value(started_notif.params.clone().expect("item/started params"))?;
+            if let ThreadItem::FileChange { .. } = started.item {
+                return Ok::<ThreadItem, anyhow::Error>(started.item);
+            }
+        }
+    })
+    .await??;
+    let ThreadItem::FileChange { id, status, .. } = started_file_change_1 else {
+        unreachable!("loop ensures we break on file change items");
+    };
+    assert_eq!(id, "patch-call-1");
+    assert_eq!(status, PatchApplyStatus::InProgress);
+
+    let server_req = timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_request_message(),
+    )
+    .await??;
+    let ServerRequest::FileChangeRequestApproval { request_id, params } = server_req else {
+        panic!("expected FileChangeRequestApproval request")
+    };
+    assert_eq!(params.item_id, "patch-call-1");
+    assert_eq!(params.thread_id, thread.id);
+    assert_eq!(params.turn_id, turn_1.id);
+
+    mcp.send_response(
+        request_id,
+        serde_json::to_value(FileChangeRequestApprovalResponse {
+            decision: FileChangeApprovalDecision::AcceptForSession,
+        })?,
+    )
+    .await?;
+
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("item/fileChange/outputDelta"),
+    )
+    .await??;
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("item/completed"),
+    )
+    .await??;
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("codex/event/task_complete"),
+    )
+    .await??;
+
+    let readme_path = workspace.join("README.md");
+    assert_eq!(std::fs::read_to_string(&readme_path)?, "new line\n");
+
+    // Second turn: apply a patch to the same file. Approval should be skipped due to AcceptForSession.
+    let turn_2_req = mcp
+        .send_turn_start_request(TurnStartParams {
+            thread_id: thread.id.clone(),
+            input: vec![V2UserInput::Text {
+                text: "apply patch 2".into(),
+            }],
+            cwd: Some(workspace.clone()),
+            ..Default::default()
+        })
+        .await?;
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_response_message(RequestId::Integer(turn_2_req)),
+    )
+    .await??;
+
+    let started_file_change_2 = timeout(DEFAULT_READ_TIMEOUT, async {
+        loop {
+            let started_notif = mcp
+                .read_stream_until_notification_message("item/started")
+                .await?;
+            let started: ItemStartedNotification =
+                serde_json::from_value(started_notif.params.clone().expect("item/started params"))?;
+            if let ThreadItem::FileChange { .. } = started.item {
+                return Ok::<ThreadItem, anyhow::Error>(started.item);
+            }
+        }
+    })
+    .await??;
+    let ThreadItem::FileChange { id, status, .. } = started_file_change_2 else {
+        unreachable!("loop ensures we break on file change items");
+    };
+    assert_eq!(id, "patch-call-2");
+    assert_eq!(status, PatchApplyStatus::InProgress);
+
+    // If the server incorrectly emits FileChangeRequestApproval, the helper below will error
+    // (it bails on unexpected JSONRPCMessage::Request), causing the test to fail.
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("item/fileChange/outputDelta"),
+    )
+    .await??;
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("item/completed"),
+    )
+    .await??;
+    timeout(
+        DEFAULT_READ_TIMEOUT,
+        mcp.read_stream_until_notification_message("codex/event/task_complete"),
+    )
+    .await??;
+
+    assert_eq!(std::fs::read_to_string(readme_path)?, "updated line\n");
+
+    Ok(())
+}
+
 #[tokio::test]
 async fn turn_start_file_change_approval_decline_v2() -> Result<()> {
     skip_if_no_network!(Ok(()));
@@ -799,7 +1058,7 @@ async fn turn_start_file_change_approval_decline_v2() -> Result<()> {
         create_apply_patch_sse_response(patch, "patch-call")?,
         create_final_assistant_message_sse_response("patch declined")?,
     ];
-    let server = create_mock_chat_completions_server(responses).await;
+    let server = create_mock_responses_server_sequence(responses).await;
     create_config_toml(&codex_home, &server.uri(), "untrusted")?;
 
     let mut mcp = McpProcess::new(&codex_home).await?;
@@ -886,7 +1145,7 @@ async fn turn_start_file_change_approval_decline_v2() -> Result<()> {
     mcp.send_response(
         request_id,
         serde_json::to_value(FileChangeRequestApprovalResponse {
-            decision: ApprovalDecision::Decline,
+            decision: FileChangeApprovalDecision::Decline,
         })?,
     )
     .await?;
@@ -937,7 +1196,7 @@ async fn command_execution_notifications_include_process_id() -> Result<()> {
         create_exec_command_sse_response("uexec-1")?,
         create_final_assistant_message_sse_response("done")?,
     ];
-    let server = create_mock_chat_completions_server(responses).await;
+    let server = create_mock_responses_server_sequence(responses).await;
     let codex_home = TempDir::new()?;
     create_config_toml(codex_home.path(), &server.uri(), "never")?;
     let config_toml = codex_home.path().join("config.toml");
@@ -1076,7 +1335,7 @@ model_provider = "mock_provider"
 [model_providers.mock_provider]
 name = "Mock provider for test"
 base_url = "{server_uri}/v1"
-wire_api = "chat"
+wire_api = "responses"
 request_max_retries = 0
 stream_max_retries = 0
 "#

codex-rs/apply-patch/BUILD.bazel

@@ -0,0 +1,11 @@
+load("//:defs.bzl", "codex_rust_crate")
+
+exports_files(["apply_patch_tool_instructions.md"])
+
+codex_rust_crate(
+    name = "apply-patch",
+    crate_name = "codex_apply_patch",
+    compile_data = [
+        "apply_patch_tool_instructions.md",
+    ],
+)

codex-rs/apply-patch/Cargo.toml

@@ -25,5 +25,6 @@ tree-sitter-bash = { workspace = true }
 [dev-dependencies]
 assert_cmd = { workspace = true }
 assert_matches = { workspace = true }
+codex-utils-cargo-bin = { workspace = true }
 pretty_assertions = { workspace = true }
 tempfile = { workspace = true }

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

@@ -227,11 +227,14 @@ fn check_start_and_end_lines_strict(
     first_line: Option<&&str>,
     last_line: Option<&&str>,
 ) -> Result<(), ParseError> {
+    let first_line = first_line.map(|line| line.trim());
+    let last_line = last_line.map(|line| line.trim());
+
     match (first_line, last_line) {
-        (Some(&first), Some(&last)) if first == BEGIN_PATCH_MARKER && last == END_PATCH_MARKER => {
+        (Some(first), Some(last)) if first == BEGIN_PATCH_MARKER && last == END_PATCH_MARKER => {
             Ok(())
         }
-        (Some(&first), _) if first != BEGIN_PATCH_MARKER => Err(InvalidPatchError(String::from(
+        (Some(first), _) if first != BEGIN_PATCH_MARKER => Err(InvalidPatchError(String::from(
             "The first line of the patch must be '*** Begin Patch'",
         ))),
         _ => Err(InvalidPatchError(String::from(
@@ -444,6 +447,25 @@ fn test_parse_patch() {
             "The last line of the patch must be '*** End Patch'".to_string()
         ))
     );
+
+    assert_eq!(
+        parse_patch_text(
+            concat!(
+                "*** Begin Patch",
+                " ",
+                "\n*** Add File: foo\n+hi\n",
+                " ",
+                "*** End Patch"
+            ),
+            ParseMode::Strict
+        )
+        .unwrap()
+        .hunks,
+        vec![AddFile {
+            path: PathBuf::from("foo"),
+            contents: "hi\n".to_string()
+        }]
+    );
     assert_eq!(
         parse_patch_text(
             "*** Begin Patch\n\

codex-rs/apply-patch/tests/fixtures/scenarios/020_delete_file_success/expected/keep.txt

@@ -0,0 +1 @@
+keep

codex-rs/apply-patch/tests/fixtures/scenarios/020_delete_file_success/input/keep.txt

@@ -0,0 +1 @@
+keep

codex-rs/apply-patch/tests/fixtures/scenarios/020_delete_file_success/input/obsolete.txt

@@ -0,0 +1 @@
+obsolete

codex-rs/apply-patch/tests/fixtures/scenarios/020_delete_file_success/patch.txt

@@ -0,0 +1,3 @@
+*** Begin Patch
+*** Delete File: obsolete.txt
+*** End Patch

codex-rs/apply-patch/tests/fixtures/scenarios/020_whitespace_padded_patch_marker_lines/expected/file.txt

@@ -0,0 +1 @@
+two

codex-rs/apply-patch/tests/fixtures/scenarios/020_whitespace_padded_patch_marker_lines/input/file.txt

@@ -0,0 +1 @@
+one

codex-rs/apply-patch/tests/fixtures/scenarios/020_whitespace_padded_patch_marker_lines/patch.txt

@@ -0,0 +1,6 @@
+*** Begin Patch 
+*** Update File: file.txt
+@@
+-one
++two
+ *** End Patch

codex-rs/apply-patch/tests/fixtures/scenarios/021_update_file_deletion_only/expected/lines.txt

@@ -0,0 +1,2 @@
+line1
+line3

codex-rs/apply-patch/tests/fixtures/scenarios/021_update_file_deletion_only/input/lines.txt

@@ -0,0 +1,3 @@
+line1
+line2
+line3

codex-rs/apply-patch/tests/fixtures/scenarios/021_update_file_deletion_only/patch.txt

@@ -0,0 +1,7 @@
+*** Begin Patch
+*** Update File: lines.txt
+@@
+ line1
+-line2
+ line3
+*** End Patch

codex-rs/apply-patch/tests/fixtures/scenarios/022_update_file_end_of_file_marker/expected/tail.txt

@@ -0,0 +1,2 @@
+first
+second updated

codex-rs/apply-patch/tests/fixtures/scenarios/022_update_file_end_of_file_marker/input/tail.txt

@@ -0,0 +1,2 @@
+first
+second

codex-rs/apply-patch/tests/fixtures/scenarios/022_update_file_end_of_file_marker/patch.txt

@@ -0,0 +1,8 @@
+*** Begin Patch
+*** Update File: tail.txt
+@@
+ first
+-second
++second updated
+*** End of File
+*** End Patch

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

@@ -1,8 +1,13 @@
-use assert_cmd::prelude::*;
+use assert_cmd::Command;
 use std::fs;
-use std::process::Command;
 use tempfile::tempdir;
 
+fn apply_patch_command() -> anyhow::Result<Command> {
+    Ok(Command::new(codex_utils_cargo_bin::cargo_bin(
+        "apply_patch",
+    )?))
+}
+
 #[test]
 fn test_apply_patch_cli_add_and_update() -> anyhow::Result<()> {
     let tmp = tempdir()?;
@@ -16,8 +21,7 @@ fn test_apply_patch_cli_add_and_update() -> anyhow::Result<()> {
 +hello
 *** End Patch"#
     );
-    Command::cargo_bin("apply_patch")
-        .expect("should find apply_patch binary")
+    apply_patch_command()?
         .arg(add_patch)
         .current_dir(tmp.path())
         .assert()
@@ -34,8 +38,7 @@ fn test_apply_patch_cli_add_and_update() -> anyhow::Result<()> {
 +world
 *** End Patch"#
     );
-    Command::cargo_bin("apply_patch")
-        .expect("should find apply_patch binary")
+    apply_patch_command()?
         .arg(update_patch)
         .current_dir(tmp.path())
         .assert()
@@ -59,10 +62,9 @@ fn test_apply_patch_cli_stdin_add_and_update() -> anyhow::Result<()> {
 +hello
 *** End Patch"#
     );
-    let mut cmd =
-        assert_cmd::Command::cargo_bin("apply_patch").expect("should find apply_patch binary");
-    cmd.current_dir(tmp.path());
-    cmd.write_stdin(add_patch)
+    apply_patch_command()?
+        .current_dir(tmp.path())
+        .write_stdin(add_patch)
         .assert()
         .success()
         .stdout(format!("Success. Updated the following files:\nA {file}\n"));
@@ -77,10 +79,9 @@ fn test_apply_patch_cli_stdin_add_and_update() -> anyhow::Result<()> {
 +world
 *** End Patch"#
     );
-    let mut cmd =
-        assert_cmd::Command::cargo_bin("apply_patch").expect("should find apply_patch binary");
-    cmd.current_dir(tmp.path());
-    cmd.write_stdin(update_patch)
+    apply_patch_command()?
+        .current_dir(tmp.path())
+        .write_stdin(update_patch)
         .assert()
         .success()
         .stdout(format!("Success. Updated the following files:\nM {file}\n"));

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

@@ -1,4 +1,4 @@
-use assert_cmd::prelude::*;
+use codex_utils_cargo_bin::find_resource;
 use pretty_assertions::assert_eq;
 use std::collections::BTreeMap;
 use std::fs;
@@ -9,7 +9,8 @@ use tempfile::tempdir;
 
 #[test]
 fn test_apply_patch_scenarios() -> anyhow::Result<()> {
-    for scenario in fs::read_dir("tests/fixtures/scenarios")? {
+    let scenarios_dir = find_resource!("tests/fixtures/scenarios")?;
+    for scenario in fs::read_dir(scenarios_dir)? {
         let scenario = scenario?;
         let path = scenario.path();
         if path.is_dir() {
@@ -36,7 +37,7 @@ fn run_apply_patch_scenario(dir: &Path) -> anyhow::Result<()> {
     // Run apply_patch in the temporary directory. We intentionally do not assert
     // on the exit status here; the scenarios are specified purely in terms of
     // final filesystem state, which we compare below.
-    Command::cargo_bin("apply_patch")?
+    Command::new(codex_utils_cargo_bin::cargo_bin("apply_patch")?)
         .arg(patch)
         .current_dir(tmp.path())
         .output()?;
@@ -82,11 +83,15 @@ fn snapshot_dir_recursive(
             continue;
         };
         let rel = stripped.to_path_buf();
-        let file_type = entry.file_type()?;
-        if file_type.is_dir() {
+
+        // Under Buck2, files in `__srcs` are often materialized as symlinks.
+        // Use `metadata()` (follows symlinks) so our fixture snapshots work
+        // under both Cargo and Buck2.
+        let metadata = fs::metadata(&path)?;
+        if metadata.is_dir() {
             entries.insert(rel.clone(), Entry::Dir);
             snapshot_dir_recursive(base, &path, entries)?;
-        } else if file_type.is_file() {
+        } else if metadata.is_file() {
             let contents = fs::read(&path)?;
             entries.insert(rel, Entry::File(contents));
         }
@@ -98,12 +103,14 @@ fn copy_dir_recursive(src: &Path, dst: &Path) -> anyhow::Result<()> {
     for entry in fs::read_dir(src)? {
         let entry = entry?;
         let path = entry.path();
-        let file_type = entry.file_type()?;
         let dest_path = dst.join(entry.file_name());
-        if file_type.is_dir() {
+
+        // See note in `snapshot_dir_recursive` about Buck2 symlink trees.
+        let metadata = fs::metadata(&path)?;
+        if metadata.is_dir() {
             fs::create_dir_all(&dest_path)?;
             copy_dir_recursive(&path, &dest_path)?;
-        } else if file_type.is_file() {
+        } else if metadata.is_file() {
             if let Some(parent) = dest_path.parent() {
                 fs::create_dir_all(parent)?;
             }