Compare commits

...

77 Commits

Author SHA1 Message Date
jif-oai
644cb5fd6c format 2025-12-03 17:02:45 +00:00
jif-oai
bec31716ff Fix merge 2025-12-03 16:59:47 +00:00
jif-oai
c51d88da6b Merge remote-tracking branch 'origin/main' into jif/fork-save
# Conflicts:
#	codex-rs/app-server/tests/common/rollout.rs
#	codex-rs/core/src/codex.rs
#	codex-rs/core/src/conversation_manager.rs
#	codex-rs/core/src/rollout/tests.rs
#	codex-rs/mcp-server/src/codex_tool_runner.rs
2025-12-03 16:47:40 +00:00
jif-oai
45f3250eec feat: codex tool tips (#7440)
<img width="551" height="316" alt="Screenshot 2025-12-01 at 12 22 26"
src="https://github.com/user-attachments/assets/6ca3deff-8ef8-4f74-a8e1-e5ea13fd6740"
/>
2025-12-03 16:29:13 +00:00
jif-oai
51307eaf07 feat: retroactive image placeholder to prevent poisoning (#6774)
If an image can't be read by the API, it will poison the entire history,
preventing any new turn on the conversation.
This detect such cases and replace the image by a placeholder
2025-12-03 11:35:56 +00:00
jif-oai
42ae738f67 feat: model warning in case of apply patch (#7494) 2025-12-03 09:07:31 +00:00
Dylan Hurd
00ef9d3784 fix(tui) Support image paste from clipboard on native Windows (#7514)
Closes #3404 

## Summary
On windows, ctrl+v does not work for the same reason that cmd+v does not
work on macos. This PR adds alt/option+v detection, which allows windows
users to paste images from the clipboard using.

We could swap between just ctrl on mac and just alt on windows, but this
felt simpler - I don't feel strongly about it.

Note that this will NOT address image pasting in WSL environments, due
to issues with WSL <> Windows clipboards. I'm planning to address that
in a separate PR since it will likely warrant some discussion.

## Testing
- [x] Tested locally on a Mac and Windows laptop
2025-12-02 22:12:49 -08:00
Robby He
f3989f6092 fix(unified_exec): use platform default shell when unified_exec shell… (#7486)
# Unified Exec Shell Selection on Windows

## Problem

reference issue #7466

The `unified_exec` handler currently deserializes model-provided tool
calls into the `ExecCommandArgs` struct:

```rust
#[derive(Debug, Deserialize)]
struct ExecCommandArgs {
    cmd: String,
    #[serde(default)]
    workdir: Option<String>,
    #[serde(default = "default_shell")]
    shell: String,
    #[serde(default = "default_login")]
    login: bool,
    #[serde(default = "default_exec_yield_time_ms")]
    yield_time_ms: u64,
    #[serde(default)]
    max_output_tokens: Option<usize>,
    #[serde(default)]
    with_escalated_permissions: Option<bool>,
    #[serde(default)]
    justification: Option<String>,
}
```

The `shell` field uses a hard-coded default:

```rust
fn default_shell() -> String {
    "/bin/bash".to_string()
}
```

When the model returns a tool call JSON that only contains `cmd` (which
is the common case), Serde fills in `shell` with this default value.
Later, `get_command` uses that value as if it were a model-provided
shell path:

```rust
fn get_command(args: &ExecCommandArgs) -> Vec<String> {
    let shell = get_shell_by_model_provided_path(&PathBuf::from(args.shell.clone()));
    shell.derive_exec_args(&args.cmd, args.login)
}
```

On Unix, this usually resolves to `/bin/bash` and works as expected.
However, on Windows this behavior is problematic:

- The hard-coded `"/bin/bash"` is not a valid Windows path.
- `get_shell_by_model_provided_path` treats this as a model-specified
shell, and tries to resolve it (e.g. via `which::which("bash")`), which
may or may not exist and may not behave as intended.
- In practice, this leads to commands being executed under a non-default
or non-existent shell on Windows (for example, WSL bash), instead of the
expected Windows PowerShell or `cmd.exe`.

The core of the issue is that **"model did not specify `shell`" is
currently interpreted as "the model explicitly requested `/bin/bash`"**,
which is both Unix-specific and wrong on Windows.

## Proposed Solution

Instead of hard-coding `"/bin/bash"` into `ExecCommandArgs`, we should
distinguish between:

1. **The model explicitly specifying a shell**, e.g.:

   ```json
   {
     "cmd": "echo hello",
     "shell": "pwsh"
   }
   ```

In this case, we *do* want to respect the model’s choice and use
`get_shell_by_model_provided_path`.

2. **The model omitting the `shell` field entirely**, e.g.:

   ```json
   {
     "cmd": "echo hello"
   }
   ```

In this case, we should *not* assume `/bin/bash`. Instead, we should use
`default_user_shell()` and let the platform decide.

To express this distinction, we can:

1. Change `shell` to be optional in `ExecCommandArgs`:

   ```rust
   #[derive(Debug, Deserialize)]
   struct ExecCommandArgs {
       cmd: String,
       #[serde(default)]
       workdir: Option<String>,
       #[serde(default)]
       shell: Option<String>,
       #[serde(default = "default_login")]
       login: bool,
       #[serde(default = "default_exec_yield_time_ms")]
       yield_time_ms: u64,
       #[serde(default)]
       max_output_tokens: Option<usize>,
       #[serde(default)]
       with_escalated_permissions: Option<bool>,
       #[serde(default)]
       justification: Option<String>,
   }
   ```

Here, the absence of `shell` in the JSON is represented as `shell:
None`, rather than a hard-coded string value.
2025-12-02 21:49:25 -08:00
Matthew Zeng
dbec741ef0 Update device code auth strings. (#7498)
- [x] Update device code auth strings.
2025-12-02 17:36:38 -08:00
Michael Bolin
06e7667d0e fix: inline function marked as dead code (#7508)
I was debugging something else and noticed we could eliminate an
instance of `#[allow(dead_code)]` pretty easily.
2025-12-03 00:50:34 +00:00
Ahmed Ibrahim
1ef1fe67ec improve resume performance (#7303)
Reading the tail can be costly if we have a very big rollout item. we
can just read the file metadata
2025-12-02 16:39:40 -08:00
Michael Bolin
ee191dbe81 fix: path resolution bug in npx (#7134)
When running `npx @openai/codex-shell-tool-mcp`, the old code derived
`__dirname` from `process.argv[1]`, which points to npx’s transient
wrapper script in
`~/.npm/_npx/134d0fb7e1a27652/node_modules/.bin/codex-shell-tool-mcp`.
That made `vendorRoot` resolve to `<npx cache>/vendor`, so the startup
checks failed with "Required binary missing" because it looked for
`codex-execve-wrapper` in the wrong place.

By relying on the real module `__dirname` and `path.resolve(__dirname,
"..", "vendor")`, the package now anchors to its installed location
under `node_modules/@openai/codex-shell-tool-mcp/`, so the bundled
binaries are found and npx launches correctly.
2025-12-02 16:37:14 -08:00
Joshua Sutton
ad9eeeb287 Ensure duplicate-length paste placeholders stay distinct (#7431)
Fix issue #7430 
Generate unique numbered placeholders for multiple large pastes of the
same length so deleting one no longer removes the others.

Signed-off-by: Joshua <joshua1s@protonmail.com>
2025-12-02 16:16:01 -08:00
Michael Bolin
6b5b9a687e feat: support --version flag for @openai/codex-shell-tool-mcp (#7504)
I find it helpful to easily verify which version is running.

Tested:

```shell
~/code/codex3/codex-rs/exec-server$ cargo run --bin codex-exec-mcp-server -- --help
    Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.19s
     Running `/Users/mbolin/code/codex3/codex-rs/target/debug/codex-exec-mcp-server --help`
Usage: codex-exec-mcp-server [OPTIONS]

Options:
      --execve <EXECVE_WRAPPER>  Executable to delegate execve(2) calls to in Bash
      --bash <BASH_PATH>         Path to Bash that has been patched to support execve() wrapping
  -h, --help                     Print help
  -V, --version                  Print version
~/code/codex3/codex-rs/exec-server$ cargo run --bin codex-exec-mcp-server -- --version
    Finished `dev` profile [unoptimized + debuginfo] target(s) in 0.17s
     Running `/Users/mbolin/code/codex3/codex-rs/target/debug/codex-exec-mcp-server --version`
codex-exec-server 0.0.0
```
2025-12-02 23:43:25 +00:00
Josh McKinney
58e1e570fa refactor: tui.rs extract several pieces (#7461)
Pull FrameRequester out of tui.rs into its own module and make a
FrameScheduler struct. This is effectively an Actor/Handler approach
(see https://ryhl.io/blog/actors-with-tokio/). Adds tests and docs.

Small refactor of pending_viewport_area logic.
2025-12-02 15:19:27 -08:00
Michael Bolin
ec93b6daf3 chore: make create_approval_requirement_for_command an async fn (#7501)
I think this might help with https://github.com/openai/codex/pull/7033
because `create_approval_requirement_for_command()` will soon need
access to `Session.state`, which is a `tokio::sync::Mutex` that needs to
be accessed via `async`.
2025-12-02 15:01:15 -08:00
liam
4d4778ec1c Trim history.jsonl when history.max_bytes is set (#6242)
This PR honors the `history.max_bytes` configuration parameter by
trimming `history.jsonl` whenever it grows past the configured limit.
While appending new entries we retain the newest record, drop the oldest
lines to stay within the byte budget, and serialize the compacted file
back to disk under the same lock to keep writers safe.
2025-12-02 14:01:05 -08:00
Owen Lin
77c457121e fix: remove serde(flatten) annotation for TurnError (#7499)
The problem with using `serde(flatten)` on Turn status is that it
conditionally serializes the `error` field, which is not the pattern we
want in API v2 where all fields on an object should always be returned.

```
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
#[serde(rename_all = "camelCase")]
#[ts(export_to = "v2/")]
pub struct Turn {
    pub id: String,
    /// Only populated on a `thread/resume` response.
    /// For all other responses and notifications returning a Turn,
    /// the items field will be an empty list.
    pub items: Vec<ThreadItem>,
    #[serde(flatten)]
    pub status: TurnStatus,
}

#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
#[serde(tag = "status", rename_all = "camelCase")]
#[ts(tag = "status", export_to = "v2/")]
pub enum TurnStatus {
    Completed,
    Interrupted,
    Failed { error: TurnError },
    InProgress,
}
```

serializes to:
```
{
  "id": "turn-123",
  "items": [],
  "status": "completed"
}

{
  "id": "turn-123",
  "items": [],
  "status": "failed",
  "error": {
    "message": "Tool timeout",
    "codexErrorInfo": null
  }
}
```

Instead we want:
```
{
  "id": "turn-123",
  "items": [],
  "status": "completed",
  "error": null
}

{
  "id": "turn-123",
  "items": [],
  "status": "failed",
  "error": {
    "message": "Tool timeout",
    "codexErrorInfo": null
  }
}
```
2025-12-02 21:39:10 +00:00
zhao-oai
5ebdc9af1b persisting credits if new snapshot does not contain credit info (#7490)
in response to incoming changes to responses headers where the header
may sometimes not contain credits info (no longer forcing a credit
check)
2025-12-02 16:23:24 -05:00
Michael Bolin
f6a7da4ac3 fix: drop lock once it is no longer needed (#7500)
I noticed this while doing a post-commit review of https://github.com/openai/codex/pull/7467.
2025-12-02 20:46:26 +00:00
zhao-oai
1d09ac89a1 execpolicy helpers (#7032)
this PR 
- adds a helper function to amend `.codexpolicy` files with new prefix
rules
- adds a utility to `Policy` allowing prefix rules to be added to
existing `Policy` structs

both additions will be helpful as we thread codexpolicy into the TUI
workflow
2025-12-02 15:05:27 -05:00
Ahmed Ibrahim
127e307f89 Show token used when context window is unknown (#7497)
- Show context window usage in tokens instead of percentage when the
window length is unknown.
2025-12-02 11:45:50 -08:00
Ahmed Ibrahim
21ad1c1c90 Use non-blocking mutex (#7467) 2025-12-02 10:50:46 -08:00
lionel-oai
349734e38d Fix: track only untracked paths in ghost snapshots (#7470)
# Ghost snapshot ignores

This PR should close #7067, #7395, #7405.

Prior to this change the ghost snapshot task ran `git status
--ignored=matching` so the report picked up literally every ignored
file. When a directory only contained entries matched by patterns such
as `dozens/*.txt`, `/test123/generated/*.html`, or `/wp-includes/*`, Git
still enumerated them and the large-untracked-dir detection treated the
parent directory as “large,” even though everything inside was
intentionally ignored.

By removing `--ignored=matching` we only capture true untracked paths
now, so those patterns stay out of the snapshot report and no longer
trigger the “large untracked directories” warning.

---------

Signed-off-by: lionelchg <lionel.cheng@hotmail.fr>
Co-authored-by: lionelchg <lionel.cheng@hotmail.fr>
2025-12-02 19:42:33 +01:00
jif-oai
2222cab9ea feat: ignore standard directories (#7483) 2025-12-02 18:42:07 +00:00
Owen Lin
c2f8c4e9f4 fix: add ts number annotations for app-server v2 types (#7492)
These will be more ergonomic to work with in Typescript.
2025-12-02 18:09:41 +00:00
jif-oai
72b95db12f feat: intercept apply_patch for unified_exec (#7446) 2025-12-02 17:54:02 +00:00
Owen Lin
37ee6bf2c3 chore: remove mention of experimental/unstable from app-server README (#7474) 2025-12-02 17:35:05 +00:00
pakrym-oai
8b1e397211 Add request logging back (#7471)
Having full requests helps debugging
2025-12-02 07:57:55 -08:00
jif-oai
85e687c74a feat: add one off commands to app-server v2 (#7452) 2025-12-02 11:56:09 +00:00
jif-oai
9ee855ec57 feat: add warning message for the model (#7445)
Add a warning message as a user turn to the model if the model does not
behave as expected (here, for example, if the model opens too many
`unified_exec` sessions)
2025-12-02 11:56:00 +00:00
jif-oai
4b78e2ab09 chore: review everywhere (#7444) 2025-12-02 11:26:27 +00:00
jif-oai
85e2fabc9f feat: alias compaction (#7442) 2025-12-02 09:21:30 +00:00
Thibault Sottiaux
a8d5ad37b8 feat: experimental support for skills.md (#7412)
This change prototypes support for Skills with the CLI. This is an
**experimental** feature for internal testing.

---------

Co-authored-by: Gav Verma <gverma@openai.com>
2025-12-01 20:22:35 -08:00
Manoel Calixto
32e4a3a4d7 fix(tui): handle WSL clipboard image paths (#3990)
Fixes #3939 
Fixes #2803

## Summary
- convert Windows clipboard file paths into their `/mnt/<drive>`
equivalents when running inside WSL so pasted images resolve correctly
- add WSL detection helpers and share them with unit tests to cover both
native Windows and WSL clipboard normalization cases
- improve the test suite by exercising Windows path handling plus a
dedicated WSL conversion scenario and keeping the code path guarded by
targeted cfgs

## Testing
- just fmt
- cargo test -p codex-tui
- cargo clippy -p codex-tui --tests
- just fix -p codex-tui

## Screenshots
_Codex TUI screenshot:_
<img width="1880" height="848" alt="describe this copied image"
src="https://github.com/user-attachments/assets/c620d43c-f45c-451e-8893-e56ae85a5eea"
/>

_GitHub docs directory screenshot:_
<img width="1064" height="478" alt="image-copied"
src="https://github.com/user-attachments/assets/eb5eef6c-eb43-45a0-8bfe-25c35bcae753"
/>

Co-authored-by: Eric Traut <etraut@openai.com>
2025-12-01 16:54:20 -08:00
Steve Mostovoy
f443555728 fix(core): enable history lookup on windows (#7457)
- Add portable history log id helper to support inode-like tracking on
Unix and creation time on Windows
- Refactor history metadata and lookup to share code paths and allow
nonzero log ids across platforms
- Add coverage for lookup stability after appends
2025-12-01 16:29:01 -08:00
Celia Chen
ff4ca9959c [app-server] Add ImageView item (#7468)
Add view_image tool call as image_view item.

Before:
```
< {
<   "method": "codex/event/view_image_tool_call",
<   "params": {
<     "conversationId": "019adc2f-2922-7e43-ace9-64f394019616",
<     "id": "0",
<     "msg": {
<       "call_id": "call_nBQDxnTfZQtgjGpVoGuDnRjz",
<       "path": "/Users/celia/code/codex/codex-rs/app-server-protocol/codex-cli-login.png",
<       "type": "view_image_tool_call"
<     }
<   }
< }
```

After:
```
< {
<   "method": "item/started",
<   "params": {
<     "item": {
<       "id": "call_nBQDxnTfZQtgjGpVoGuDnRjz",
<       "path": "/Users/celia/code/codex/codex-rs/app-server-protocol/codex-cli-login.png",
<       "type": "imageView"
<     },
<     "threadId": "019adc2f-2922-7e43-ace9-64f394019616",
<     "turnId": "0"
<   }
< }

< {
<   "method": "item/completed",
<   "params": {
<     "item": {
<       "id": "call_nBQDxnTfZQtgjGpVoGuDnRjz",
<       "path": "/Users/celia/code/codex/codex-rs/app-server-protocol/codex-cli-login.png",
<       "type": "imageView"
<     },
<     "threadId": "019adc2f-2922-7e43-ace9-64f394019616",
<     "turnId": "0"
<   }
< }
```
2025-12-01 23:56:05 +00:00
Dylan Hurd
5b25915d7e fix(apply_patch) tests for shell_command (#7307)
## Summary
Adds test coverage for invocations of apply_patch via shell_command with
heredoc, to validate behavior.

## Testing
- [x] These are tests
2025-12-01 15:09:22 -08:00
Michael Bolin
c0564edebe chore: update to rmcp@0.10.0 to pick up support for custom client notifications (#7462)
In https://github.com/openai/codex/pull/7112, I updated our `rmcp`
dependency to point to a personal fork while I tried to upstream my
proposed change. Now that
https://github.com/modelcontextprotocol/rust-sdk/pull/556 has been
upstreamed and included in the `0.10.0` release of the crate, we can go
back to using the mainline release.
2025-12-01 14:01:50 -08:00
linuxmetel
c936c68c84 fix: prevent MCP startup failure on missing 'type' field (#7417)
Fix the issue #7416 that the codex-cli produce an error "MCP startup
failure on missing 'type' field" in the startup.

- Cause: serde in `convert_to_rmcp`
(`codex-rs/rmcp-client/src/utils.rs`) failed because no `r#type` value
was provided
- Fix: set a default `r#type` value in the corresponding structs
2025-12-01 13:58:20 -05:00
Kaden Gruizenga
41760f8a09 docs: clarify codex max defaults and xhigh availability (#7449)
## Summary
Adds the missing `xhigh` reasoning level everywhere it should have been
documented, and makes clear it only works with `gpt-5.1-codex-max`.

## Changes

* `docs/config.md`

* Add `xhigh` to the official list of reasoning levels with a note that
`xhigh` is exclusive to Codex Max.

* `docs/example-config.md`

* Update the example comment adding `xhigh` as a valid option but only
for Codex Max.

* `docs/faq.md`

  * Update the model recommendation to `GPT-5.1 Codex Max`.
* Mention that users can choose `high` or the newly documented `xhigh`
level when using Codex Max.
2025-12-01 10:46:53 -08:00
Albert O'Shea
440c7acd8f fix: nix build missing rmcp output hash (#7436)
Output hash for `rmcp-0.9.0` was missing from the nix package, (i.e.
`error: No hash was found while vendoring the git dependency
rmcp-0.9.0.`) blocking the build.
2025-12-01 10:45:31 -08:00
Ali Towaiji
0cc3b50228 Fix recent_commits(limit=0) returning 1 commit instead of 0 (#7334)
Fixes #7333

This is a small bug fix.

This PR fixes an inconsistency in `recent_commits` where `limit == 0`
still returns 1 commit due to the use of `limit.max(1)` when
constructing the `git log -n` argument.

Expected behavior: requesting 0 commits should return an empty list.

This PR:
- returns an empty `Vec` when `limit == 0`
- adds a test for `recent_commits(limit == 0)` that fails before the
change and passes afterwards
- maintains existing behavior for `limit > 0`

This aligns behavior with API expectations and avoids downstream
consumers misinterpreting the repository as having commit history when
`limit == 0` is used to explicitly request none.

Happy to adjust if the current behavior is intentional.
2025-12-01 10:14:36 -08:00
Owen Lin
8532876ad8 [app-server] fix: emit item/fileChange/outputDelta for file change items (#7399) 2025-12-01 17:52:34 +00:00
Owen Lin
44d92675eb [app-server] fix: ensure thread_id and turn_id are on all events (#7408)
This is an improvement for client-side developer ergonomics by
simplifying the state the client needs to keep track of.
2025-12-01 08:50:47 -08:00
jif-oai
a421eba31f fix: disable review rollout filtering (#7371) 2025-12-01 09:04:13 +00:00
Celia Chen
40006808a3 [app-server] add turn/plan/updated event (#7329)
transform `EventMsg::PlanDate` to v2 `turn/plan/updated` event. similar
to `turn/diff/updated`.
2025-11-30 21:09:59 -08:00
dependabot[bot]
ba58184349 chore(deps): bump image from 0.25.8 to 0.25.9 in /codex-rs (#7421)
Bumps [image](https://github.com/image-rs/image) from 0.25.8 to 0.25.9.
<details>
<summary>Changelog</summary>
<p><em>Sourced from <a
href="https://github.com/image-rs/image/blob/main/CHANGES.md">image's
changelog</a>.</em></p>
<blockquote>
<h3>Version 0.25.9</h3>
<p>Features:</p>
<ul>
<li>Support extracting XMP metadata from PNG, JPEG, GIF, WebP and TIFF
files (<a
href="https://redirect.github.com/image-rs/image/issues/2567">#2567</a>,
<a
href="https://redirect.github.com/image-rs/image/issues/2634">#2634</a>,
<a
href="https://redirect.github.com/image-rs/image/issues/2644">#2644</a>)</li>
<li>Support reading IPTC metadata from PNG and JPG files (<a
href="https://redirect.github.com/image-rs/image/issues/2611">#2611</a>)</li>
<li>Support reading ICC profile from GIF files (<a
href="https://redirect.github.com/image-rs/image/issues/2644">#2644</a>)</li>
<li>Allow setting a specific DEFLATE compression level when writing PNG
(<a
href="https://redirect.github.com/image-rs/image/issues/2583">#2583</a>)</li>
<li>Initial support for 16-bit CMYK TIFF files (<a
href="https://redirect.github.com/image-rs/image/issues/2588">#2588</a>)</li>
<li>Allow extracting the alpha channel of a <code>Pixel</code> in a
generic way (<a
href="https://redirect.github.com/image-rs/image/issues/2638">#2638</a>)</li>
</ul>
<p>Structural changes:</p>
<ul>
<li>EXR format decoding now only uses multi-threading via Rayon when the
<code>rayon</code> feature is enabled (<a
href="https://redirect.github.com/image-rs/image/issues/2643">#2643</a>)</li>
<li>Upgraded zune-jpeg to 0.5.x, ravif to 0.12.x, gif to 0.14.x</li>
<li>pnm: parse integers in PBM/PGM/PPM headers without allocations (<a
href="https://redirect.github.com/image-rs/image/issues/2620">#2620</a>)</li>
<li>Replace <code>doc_auto_cfg</code> with <code>doc_cfg</code> (<a
href="https://redirect.github.com/image-rs/image/issues/2637">#2637</a>)</li>
</ul>
<p>Bug fixes:</p>
<ul>
<li>Do not encode empty JPEG images (<a
href="https://redirect.github.com/image-rs/image/issues/2624">#2624</a>)</li>
<li>tga: reject empty images (<a
href="https://redirect.github.com/image-rs/image/issues/2614">#2614</a>)</li>
<li>tga: fix orientation flip for color mapped images (<a
href="https://redirect.github.com/image-rs/image/issues/2607">#2607</a>)</li>
<li>tga: adjust colormap lookup to match tga 2.0 spec (<a
href="https://redirect.github.com/image-rs/image/issues/2608">#2608</a>)</li>
</ul>
</blockquote>
</details>
<details>
<summary>Commits</summary>
<ul>
<li><a
href="5ceb6af6c2"><code>5ceb6af</code></a>
Merge pull request <a
href="https://redirect.github.com/image-rs/image/issues/2640">#2640</a>
from Shnatsel/release-v0.25.9</li>
<li><a
href="282d7b345c"><code>282d7b3</code></a>
Merge pull request <a
href="https://redirect.github.com/image-rs/image/issues/2646">#2646</a>
from oligamiq/main</li>
<li><a
href="5412aeee5a"><code>5412aee</code></a>
Amend the note in accordance with the advice of 197g.</li>
<li><a
href="4e8a4ed2e8"><code>4e8a4ed</code></a>
Clarify default features in README and add usage note</li>
<li><a
href="ca8fa528ff"><code>ca8fa52</code></a>
Merge pull request <a
href="https://redirect.github.com/image-rs/image/issues/2644">#2644</a>
from image-rs/gif-0.14</li>
<li><a
href="d9bc8fe790"><code>d9bc8fe</code></a>
mention GIF 0.14 changes</li>
<li><a
href="053220a0b1"><code>053220a</code></a>
Provide gif's XMP and ICC metadata</li>
<li><a
href="2ec20b3b3b"><code>2ec20b3</code></a>
Prepare codec with gif@0.14</li>
<li><a
href="31939facce"><code>31939fa</code></a>
Mention EXR rayon change</li>
<li><a
href="c7f68be265"><code>c7f68be</code></a>
Merge pull request <a
href="https://redirect.github.com/image-rs/image/issues/2643">#2643</a>
from Shnatsel/really-optional-rayon</li>
<li>Additional commits viewable in <a
href="https://github.com/image-rs/image/compare/v0.25.8...v0.25.9">compare
view</a></li>
</ul>
</details>
<br />


[![Dependabot compatibility
score](https://dependabot-badges.githubapp.com/badges/compatibility_score?dependency-name=image&package-manager=cargo&previous-version=0.25.8&new-version=0.25.9)](https://docs.github.com/en/github/managing-security-vulnerabilities/about-dependabot-security-updates#about-compatibility-scores)

Dependabot will resolve any conflicts with this PR as long as you don't
alter it yourself. You can also trigger a rebase manually by commenting
`@dependabot rebase`.

[//]: # (dependabot-automerge-start)
[//]: # (dependabot-automerge-end)

---

<details>
<summary>Dependabot commands and options</summary>
<br />

You can trigger Dependabot actions by commenting on this PR:
- `@dependabot rebase` will rebase this PR
- `@dependabot recreate` will recreate this PR, overwriting any edits
that have been made to it
- `@dependabot merge` will merge this PR after your CI passes on it
- `@dependabot squash and merge` will squash and merge this PR after
your CI passes on it
- `@dependabot cancel merge` will cancel a previously requested merge
and block automerging
- `@dependabot reopen` will reopen this PR if it is closed
- `@dependabot close` will close this PR and stop Dependabot recreating
it. You can achieve the same result by closing it manually
- `@dependabot show <dependency name> ignore conditions` will show all
of the ignore conditions of the specified dependency
- `@dependabot ignore this major version` will close this PR and stop
Dependabot creating any more for this major version (unless you reopen
the PR or upgrade to it yourself)
- `@dependabot ignore this minor version` will close this PR and stop
Dependabot creating any more for this minor version (unless you reopen
the PR or upgrade to it yourself)
- `@dependabot ignore this dependency` will close this PR and stop
Dependabot creating any more for this dependency (unless you reopen the
PR or upgrade to it yourself)


</details>

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2025-11-30 20:50:51 -08:00
Eric Traut
14df5c9492 Fixed CLA action to properly exempt dependabot (#7429) 2025-11-30 20:45:17 -08:00
dependabot[bot]
cb85a7b96e chore(deps): bump tracing from 0.1.41 to 0.1.43 in /codex-rs (#7428)
Bumps [tracing](https://github.com/tokio-rs/tracing) from 0.1.41 to
0.1.43.
<details>
<summary>Release notes</summary>
<p><em>Sourced from <a
href="https://github.com/tokio-rs/tracing/releases">tracing's
releases</a>.</em></p>
<blockquote>
<h2>tracing 0.1.43</h2>
<h4>Important</h4>
<p>The previous release [0.1.42] was yanked because <a
href="https://redirect.github.com/tokio-rs/tracing/issues/3382">#3382</a>
was a breaking change.
See further details in <a
href="https://redirect.github.com/tokio-rs/tracing/issues/3424">#3424</a>.
This release contains all the changes from that
version, plus a revert for the problematic part of the breaking PR.</p>
<h3>Fixed</h3>
<ul>
<li>Revert &quot;make <code>valueset</code> macro sanitary&quot; (<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3425">#3425</a>)</li>
</ul>
<p><a
href="https://redirect.github.com/tokio-rs/tracing/issues/3382">#3382</a>:
<a
href="https://redirect.github.com/tokio-rs/tracing/pull/3382">tokio-rs/tracing#3382</a>
<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3424">#3424</a>:
<a
href="https://redirect.github.com/tokio-rs/tracing/pull/3424">tokio-rs/tracing#3424</a>
<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3425">#3425</a>:
<a
href="https://redirect.github.com/tokio-rs/tracing/pull/3425">tokio-rs/tracing#3425</a>
[0.1.42]: <a
href="https://github.com/tokio-rs/tracing/releases/tag/tracing-0.1.42">https://github.com/tokio-rs/tracing/releases/tag/tracing-0.1.42</a></p>
<h2>tracing 0.1.42</h2>
<h3>Important</h3>
<p>The [<code>Span::record_all</code>] method has been removed from the
documented API. It
was always unsuable via the documented API as it requried a
<code>ValueSet</code> which
has no publically documented constructors. The method remains, but
should not
be used outside of <code>tracing</code> macros.</p>
<h3>Added</h3>
<ul>
<li><strong>attributes</strong>: Support constant expressions as
instrument field names (<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3158">#3158</a>)</li>
<li>Add <code>record_all!</code> macro for recording multiple values in
one call (<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3227">#3227</a>)</li>
<li><strong>core</strong>: Improve code generation at trace points
significantly (<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3398">#3398</a>)</li>
</ul>
<h3>Changed</h3>
<ul>
<li><code>tracing-core</code>: updated to 0.1.35 (<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3414">#3414</a>)</li>
<li><code>tracing-attributes</code>: updated to 0.1.31 (<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3417">#3417</a>)</li>
</ul>
<h3>Fixed</h3>
<ul>
<li>Fix &quot;name / parent&quot; variant of <code>event!</code> (<a
href="https://redirect.github.com/tokio-rs/tracing/issues/2983">#2983</a>)</li>
<li>Remove 'r#' prefix from raw identifiers in field names (<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3130">#3130</a>)</li>
<li>Fix perf regression when <code>release_max_level_*</code> not set
(<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3373">#3373</a>)</li>
<li>Use imported instead of fully qualified path (<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3374">#3374</a>)</li>
<li>Make <code>valueset</code> macro sanitary (<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3382">#3382</a>)</li>
</ul>
<h3>Documented</h3>
<ul>
<li><strong>core</strong>: Add missing <code>dyn</code> keyword in
<code>Visit</code> documentation code sample (<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3387">#3387</a>)</li>
</ul>
<p><a
href="https://redirect.github.com/tokio-rs/tracing/issues/2983">#2983</a>:
<a
href="https://redirect.github.com/tokio-rs/tracing/pull/%5B#2983%5D(https://redirect.github.com/tokio-rs/tracing/issues/2983)">tokio-rs/tracing#2983</a>
<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3130">#3130</a>:
<a
href="https://redirect.github.com/tokio-rs/tracing/pull/%5B#3130%5D(https://redirect.github.com/tokio-rs/tracing/issues/3130)">tokio-rs/tracing#3130</a>
<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3158">#3158</a>:
<a
href="https://redirect.github.com/tokio-rs/tracing/pull/%5B#3158%5D(https://redirect.github.com/tokio-rs/tracing/issues/3158)">tokio-rs/tracing#3158</a></p>
<!-- raw HTML omitted -->
</blockquote>
<p>... (truncated)</p>
</details>
<details>
<summary>Commits</summary>
<ul>
<li><a
href="64e1c8d3ae"><code>64e1c8d</code></a>
chore: prepare tracing 0.1.43 (<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3427">#3427</a>)</li>
<li><a
href="7c44f7bb21"><code>7c44f7b</code></a>
tracing: revert &quot;make <code>valueset</code> macro sanitary&quot;
(<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3425">#3425</a>)</li>
<li><a
href="cdaf661c13"><code>cdaf661</code></a>
chore: prepare tracing-mock 0.1.0-beta.2 (<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3422">#3422</a>)</li>
<li><a
href="a164fd3021"><code>a164fd3</code></a>
chore: prepare tracing-journald 0.3.2 (<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3421">#3421</a>)</li>
<li><a
href="405397b8cc"><code>405397b</code></a>
chore: prepare tracing-appender 0.2.4 (<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3420">#3420</a>)</li>
<li><a
href="a9eeed7394"><code>a9eeed7</code></a>
chore: prepare tracing-subscriber 0.3.21 (<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3419">#3419</a>)</li>
<li><a
href="5bd5505478"><code>5bd5505</code></a>
chore: prepare tracing 0.1.42 (<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3418">#3418</a>)</li>
<li><a
href="55086231ec"><code>5508623</code></a>
chore: prepare tracing-attributes 0.1.31 (<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3417">#3417</a>)</li>
<li><a
href="d92b4c0feb"><code>d92b4c0</code></a>
chore: prepare tracing-core 0.1.35 (<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3414">#3414</a>)</li>
<li><a
href="9751b6e776"><code>9751b6e</code></a>
chore: run <code>tracing-subscriber</code> tests with all features (<a
href="https://redirect.github.com/tokio-rs/tracing/issues/3412">#3412</a>)</li>
<li>Additional commits viewable in <a
href="https://github.com/tokio-rs/tracing/compare/tracing-0.1.41...tracing-0.1.43">compare
view</a></li>
</ul>
</details>
<br />


[![Dependabot compatibility
score](https://dependabot-badges.githubapp.com/badges/compatibility_score?dependency-name=tracing&package-manager=cargo&previous-version=0.1.41&new-version=0.1.43)](https://docs.github.com/en/github/managing-security-vulnerabilities/about-dependabot-security-updates#about-compatibility-scores)

Dependabot will resolve any conflicts with this PR as long as you don't
alter it yourself. You can also trigger a rebase manually by commenting
`@dependabot rebase`.

[//]: # (dependabot-automerge-start)
[//]: # (dependabot-automerge-end)

---

<details>
<summary>Dependabot commands and options</summary>
<br />

You can trigger Dependabot actions by commenting on this PR:
- `@dependabot rebase` will rebase this PR
- `@dependabot recreate` will recreate this PR, overwriting any edits
that have been made to it
- `@dependabot merge` will merge this PR after your CI passes on it
- `@dependabot squash and merge` will squash and merge this PR after
your CI passes on it
- `@dependabot cancel merge` will cancel a previously requested merge
and block automerging
- `@dependabot reopen` will reopen this PR if it is closed
- `@dependabot close` will close this PR and stop Dependabot recreating
it. You can achieve the same result by closing it manually
- `@dependabot show <dependency name> ignore conditions` will show all
of the ignore conditions of the specified dependency
- `@dependabot ignore this major version` will close this PR and stop
Dependabot creating any more for this major version (unless you reopen
the PR or upgrade to it yourself)
- `@dependabot ignore this minor version` will close this PR and stop
Dependabot creating any more for this minor version (unless you reopen
the PR or upgrade to it yourself)
- `@dependabot ignore this dependency` will close this PR and stop
Dependabot creating any more for this dependency (unless you reopen the
PR or upgrade to it yourself)


</details>

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Eric Traut <etraut@openai.com>
2025-11-30 20:36:03 -08:00
dependabot[bot]
3f12f1140f chore(deps): bump reqwest from 0.12.23 to 0.12.24 in /codex-rs (#7424)
Bumps [reqwest](https://github.com/seanmonstar/reqwest) from 0.12.23 to
0.12.24.
<details>
<summary>Release notes</summary>
<p><em>Sourced from <a
href="https://github.com/seanmonstar/reqwest/releases">reqwest's
releases</a>.</em></p>
<blockquote>
<h2>v0.12.24</h2>
<h2>Highlights</h2>
<ul>
<li>Refactor cookie handling to an internal middleware.</li>
<li>Refactor internal random generator.</li>
<li>Refactor base64 encoding to reduce a copy.</li>
<li>Documentation updates.</li>
</ul>
<h2>What's Changed</h2>
<ul>
<li>build(deps): silence unused deps in WASM build by <a
href="https://github.com/0x676e67"><code>@​0x676e67</code></a> in <a
href="https://redirect.github.com/seanmonstar/reqwest/pull/2799">seanmonstar/reqwest#2799</a></li>
<li>perf(util): avoid extra copy when base64 encoding by <a
href="https://github.com/0x676e67"><code>@​0x676e67</code></a> in <a
href="https://redirect.github.com/seanmonstar/reqwest/pull/2805">seanmonstar/reqwest#2805</a></li>
<li>docs: fix method name in changelog entry by <a
href="https://github.com/johannespfrang"><code>@​johannespfrang</code></a>
in <a
href="https://redirect.github.com/seanmonstar/reqwest/pull/2807">seanmonstar/reqwest#2807</a></li>
<li>chore: Align the name usage of TotalTimeout by <a
href="https://github.com/Xuanwo"><code>@​Xuanwo</code></a> in <a
href="https://redirect.github.com/seanmonstar/reqwest/pull/2657">seanmonstar/reqwest#2657</a></li>
<li>refactor(cookie): add <code>CookieService</code> by <a
href="https://github.com/linyihai"><code>@​linyihai</code></a> in <a
href="https://redirect.github.com/seanmonstar/reqwest/pull/2787">seanmonstar/reqwest#2787</a></li>
<li>Fixes typo in retry max_retries_per_request doc comment re 2813 by
<a href="https://github.com/dmackinn"><code>@​dmackinn</code></a> in <a
href="https://redirect.github.com/seanmonstar/reqwest/pull/2824">seanmonstar/reqwest#2824</a></li>
<li>test(multipart): fix build failure with
<code>no-default-features</code> by <a
href="https://github.com/0x676e67"><code>@​0x676e67</code></a> in <a
href="https://redirect.github.com/seanmonstar/reqwest/pull/2801">seanmonstar/reqwest#2801</a></li>
<li>refactor(cookie): avoid duplicate cookie insertion by <a
href="https://github.com/0x676e67"><code>@​0x676e67</code></a> in <a
href="https://redirect.github.com/seanmonstar/reqwest/pull/2834">seanmonstar/reqwest#2834</a></li>
</ul>
<h2>New Contributors</h2>
<ul>
<li><a
href="https://github.com/johannespfrang"><code>@​johannespfrang</code></a>
made their first contribution in <a
href="https://redirect.github.com/seanmonstar/reqwest/pull/2807">seanmonstar/reqwest#2807</a></li>
<li><a href="https://github.com/dmackinn"><code>@​dmackinn</code></a>
made their first contribution in <a
href="https://redirect.github.com/seanmonstar/reqwest/pull/2824">seanmonstar/reqwest#2824</a></li>
</ul>
<p><strong>Full Changelog</strong>: <a
href="https://github.com/seanmonstar/reqwest/compare/v0.12.23...v0.12.24">https://github.com/seanmonstar/reqwest/compare/v0.12.23...v0.12.24</a></p>
</blockquote>
</details>
<details>
<summary>Changelog</summary>
<p><em>Sourced from <a
href="https://github.com/seanmonstar/reqwest/blob/master/CHANGELOG.md">reqwest's
changelog</a>.</em></p>
<blockquote>
<h2>v0.12.24</h2>
<ul>
<li>Refactor cookie handling to an internal middleware.</li>
<li>Refactor internal random generator.</li>
<li>Refactor base64 encoding to reduce a copy.</li>
<li>Documentation updates.</li>
</ul>
</blockquote>
</details>
<details>
<summary>Commits</summary>
<ul>
<li><a
href="b126ca49da"><code>b126ca4</code></a>
v0.12.24</li>
<li><a
href="4023493096"><code>4023493</code></a>
refactor: change fast_random from xorshift to siphash a counter</li>
<li><a
href="fd61bc93e6"><code>fd61bc9</code></a>
refactor(cookie): avoid duplicate cookie insertion (<a
href="https://redirect.github.com/seanmonstar/reqwest/issues/2834">#2834</a>)</li>
<li><a
href="0bfa526776"><code>0bfa526</code></a>
test(multipart): fix build failure with <code>no-default-features</code>
(<a
href="https://redirect.github.com/seanmonstar/reqwest/issues/2801">#2801</a>)</li>
<li><a
href="994b8a0b7a"><code>994b8a0</code></a>
docs: typo in retry max_retries_per_request (<a
href="https://redirect.github.com/seanmonstar/reqwest/issues/2824">#2824</a>)</li>
<li><a
href="da0702b762"><code>da0702b</code></a>
refactor(cookie): de-duplicate cookie support as
<code>CookieService</code> middleware (...</li>
<li><a
href="7ebddeaa87"><code>7ebddea</code></a>
chore: align internal name usage of TotalTimeout (<a
href="https://redirect.github.com/seanmonstar/reqwest/issues/2657">#2657</a>)</li>
<li><a
href="b540a4e746"><code>b540a4e</code></a>
chore(readme): use correct CI status badge</li>
<li><a
href="e4550c4cc5"><code>e4550c4</code></a>
docs: fix method name in changelog entry (<a
href="https://redirect.github.com/seanmonstar/reqwest/issues/2807">#2807</a>)</li>
<li><a
href="f4694a2922"><code>f4694a2</code></a>
perf(util): avoid extra copy when base64 encoding (<a
href="https://redirect.github.com/seanmonstar/reqwest/issues/2805">#2805</a>)</li>
<li>Additional commits viewable in <a
href="https://github.com/seanmonstar/reqwest/compare/v0.12.23...v0.12.24">compare
view</a></li>
</ul>
</details>
<br />


[![Dependabot compatibility
score](https://dependabot-badges.githubapp.com/badges/compatibility_score?dependency-name=reqwest&package-manager=cargo&previous-version=0.12.23&new-version=0.12.24)](https://docs.github.com/en/github/managing-security-vulnerabilities/about-dependabot-security-updates#about-compatibility-scores)

Dependabot will resolve any conflicts with this PR as long as you don't
alter it yourself. You can also trigger a rebase manually by commenting
`@dependabot rebase`.

[//]: # (dependabot-automerge-start)
[//]: # (dependabot-automerge-end)

---

<details>
<summary>Dependabot commands and options</summary>
<br />

You can trigger Dependabot actions by commenting on this PR:
- `@dependabot rebase` will rebase this PR
- `@dependabot recreate` will recreate this PR, overwriting any edits
that have been made to it
- `@dependabot merge` will merge this PR after your CI passes on it
- `@dependabot squash and merge` will squash and merge this PR after
your CI passes on it
- `@dependabot cancel merge` will cancel a previously requested merge
and block automerging
- `@dependabot reopen` will reopen this PR if it is closed
- `@dependabot close` will close this PR and stop Dependabot recreating
it. You can achieve the same result by closing it manually
- `@dependabot show <dependency name> ignore conditions` will show all
of the ignore conditions of the specified dependency
- `@dependabot ignore this major version` will close this PR and stop
Dependabot creating any more for this major version (unless you reopen
the PR or upgrade to it yourself)
- `@dependabot ignore this minor version` will close this PR and stop
Dependabot creating any more for this minor version (unless you reopen
the PR or upgrade to it yourself)
- `@dependabot ignore this dependency` will close this PR and stop
Dependabot creating any more for this dependency (unless you reopen the
PR or upgrade to it yourself)


</details>

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Eric Traut <etraut@openai.com>
2025-11-30 20:35:49 -08:00
dependabot[bot]
c22cd2e953 chore(deps): bump serde_with from 3.14.0 to 3.16.1 in /codex-rs (#7422)
Bumps [serde_with](https://github.com/jonasbb/serde_with) from 3.14.0 to
3.16.1.
<details>
<summary>Release notes</summary>
<p><em>Sourced from <a
href="https://github.com/jonasbb/serde_with/releases">serde_with's
releases</a>.</em></p>
<blockquote>
<h2>serde_with v3.16.1</h2>
<h3>Fixed</h3>
<ul>
<li>Fix <code>JsonSchemaAs</code> of <code>SetPreventDuplicates</code>
and <code>SetLastValueWins</code>. (<a
href="https://redirect.github.com/jonasbb/serde_with/issues/906">#906</a>,
<a
href="https://redirect.github.com/jonasbb/serde_with/issues/907">#907</a>)</li>
</ul>
<h2>serde_with v3.16.0</h2>
<h3>Added</h3>
<ul>
<li>Added support for <code>smallvec</code> v1 under the
<code>smallvec_1</code> feature flag by <a
href="https://github.com/isharma228"><code>@​isharma228</code></a> (<a
href="https://redirect.github.com/jonasbb/serde_with/issues/895">#895</a>)</li>
<li>Add <code>JsonSchemaAs</code> implementation for
<code>json::JsonString</code> by <a
href="https://github.com/yogevm15"><code>@​yogevm15</code></a> (<a
href="https://redirect.github.com/jonasbb/serde_with/issues/901">#901</a>)</li>
</ul>
<h2>serde_with v3.15.1</h2>
<h3>Fixed</h3>
<ul>
<li>Fix building of the documentation by updating references to use
<code>serde_core</code>.</li>
</ul>
<h2>serde_with v3.15.0</h2>
<h3>Added</h3>
<ul>
<li>
<p>Added error inspection to <code>VecSkipError</code> and
<code>MapSkipError</code> by <a
href="https://github.com/michelhe"><code>@​michelhe</code></a> (<a
href="https://redirect.github.com/jonasbb/serde_with/issues/878">#878</a>)
This allows interacting with the previously hidden error, for example
for logging.
Checkout the newly added example to both types.</p>
</li>
<li>
<p>Allow documenting the types generated by <code>serde_conv!</code>.
The <code>serde_conv!</code> macro now acceps outer attributes before
the optional visibility modifier.
This allow adding doc comments in the shape of <code>#[doc =
&quot;...&quot;]</code> or any other attributes, such as lint
modifiers.</p>
<pre lang="rust"><code>serde_conv!(
    #[doc = &quot;Serialize bools as string&quot;]
    #[allow(dead_code)]
    pub BoolAsString,
    bool,
    |x: &amp;bool| ::std::string::ToString::to_string(x),
    |x: ::std::string::String| x.parse()
);
</code></pre>
</li>
<li>
<p>Add support for <code>hashbrown</code> v0.16 (<a
href="https://redirect.github.com/jonasbb/serde_with/issues/877">#877</a>)</p>
<p>This extends the existing support for <code>hashbrown</code> v0.14
and v0.15 to the newly released version.</p>
</li>
</ul>
<h3>Changed</h3>
<ul>
<li>Bump MSRV to 1.76, since that is required for <code>toml</code>
dev-dependency.</li>
</ul>
<h2>serde_with v3.14.1</h2>
<h3>Fixed</h3>
<ul>
<li>Show macro expansion in the docs.rs generated rustdoc.</li>
</ul>
<!-- raw HTML omitted -->
</blockquote>
<p>... (truncated)</p>
</details>
<details>
<summary>Commits</summary>
<ul>
<li><a
href="8513323fda"><code>8513323</code></a>
Bump version to 3.16.1 (<a
href="https://redirect.github.com/jonasbb/serde_with/issues/908">#908</a>)</li>
<li><a
href="5392bbe75e"><code>5392bbe</code></a>
Bump version to 3.16.1</li>
<li><a
href="1e54f1cd38"><code>1e54f1c</code></a>
Fix duplicate schema set definitions for schemars 0.8, 0.9, and 1.0 (<a
href="https://redirect.github.com/jonasbb/serde_with/issues/907">#907</a>)</li>
<li><a
href="0650180645"><code>0650180</code></a>
Fix duplicate schema set definitions for schemars 0.8, 0.9, and 1.0</li>
<li><a
href="41d1033438"><code>41d1033</code></a>
Fix test conditions for schemars tests to include &quot;hex&quot;
feature</li>
<li><a
href="2eed58af05"><code>2eed58a</code></a>
Bump the github-actions group across 1 directory with 2 updates (<a
href="https://redirect.github.com/jonasbb/serde_with/issues/905">#905</a>)</li>
<li><a
href="ed040f2330"><code>ed040f2</code></a>
Bump the github-actions group across 1 directory with 2 updates</li>
<li><a
href="fa2129b1b9"><code>fa2129b</code></a>
Bump ron from 0.11.0 to 0.12.0 (<a
href="https://redirect.github.com/jonasbb/serde_with/issues/904">#904</a>)</li>
<li><a
href="b55cb99757"><code>b55cb99</code></a>
Bump ron from 0.11.0 to 0.12.0</li>
<li><a
href="066b9d4019"><code>066b9d4</code></a>
Bump version to 3.16.0 (<a
href="https://redirect.github.com/jonasbb/serde_with/issues/903">#903</a>)</li>
<li>Additional commits viewable in <a
href="https://github.com/jonasbb/serde_with/compare/v3.14.0...v3.16.1">compare
view</a></li>
</ul>
</details>
<br />


[![Dependabot compatibility
score](https://dependabot-badges.githubapp.com/badges/compatibility_score?dependency-name=serde_with&package-manager=cargo&previous-version=3.14.0&new-version=3.16.1)](https://docs.github.com/en/github/managing-security-vulnerabilities/about-dependabot-security-updates#about-compatibility-scores)

Dependabot will resolve any conflicts with this PR as long as you don't
alter it yourself. You can also trigger a rebase manually by commenting
`@dependabot rebase`.

[//]: # (dependabot-automerge-start)
[//]: # (dependabot-automerge-end)

---

<details>
<summary>Dependabot commands and options</summary>
<br />

You can trigger Dependabot actions by commenting on this PR:
- `@dependabot rebase` will rebase this PR
- `@dependabot recreate` will recreate this PR, overwriting any edits
that have been made to it
- `@dependabot merge` will merge this PR after your CI passes on it
- `@dependabot squash and merge` will squash and merge this PR after
your CI passes on it
- `@dependabot cancel merge` will cancel a previously requested merge
and block automerging
- `@dependabot reopen` will reopen this PR if it is closed
- `@dependabot close` will close this PR and stop Dependabot recreating
it. You can achieve the same result by closing it manually
- `@dependabot show <dependency name> ignore conditions` will show all
of the ignore conditions of the specified dependency
- `@dependabot ignore this major version` will close this PR and stop
Dependabot creating any more for this major version (unless you reopen
the PR or upgrade to it yourself)
- `@dependabot ignore this minor version` will close this PR and stop
Dependabot creating any more for this minor version (unless you reopen
the PR or upgrade to it yourself)
- `@dependabot ignore this dependency` will close this PR and stop
Dependabot creating any more for this dependency (unless you reopen the
PR or upgrade to it yourself)


</details>

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Eric Traut <etraut@openai.com>
2025-11-30 20:35:32 -08:00
dependabot[bot]
ebd485b1a0 chore(deps): bump arboard from 3.6.0 to 3.6.1 in /codex-rs (#7426)
Bumps [arboard](https://github.com/1Password/arboard) from 3.6.0 to
3.6.1.
<details>
<summary>Release notes</summary>
<p><em>Sourced from <a
href="https://github.com/1Password/arboard/releases">arboard's
releases</a>.</em></p>
<blockquote>
<h2>v3.6.1</h2>
<p>This release focuses on improving compatibility with data in the real
world and bug fixes. It also includes a new <code>Set</code> API for
working with file paths via drag-and-drop interfaces across Linux,
macOS, and Windows.</p>
<p>This release also marks the start of exclusively publishing
changelogs via GitHub Releases. The old <code>CHANGELOG.md</code> has
been removed due to maintenance overhead and duplication. <a
href="https://github.com/1Password/arboard/releases/tag/v3.6.0">v3.6.0</a>
is the last revision to include this file.</p>
<h3>Added</h3>
<ul>
<li>Add support for pasting lists of files via
<code>Set::file_list</code> interface by <a
href="https://github.com/Gae24"><code>@​Gae24</code></a> in <a
href="https://redirect.github.com/1Password/arboard/pull/181">1Password/arboard#181</a></li>
<li>Support <code>windows-sys</code> 0.60 in <code>arboard</code>'s
allowed version range by <a
href="https://github.com/complexspaces"><code>@​complexspaces</code></a>
in <a
href="https://redirect.github.com/1Password/arboard/pull/201">1Password/arboard#201</a></li>
</ul>
<h3>Changed</h3>
<ul>
<li>Fix grammar and typos by <a
href="https://github.com/complexspaces"><code>@​complexspaces</code></a>
and <a href="https://github.com/gagath"><code>@​gagath</code></a> in <a
href="https://redirect.github.com/1Password/arboard/pull/194">1Password/arboard#194</a>
and <a
href="https://redirect.github.com/1Password/arboard/pull/196">1Password/arboard#196</a></li>
<li>Prefer PNG when pasting images on Windows by <a
href="https://github.com/wcassels"><code>@​wcassels</code></a> in <a
href="https://redirect.github.com/1Password/arboard/pull/198">1Password/arboard#198</a>
<ul>
<li>Note: This change greatly increases compatibility for
&quot;complicated&quot; images that contain alpha values and/or
transparent pixels. Support for transparency in <code>BITMAP</code>
formats is ill-defined and inconsistently implemented in the wild, but
is consistent in <code>PNG</code>. Most applications loading images onto
the clipboard include <code>PNG</code>-encoded data already.</li>
</ul>
</li>
<li>Bitmap images pasted on Windows now use the <code>image</code> crate
instead of a homegrown internal parser.
<ul>
<li>This <strong>should not</strong> regress any existing Bitmap use
cases and instead will provide more consistent and robust parsing. If
you notice something now broken, please open an issue!</li>
</ul>
</li>
</ul>
<h3>Fixed</h3>
<ul>
<li>Remove silent dropping of file paths when non-UTF8 was mixed in on
Linux by <a href="https://github.com/Gae24"><code>@​Gae24</code></a> in
<a
href="https://redirect.github.com/1Password/arboard/pull/197">1Password/arboard#197</a></li>
<li>Fix parsing of 24-bit bitmaps on Windows by <a
href="https://github.com/wcassels"><code>@​wcassels</code></a> in <a
href="https://redirect.github.com/1Password/arboard/pull/198">1Password/arboard#198</a>
<ul>
<li>Example: Images with transparency copied by Firefox are now handled
correctly, among others.</li>
</ul>
</li>
</ul>
<h2>New Contributors</h2>
<ul>
<li><a href="https://github.com/gagath"><code>@​gagath</code></a> made
their first contribution in <a
href="https://redirect.github.com/1Password/arboard/pull/196">1Password/arboard#196</a></li>
<li><a href="https://github.com/wcassels"><code>@​wcassels</code></a>
made their first contribution in <a
href="https://redirect.github.com/1Password/arboard/pull/198">1Password/arboard#198</a></li>
</ul>
<p><strong>Full Changelog</strong>: <a
href="https://github.com/1Password/arboard/compare/v3.6.0...v3.6.1">https://github.com/1Password/arboard/compare/v3.6.0...v3.6.1</a></p>
</blockquote>
</details>
<details>
<summary>Commits</summary>
<ul>
<li><a
href="a3750c79a5"><code>a3750c7</code></a>
Release 3.6.1</li>
<li><a
href="edcce2cd6b"><code>edcce2c</code></a>
Remove CHANGELOG.md in favor of GitHub releases</li>
<li><a
href="26a96a6199"><code>26a96a6</code></a>
Bump windows-sys semver range to support 0.60.x</li>
<li><a
href="7bdd1c1175"><code>7bdd1c1</code></a>
Update errno for windows-sys 0.60 flexibility</li>
<li><a
href="55c0b260c4"><code>55c0b26</code></a>
read/write_unaligned rather than using manual field offsets</li>
<li><a
href="ff15a093d6"><code>ff15a09</code></a>
Return conversionFailure instead of adhoc errors</li>
<li><a
href="16ef18113f"><code>16ef181</code></a>
Implement fetching PNG on Windows and prefer over DIB when
available</li>
<li><a
href="a3c64f9a93"><code>a3c64f9</code></a>
Add a couple of end-to-end DIBV5 tests</li>
<li><a
href="e6008eaa91"><code>e6008ea</code></a>
Use image for reading DIB and try to make it do the right thing for
32-bit BI...</li>
<li><a
href="17ef05ce13"><code>17ef05c</code></a>
add <code>file_list</code> to <code>Set</code> interface (<a
href="https://redirect.github.com/1Password/arboard/issues/181">#181</a>)</li>
<li>Additional commits viewable in <a
href="https://github.com/1Password/arboard/compare/v3.6.0...v3.6.1">compare
view</a></li>
</ul>
</details>
<br />


[![Dependabot compatibility
score](https://dependabot-badges.githubapp.com/badges/compatibility_score?dependency-name=arboard&package-manager=cargo&previous-version=3.6.0&new-version=3.6.1)](https://docs.github.com/en/github/managing-security-vulnerabilities/about-dependabot-security-updates#about-compatibility-scores)

Dependabot will resolve any conflicts with this PR as long as you don't
alter it yourself. You can also trigger a rebase manually by commenting
`@dependabot rebase`.

[//]: # (dependabot-automerge-start)
[//]: # (dependabot-automerge-end)

---

<details>
<summary>Dependabot commands and options</summary>
<br />

You can trigger Dependabot actions by commenting on this PR:
- `@dependabot rebase` will rebase this PR
- `@dependabot recreate` will recreate this PR, overwriting any edits
that have been made to it
- `@dependabot merge` will merge this PR after your CI passes on it
- `@dependabot squash and merge` will squash and merge this PR after
your CI passes on it
- `@dependabot cancel merge` will cancel a previously requested merge
and block automerging
- `@dependabot reopen` will reopen this PR if it is closed
- `@dependabot close` will close this PR and stop Dependabot recreating
it. You can achieve the same result by closing it manually
- `@dependabot show <dependency name> ignore conditions` will show all
of the ignore conditions of the specified dependency
- `@dependabot ignore this major version` will close this PR and stop
Dependabot creating any more for this major version (unless you reopen
the PR or upgrade to it yourself)
- `@dependabot ignore this minor version` will close this PR and stop
Dependabot creating any more for this minor version (unless you reopen
the PR or upgrade to it yourself)
- `@dependabot ignore this dependency` will close this PR and stop
Dependabot creating any more for this dependency (unless you reopen the
PR or upgrade to it yourself)


</details>

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2025-11-30 20:20:45 -08:00
jif-oai
457c9fdb87 chore: better session recycling (#7368) 2025-11-30 12:42:26 -08:00
jif-oai
6eeaf46ac1 fix: other flaky tests (#7372) 2025-11-28 15:29:44 +00:00
jif-oai
aaec8abf58 feat: detached review (#7292) 2025-11-28 11:34:57 +00:00
Job Chong
cbd7d0d543 chore: improve rollout session init errors (#7336)
Title: Improve rollout session initialization error messages

Issue: https://github.com/openai/codex/issues/7283

What: add targeted mapping for rollout/session initialization errors so
users get actionable messages when Codex cannot access session files.

Why: session creation previously returned a generic internal error,
hiding permissions/FS issues and making support harder.

How:
- Added rollout::error::map_session_init_error to translate the more
common io::Error kinds into user-facing hints (permission, missing dir,
file blocking, corruption). Others are passed through directly with
`CodexErr::Fatal`.
- Reused the mapper in Codex session creation to preserve root causes
instead of returning InternalAgentDied.
2025-11-27 00:20:33 -08:00
Eric Traut
fabdbfef9c Fixes two bugs in example-config.md documentation (#7324)
This PR is a modified version of [a
PR](https://github.com/openai/codex/pull/7316) submitted by @yydrowz3.
* Removes a redundant `experimental_sandbox_command_assessment` flag
* Moves `mcp_oauth_credentials_store` from the `[features]` table, where
it doesn't belong
2025-11-26 09:52:13 -08:00
lionel-oai
8b314e2d04 doc: fix relative links and add tips (#7319)
This PR is a documentation only one which:
- addresses the #7231 by adding a paragraph in `docs/getting-started.md`
in the tips category to encourage users to load everything needed in
their environment
- corrects link referencing in `docs/platform-sandboxing.md` so that the
page link opens at the right section
- removes the explicit heading IDs like {#my-id} in `docs/advanced.md`
which are not supported by GitHub and are **not** rendered in the UI:

<img width="1198" height="849" alt="Screenshot 2025-11-26 at 16 25 31"
src="https://github.com/user-attachments/assets/308d33c3-81d3-4785-a6c1-e9377e6d3ea6"
/>

This caused the following links in `README.md` to not work in `main` but
to work in this branch (you can test by going to
https://github.com/openai/codex/blob/docs/getting-started-enhancement/README.md)
- the MCP link goes straight to the correct section now:

```markdown
  - [**Advanced**](./docs/advanced.md)
  - [Tracing / verbose logging](./docs/advanced.md#tracing--verbose-logging)
  - [Model Context Protocol (MCP)](./docs/advanced.md#model-context-protocol-mcp)
```

---------

Signed-off-by: lionel-oai <lionel@openai.com>
Signed-off-by: lionelchg <lionel.cheng@hotmail.fr>
Co-authored-by: lionelchg <lionel.cheng@hotmail.fr>
2025-11-26 09:35:08 -08:00
jif-oai
963009737f nit: drop file (#7314) 2025-11-26 11:30:34 +00:00
Eric Traut
e953092949 Fixed regression in experimental "sandbox command assessment" feature (#7308)
Recent model updates caused the experimental "sandbox tool assessment"
to time out most of the time leaving the user without any risk
assessment or tool summary. This change explicitly sets the reasoning
effort to medium and bumps the timeout.

This change has no effect if the user hasn't enabled the
`experimental_sandbox_command_assessment` feature flag.
2025-11-25 16:15:13 -08:00
jif-oai
28ff364c3a feat: update process ID for event handling (#7261) 2025-11-25 14:21:05 -08:00
jif-oai
63a3f3941a Merge branch 'main' into jif/fork 2025-11-20 12:36:40 +00:00
jif-oai
7ceabac707 nit fix 2025-11-19 20:40:35 +00:00
jif-oai
2a24ae36c2 Merge remote-tracking branch 'origin/main' into jif/fork
# Conflicts:
#	codex-rs/core/src/codex.rs
2025-11-19 20:21:19 +00:00
jif-oai
5071cc8fff Add a few tests 2025-11-19 15:30:04 +00:00
jif-oai
9c765f1217 Compilation nits 2025-11-19 15:21:58 +00:00
jif-oai
7116d2a6a4 More tests 2025-11-19 15:16:12 +00:00
jif-oai
94f1a61df5 Clippy 2025-11-19 15:06:51 +00:00
jif-oai
7ec1311aff Decoupling of TUI 2025-11-19 14:53:17 +00:00
jif-oai
b9f260057c Optional command 2025-11-19 14:21:04 +00:00
jif-oai
72af9e3092 Clippy 2025-11-19 12:39:28 +00:00
jif-oai
8f0d83eb11 Clean errors 2025-11-19 12:31:27 +00:00
jif-oai
26d667e152 Add tests 2025-11-19 12:13:36 +00:00
jif-oai
d1cf2b967c One fix 2025-11-19 11:50:42 +00:00
jif-oai
78afe914e1 Merge remote-tracking branch 'origin/main' into jif/fork
# Conflicts:
#	codex-rs/cli/src/main.rs
#	codex-rs/tui/src/cli.rs
2025-11-19 11:41:17 +00:00
jif-oai
1a8c1a4d9a V1 2025-11-18 16:50:45 +00:00
123 changed files with 7293 additions and 1975 deletions

View File

@@ -46,7 +46,4 @@ jobs:
path-to-document: https://github.com/openai/codex/blob/main/docs/CLA.md
path-to-signatures: signatures/cla.json
branch: cla-signatures
allowlist: |
codex
dependabot
dependabot[bot]
allowlist: codex,dependabot,dependabot[bot],github-actions[bot]

113
codex-rs/Cargo.lock generated
View File

@@ -198,9 +198,9 @@ dependencies = [
[[package]]
name = "arboard"
version = "3.6.0"
version = "3.6.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "55f533f8e0af236ffe5eb979b99381df3258853f00ba2e44b6e1955292c75227"
checksum = "0348a1c054491f4bfe6ab86a7b6ab1e44e45d899005de92f58b3df180b36ddaf"
dependencies = [
"clipboard-win",
"image",
@@ -212,7 +212,7 @@ dependencies = [
"objc2-foundation",
"parking_lot",
"percent-encoding",
"windows-sys 0.52.0",
"windows-sys 0.60.2",
"wl-clipboard-rs",
"x11rb",
]
@@ -1068,6 +1068,7 @@ dependencies = [
"serde_json",
"thiserror 2.0.17",
"tokio",
"tracing",
]
[[package]]
@@ -1144,6 +1145,7 @@ dependencies = [
"codex-apply-patch",
"codex-arg0",
"codex-async-utils",
"codex-core",
"codex-execpolicy",
"codex-file-search",
"codex-git",
@@ -1185,6 +1187,7 @@ dependencies = [
"seccompiler",
"serde",
"serde_json",
"serde_yaml",
"serial_test",
"sha1",
"sha2",
@@ -1280,6 +1283,7 @@ dependencies = [
"serde_json",
"shlex",
"starlark",
"tempfile",
"thiserror 2.0.17",
]
@@ -1611,6 +1615,7 @@ dependencies = [
"textwrap 0.16.2",
"tokio",
"tokio-stream",
"tokio-util",
"toml",
"tracing",
"tracing-appender",
@@ -2535,7 +2540,7 @@ checksum = "0ce92ff622d6dadf7349484f42c93271a0d49b7cc4d466a936405bacbe10aa78"
dependencies = [
"cfg-if",
"rustix 1.0.8",
"windows-sys 0.52.0",
"windows-sys 0.59.0",
]
[[package]]
@@ -3292,9 +3297,9 @@ dependencies = [
[[package]]
name = "image"
version = "0.25.8"
version = "0.25.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "529feb3e6769d234375c4cf1ee2ce713682b8e76538cb13f9fc23e1400a591e7"
checksum = "e6506c6c10786659413faa717ceebcb8f70731c0a60cbae39795fdf114519c1a"
dependencies = [
"bytemuck",
"byteorder-lite",
@@ -3302,8 +3307,8 @@ dependencies = [
"num-traits",
"png",
"tiff",
"zune-core",
"zune-jpeg",
"zune-core 0.5.0",
"zune-jpeg 0.5.5",
]
[[package]]
@@ -3439,7 +3444,7 @@ checksum = "e04d7f318608d35d4b61ddd75cbdaee86b023ebe2bd5a66ee0915f0bf93095a9"
dependencies = [
"hermit-abi",
"libc",
"windows-sys 0.52.0",
"windows-sys 0.59.0",
]
[[package]]
@@ -4463,6 +4468,12 @@ version = "1.0.15"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "57c0d7b74b563b49d38dae00a0c37d4d6de9b432382b2892f0574ddcae73fd0a"
[[package]]
name = "pastey"
version = "0.2.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "57d6c094ee800037dff99e02cab0eaf3142826586742a270ab3d7a62656bd27a"
[[package]]
name = "path-absolutize"
version = "3.1.1"
@@ -5078,9 +5089,9 @@ checksum = "2b15c43186be67a4fd63bee50d0303afffcef381492ebe2c5d87f324e1b8815c"
[[package]]
name = "reqwest"
version = "0.12.23"
version = "0.12.24"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "d429f34c8092b2d42c7c93cec323bb4adeb7c67698f70839adec842ec10c7ceb"
checksum = "9d0946410b9f7b082a427e4ef5c8ff541a88b357bc6c637c40db3a68ac70a36f"
dependencies = [
"base64",
"bytes",
@@ -5141,8 +5152,9 @@ dependencies = [
[[package]]
name = "rmcp"
version = "0.9.0"
source = "git+https://github.com/bolinfest/rust-sdk?branch=pr556#4d9cc16f4c76c84486344f542ed9a3e9364019ba"
version = "0.10.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "38b18323edc657390a6ed4d7a9110b0dec2dc3ed128eb2a123edfbafabdbddc5"
dependencies = [
"async-trait",
"base64",
@@ -5153,7 +5165,7 @@ dependencies = [
"http-body",
"http-body-util",
"oauth2",
"paste",
"pastey",
"pin-project-lite",
"process-wrap",
"rand 0.9.2",
@@ -5175,8 +5187,9 @@ dependencies = [
[[package]]
name = "rmcp-macros"
version = "0.9.0"
source = "git+https://github.com/bolinfest/rust-sdk?branch=pr556#4d9cc16f4c76c84486344f542ed9a3e9364019ba"
version = "0.10.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "c75d0a62676bf8c8003c4e3c348e2ceb6a7b3e48323681aaf177fdccdac2ce50"
dependencies = [
"darling 0.21.3",
"proc-macro2",
@@ -5216,7 +5229,7 @@ dependencies = [
"errno",
"libc",
"linux-raw-sys 0.4.15",
"windows-sys 0.52.0",
"windows-sys 0.59.0",
]
[[package]]
@@ -5735,9 +5748,9 @@ dependencies = [
[[package]]
name = "serde_with"
version = "3.14.0"
version = "3.16.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f2c45cd61fefa9db6f254525d46e392b852e0e61d9a1fd36e5bd183450a556d5"
checksum = "4fa237f2807440d238e0364a218270b98f767a00d3dada77b1c53ae88940e2e7"
dependencies = [
"base64",
"chrono",
@@ -5746,8 +5759,7 @@ dependencies = [
"indexmap 2.12.0",
"schemars 0.9.0",
"schemars 1.0.4",
"serde",
"serde_derive",
"serde_core",
"serde_json",
"serde_with_macros",
"time",
@@ -5755,16 +5767,29 @@ dependencies = [
[[package]]
name = "serde_with_macros"
version = "3.14.0"
version = "3.16.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "de90945e6565ce0d9a25098082ed4ee4002e047cb59892c318d66821e14bb30f"
checksum = "52a8e3ca0ca629121f70ab50f95249e5a6f925cc0f6ffe8256c45b728875706c"
dependencies = [
"darling 0.20.11",
"darling 0.21.3",
"proc-macro2",
"quote",
"syn 2.0.104",
]
[[package]]
name = "serde_yaml"
version = "0.9.34+deprecated"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "6a8b1a1a2ebf674015cc02edccce75287f1a0130d394307b36743c2f5d504b47"
dependencies = [
"indexmap 2.12.0",
"itoa",
"ryu",
"serde",
"unsafe-libyaml",
]
[[package]]
name = "serial2"
version = "0.2.31"
@@ -6408,7 +6433,7 @@ dependencies = [
"half",
"quick-error",
"weezl",
"zune-jpeg",
"zune-jpeg 0.4.19",
]
[[package]]
@@ -6576,6 +6601,7 @@ dependencies = [
"futures-sink",
"futures-util",
"pin-project-lite",
"slab",
"tokio",
]
@@ -6713,9 +6739,9 @@ checksum = "8df9b6e13f2d32c91b9bd719c00d1958837bc7dec474d94952798cc8e69eeec3"
[[package]]
name = "tracing"
version = "0.1.41"
version = "0.1.43"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "784e0ac535deb450455cbfa28a6f0df145ea1bb7ae51b821cf5e7927fdcfbdd0"
checksum = "2d15d90a0b5c19378952d479dc858407149d7bb45a14de0142f6c534b16fc647"
dependencies = [
"log",
"pin-project-lite",
@@ -6737,9 +6763,9 @@ dependencies = [
[[package]]
name = "tracing-attributes"
version = "0.1.30"
version = "0.1.31"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "81383ab64e72a7a8b8e13130c49e3dab29def6d0c7d76a03087b3cf71c5c6903"
checksum = "7490cfa5ec963746568740651ac6781f701c9c5ea257c58e057f3ba8cf69e8da"
dependencies = [
"proc-macro2",
"quote",
@@ -6748,9 +6774,9 @@ dependencies = [
[[package]]
name = "tracing-core"
version = "0.1.34"
version = "0.1.35"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "b9d12581f227e93f094d3af2ae690a574abb8a2b9b7a96e7cfe9647b2b617678"
checksum = "7a04e24fab5c89c6a36eb8558c9656f30d81de51dfa4d3b45f26b21d61fa0a6c"
dependencies = [
"once_cell",
"valuable",
@@ -6979,6 +7005,12 @@ version = "0.2.6"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ebc1c04c71510c7f702b52b7c350734c9ff1295c464a03335b00bb84fc54f853"
[[package]]
name = "unsafe-libyaml"
version = "0.2.11"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "673aac59facbab8a9007c7f6108d11f63b603f7cabff99fabf650fea5c32b861"
[[package]]
name = "untrusted"
version = "0.9.0"
@@ -7368,7 +7400,7 @@ version = "0.1.9"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "cf221c93e13a30d793f7645a0e7762c55d169dbb0a49671918a2319d289b10bb"
dependencies = [
"windows-sys 0.52.0",
"windows-sys 0.59.0",
]
[[package]]
@@ -8093,13 +8125,28 @@ version = "0.4.12"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "3f423a2c17029964870cfaabb1f13dfab7d092a62a29a89264f4d36990ca414a"
[[package]]
name = "zune-core"
version = "0.5.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "111f7d9820f05fd715df3144e254d6fc02ee4088b0644c0ffd0efc9e6d9d2773"
[[package]]
name = "zune-jpeg"
version = "0.4.19"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "2c9e525af0a6a658e031e95f14b7f889976b74a11ba0eca5a5fc9ac8a1c43a6a"
dependencies = [
"zune-core",
"zune-core 0.4.12",
]
[[package]]
name = "zune-jpeg"
version = "0.5.5"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "dc6fb7703e32e9a07fb3f757360338b3a567a5054f21b5f52a666752e333d58e"
dependencies = [
"zune-core 0.5.0",
]
[[package]]

View File

@@ -59,15 +59,15 @@ license = "Apache-2.0"
# Internal
app_test_support = { path = "app-server/tests/common" }
codex-ansi-escape = { path = "ansi-escape" }
codex-api = { path = "codex-api" }
codex-app-server = { path = "app-server" }
codex-app-server-protocol = { path = "app-server-protocol" }
codex-apply-patch = { path = "apply-patch" }
codex-arg0 = { path = "arg0" }
codex-async-utils = { path = "async-utils" }
codex-backend-client = { path = "backend-client" }
codex-api = { path = "codex-api" }
codex-client = { path = "codex-client" }
codex-chatgpt = { path = "chatgpt" }
codex-client = { path = "codex-client" }
codex-common = { path = "common" }
codex-core = { path = "core" }
codex-exec = { path = "exec" }
@@ -136,7 +136,7 @@ icu_decimal = "2.1"
icu_locale_core = "2.1"
icu_provider = { version = "2.1", features = ["sync"] }
ignore = "0.4.23"
image = { version = "^0.25.8", default-features = false }
image = { version = "^0.25.9", default-features = false }
indexmap = "2.12.0"
insta = "1.43.2"
itertools = "0.14.0"
@@ -169,16 +169,17 @@ pulldown-cmark = "0.10"
rand = "0.9"
ratatui = "0.29.0"
ratatui-macros = "0.6.0"
regex-lite = "0.1.7"
regex = "1.12.2"
regex-lite = "0.1.7"
reqwest = "0.12"
rmcp = { version = "0.9.0", default-features = false }
rmcp = { version = "0.10.0", default-features = false }
schemars = "0.8.22"
seccompiler = "0.5.0"
sentry = "0.34.0"
serde = "1"
serde_json = "1"
serde_with = "3.14"
serde_yaml = "0.9"
serde_with = "3.16"
serial_test = "3.2.0"
sha1 = "0.10.6"
sha2 = "0.10"
@@ -203,7 +204,7 @@ tokio-util = "0.7.16"
toml = "0.9.5"
toml_edit = "0.23.5"
tonic = "0.13.1"
tracing = "0.1.41"
tracing = "0.1.43"
tracing-appender = "0.2.3"
tracing-subscriber = "0.3.20"
tracing-test = "0.2.5"
@@ -288,7 +289,6 @@ opt-level = 0
# ratatui = { path = "../../ratatui" }
crossterm = { git = "https://github.com/nornagon/crossterm", branch = "nornagon/color-query" }
ratatui = { git = "https://github.com/nornagon/ratatui", branch = "nornagon-v0.29.0-patch" }
rmcp = { git = "https://github.com/bolinfest/rust-sdk", branch = "pr556" }
# Uncomment to debug local changes.
# rmcp = { path = "../../rust-sdk/crates/rmcp" }

View File

@@ -131,7 +131,7 @@ client_request_definitions! {
},
ReviewStart => "review/start" {
params: v2::ReviewStartParams,
response: v2::TurnStartResponse,
response: v2::ReviewStartResponse,
},
ModelList => "model/list" {
@@ -164,6 +164,12 @@ client_request_definitions! {
response: v2::FeedbackUploadResponse,
},
/// Execute a command (argv vector) under the server's sandbox.
OneOffCommandExec => "command/exec" {
params: v2::CommandExecParams,
response: v2::CommandExecResponse,
},
ConfigRead => "config/read" {
params: v2::ConfigReadParams,
response: v2::ConfigReadResponse,
@@ -506,10 +512,12 @@ server_notification_definitions! {
TurnStarted => "turn/started" (v2::TurnStartedNotification),
TurnCompleted => "turn/completed" (v2::TurnCompletedNotification),
TurnDiffUpdated => "turn/diff/updated" (v2::TurnDiffUpdatedNotification),
TurnPlanUpdated => "turn/plan/updated" (v2::TurnPlanUpdatedNotification),
ItemStarted => "item/started" (v2::ItemStartedNotification),
ItemCompleted => "item/completed" (v2::ItemCompletedNotification),
AgentMessageDelta => "item/agentMessage/delta" (v2::AgentMessageDeltaNotification),
CommandExecutionOutputDelta => "item/commandExecution/outputDelta" (v2::CommandExecutionOutputDeltaNotification),
FileChangeOutputDelta => "item/fileChange/outputDelta" (v2::FileChangeOutputDeltaNotification),
McpToolCallProgress => "item/mcpToolCall/progress" (v2::McpToolCallProgressNotification),
AccountUpdated => "account/updated" (v2::AccountUpdatedNotification),
AccountRateLimitsUpdated => "account/rateLimits/updated" (v2::AccountRateLimitsUpdatedNotification),

View File

@@ -0,0 +1,15 @@
use crate::protocol::v1;
use crate::protocol::v2;
impl From<v1::ExecOneOffCommandParams> for v2::CommandExecParams {
fn from(value: v1::ExecOneOffCommandParams) -> Self {
Self {
command: value.command,
timeout_ms: value
.timeout_ms
.map(|timeout| i64::try_from(timeout).unwrap_or(60_000)),
cwd: value.cwd,
sandbox_policy: value.sandbox_policy.map(std::convert::Into::into),
}
}
}

View File

@@ -2,6 +2,7 @@
// Exposes protocol pieces used by `lib.rs` via `pub use protocol::common::*;`.
pub mod common;
mod mappers;
pub mod thread_history;
pub mod v1;
pub mod v2;

View File

@@ -1,5 +1,6 @@
use crate::protocol::v2::ThreadItem;
use crate::protocol::v2::Turn;
use crate::protocol::v2::TurnError;
use crate::protocol::v2::TurnStatus;
use crate::protocol::v2::UserInput;
use codex_protocol::protocol::AgentReasoningEvent;
@@ -142,6 +143,7 @@ impl ThreadHistoryBuilder {
PendingTurn {
id: self.next_turn_id(),
items: Vec::new(),
error: None,
status: TurnStatus::Completed,
}
}
@@ -190,6 +192,7 @@ impl ThreadHistoryBuilder {
struct PendingTurn {
id: String,
items: Vec<ThreadItem>,
error: Option<TurnError>,
status: TurnStatus,
}
@@ -198,6 +201,7 @@ impl From<PendingTurn> for Turn {
Self {
id: value.id,
items: value.items,
error: value.error,
status: value.status,
}
}

View File

@@ -11,6 +11,8 @@ use codex_protocol::items::AgentMessageContent as CoreAgentMessageContent;
use codex_protocol::items::TurnItem as CoreTurnItem;
use codex_protocol::models::ResponseItem;
use codex_protocol::parse_command::ParsedCommand as CoreParsedCommand;
use codex_protocol::plan_tool::PlanItemArg as CorePlanItemArg;
use codex_protocol::plan_tool::StepStatus as CorePlanStepStatus;
use codex_protocol::protocol::CodexErrorInfo as CoreCodexErrorInfo;
use codex_protocol::protocol::CreditsSnapshot as CoreCreditsSnapshot;
use codex_protocol::protocol::RateLimitSnapshot as CoreRateLimitSnapshot;
@@ -130,6 +132,12 @@ v2_enum_from_core!(
}
);
v2_enum_from_core!(
pub enum ReviewDelivery from codex_protocol::protocol::ReviewDelivery {
Inline, Detached
}
);
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Eq, JsonSchema, TS)]
#[serde(rename_all = "camelCase")]
#[ts(export_to = "v2/")]
@@ -624,6 +632,26 @@ pub struct FeedbackUploadResponse {
pub thread_id: String,
}
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
#[serde(rename_all = "camelCase")]
#[ts(export_to = "v2/")]
pub struct CommandExecParams {
pub command: Vec<String>,
#[ts(type = "number | null")]
pub timeout_ms: Option<i64>,
pub cwd: Option<PathBuf>,
pub sandbox_policy: Option<SandboxPolicy>,
}
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
#[serde(rename_all = "camelCase")]
#[ts(export_to = "v2/")]
pub struct CommandExecResponse {
pub exit_code: i32,
pub stdout: String,
pub stderr: String,
}
// === Threads, Turns, and Items ===
// Thread APIs
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Default, JsonSchema, TS)]
@@ -758,6 +786,7 @@ pub struct Thread {
/// Model provider used for this thread (for example, 'openai').
pub model_provider: String,
/// Unix timestamp (in seconds) when the thread was created.
#[ts(type = "number")]
pub created_at: i64,
/// [UNSTABLE] Path to the thread on disk.
pub path: PathBuf,
@@ -848,8 +877,9 @@ pub struct Turn {
/// For all other responses and notifications returning a Turn,
/// the items field will be an empty list.
pub items: Vec<ThreadItem>,
#[serde(flatten)]
pub status: TurnStatus,
/// Only populated when the Turn's status is failed.
pub error: Option<TurnError>,
}
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS, Error)]
@@ -871,12 +901,12 @@ pub struct ErrorNotification {
}
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
#[serde(tag = "status", rename_all = "camelCase")]
#[ts(tag = "status", export_to = "v2/")]
#[serde(rename_all = "camelCase")]
#[ts(export_to = "v2/")]
pub enum TurnStatus {
Completed,
Interrupted,
Failed { error: TurnError },
Failed,
InProgress,
}
@@ -908,9 +938,22 @@ pub struct ReviewStartParams {
pub thread_id: String,
pub target: ReviewTarget,
/// When true, also append the final review message to the original thread.
/// Where to run the review: inline (default) on the current thread or
/// detached on a new thread (returned in `reviewThreadId`).
#[serde(default)]
pub append_to_original_thread: bool,
pub delivery: Option<ReviewDelivery>,
}
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
#[serde(rename_all = "camelCase")]
#[ts(export_to = "v2/")]
pub struct ReviewStartResponse {
pub turn: Turn,
/// Identifies the thread where the review runs.
///
/// For inline reviews, this is the original thread id.
/// For detached reviews, this is the id of the new review thread.
pub review_thread_id: String,
}
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
@@ -1020,6 +1063,8 @@ pub enum ThreadItem {
command: String,
/// The command's working directory.
cwd: PathBuf,
/// Identifier for the underlying PTY process (when available).
process_id: Option<String>,
status: CommandExecutionStatus,
/// A best-effort parsing of the command to understand the action(s) it will perform.
/// This returns a list of CommandAction objects because a single shell command may
@@ -1030,6 +1075,7 @@ pub enum ThreadItem {
/// The command's exit code.
exit_code: Option<i32>,
/// The duration of the command execution in milliseconds.
#[ts(type = "number | null")]
duration_ms: Option<i64>,
},
#[serde(rename_all = "camelCase")]
@@ -1061,7 +1107,10 @@ pub enum ThreadItem {
ImageView { id: String, path: String },
#[serde(rename_all = "camelCase")]
#[ts(rename_all = "camelCase")]
CodeReview { id: String, review: String },
EnteredReviewMode { id: String, review: String },
#[serde(rename_all = "camelCase")]
#[ts(rename_all = "camelCase")]
ExitedReviewMode { id: String, review: String },
}
impl From<CoreTurnItem> for ThreadItem {
@@ -1206,10 +1255,56 @@ pub struct TurnCompletedNotification {
/// Notification that the turn-level unified diff has changed.
/// Contains the latest aggregated diff across all file changes in the turn.
pub struct TurnDiffUpdatedNotification {
pub thread_id: String,
pub turn_id: String,
pub diff: String,
}
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
#[serde(rename_all = "camelCase")]
#[ts(export_to = "v2/")]
pub struct TurnPlanUpdatedNotification {
pub turn_id: String,
pub explanation: Option<String>,
pub plan: Vec<TurnPlanStep>,
}
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Eq, JsonSchema, TS)]
#[serde(rename_all = "camelCase")]
#[ts(export_to = "v2/")]
pub struct TurnPlanStep {
pub step: String,
pub status: TurnPlanStepStatus,
}
#[derive(Serialize, Deserialize, Debug, Clone, Copy, PartialEq, Eq, JsonSchema, TS)]
#[serde(rename_all = "camelCase")]
#[ts(export_to = "v2/")]
pub enum TurnPlanStepStatus {
Pending,
InProgress,
Completed,
}
impl From<CorePlanItemArg> for TurnPlanStep {
fn from(value: CorePlanItemArg) -> Self {
Self {
step: value.step,
status: value.status.into(),
}
}
}
impl From<CorePlanStepStatus> for TurnPlanStepStatus {
fn from(value: CorePlanStepStatus) -> Self {
match value {
CorePlanStepStatus::Pending => Self::Pending,
CorePlanStepStatus::InProgress => Self::InProgress,
CorePlanStepStatus::Completed => Self::Completed,
}
}
}
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
#[serde(rename_all = "camelCase")]
#[ts(export_to = "v2/")]
@@ -1233,6 +1328,8 @@ pub struct ItemCompletedNotification {
#[serde(rename_all = "camelCase")]
#[ts(export_to = "v2/")]
pub struct AgentMessageDeltaNotification {
pub thread_id: String,
pub turn_id: String,
pub item_id: String,
pub delta: String,
}
@@ -1241,8 +1338,11 @@ pub struct AgentMessageDeltaNotification {
#[serde(rename_all = "camelCase")]
#[ts(export_to = "v2/")]
pub struct ReasoningSummaryTextDeltaNotification {
pub thread_id: String,
pub turn_id: String,
pub item_id: String,
pub delta: String,
#[ts(type = "number")]
pub summary_index: i64,
}
@@ -1250,7 +1350,10 @@ pub struct ReasoningSummaryTextDeltaNotification {
#[serde(rename_all = "camelCase")]
#[ts(export_to = "v2/")]
pub struct ReasoningSummaryPartAddedNotification {
pub thread_id: String,
pub turn_id: String,
pub item_id: String,
#[ts(type = "number")]
pub summary_index: i64,
}
@@ -1258,8 +1361,11 @@ pub struct ReasoningSummaryPartAddedNotification {
#[serde(rename_all = "camelCase")]
#[ts(export_to = "v2/")]
pub struct ReasoningTextDeltaNotification {
pub thread_id: String,
pub turn_id: String,
pub item_id: String,
pub delta: String,
#[ts(type = "number")]
pub content_index: i64,
}
@@ -1267,6 +1373,18 @@ pub struct ReasoningTextDeltaNotification {
#[serde(rename_all = "camelCase")]
#[ts(export_to = "v2/")]
pub struct CommandExecutionOutputDeltaNotification {
pub thread_id: String,
pub turn_id: String,
pub item_id: String,
pub delta: String,
}
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, JsonSchema, TS)]
#[serde(rename_all = "camelCase")]
#[ts(export_to = "v2/")]
pub struct FileChangeOutputDeltaNotification {
pub thread_id: String,
pub turn_id: String,
pub item_id: String,
pub delta: String,
}
@@ -1275,6 +1393,8 @@ pub struct CommandExecutionOutputDeltaNotification {
#[serde(rename_all = "camelCase")]
#[ts(export_to = "v2/")]
pub struct McpToolCallProgressNotification {
pub thread_id: String,
pub turn_id: String,
pub item_id: String,
pub message: String,
}
@@ -1380,7 +1500,9 @@ impl From<CoreRateLimitSnapshot> for RateLimitSnapshot {
#[ts(export_to = "v2/")]
pub struct RateLimitWindow {
pub used_percent: i32,
#[ts(type = "number | null")]
pub window_duration_mins: Option<i64>,
#[ts(type = "number | null")]
pub resets_at: Option<i64>,
}

View File

@@ -563,7 +563,9 @@ impl CodexClient {
ServerNotification::TurnCompleted(payload) => {
if payload.turn.id == turn_id {
println!("\n< turn/completed notification: {:?}", payload.turn.status);
if let TurnStatus::Failed { error } = &payload.turn.status {
if payload.turn.status == TurnStatus::Failed
&& let Some(error) = payload.turn.error
{
println!("[turn error] {}", error.message);
}
break;

View File

@@ -1,6 +1,6 @@
# codex-app-server
`codex app-server` is the interface Codex uses to power rich interfaces such as the [Codex VS Code extension](https://marketplace.visualstudio.com/items?itemName=openai.chatgpt). The message schema is currently unstable, but those who wish to build experimental UIs on top of Codex may find it valuable.
`codex app-server` is the interface Codex uses to power rich interfaces such as the [Codex VS Code extension](https://marketplace.visualstudio.com/items?itemName=openai.chatgpt).
## Table of Contents
- [Protocol](#protocol)
@@ -65,7 +65,8 @@ The JSON-RPC API exposes dedicated methods for managing Codex conversations. Thr
- `thread/archive` — move a threads rollout file into the archived directory; returns `{}` 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 Codexs automated reviewer for a thread; responds like `turn/start` and emits a `item/completed` notification with a `codeReview` item when results are ready.
- `review/start` — kick off Codexs 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.
- `command/exec` — run a single command under the server sandbox without starting a thread/turn (handy for utilities and validation).
### 1) Start or resume a thread
@@ -190,49 +191,75 @@ Use `review/start` to run Codexs reviewer on the currently checked-out projec
- `{"type":"baseBranch","branch":"main"}` — diff against the provided branchs upstream (see prompt for the exact `git merge-base`/`git diff` instructions Codex will run).
- `{"type":"commit","sha":"abc1234","title":"Optional subject"}` — review a specific commit.
- `{"type":"custom","instructions":"Free-form reviewer instructions"}` — fallback prompt equivalent to the legacy manual review request.
- `appendToOriginalThread` (bool, default `false`) — when `true`, Codex also records a final assistant-style message with the review summary in the original thread. When `false`, only the `codeReview` item is emitted for the review run and no extra message is added to the original thread.
- `delivery` (`"inline"` or `"detached"`, default `"inline"`) — where the review runs:
- `"inline"`: run the review as a new turn on the existing thread. The responses `reviewThreadId` equals the original `threadId`, and no new `thread/started` notification is emitted.
- `"detached"`: fork a new review thread from the parent conversation and run the review there. The responses `reviewThreadId` is the id of this new review thread, and the server emits a `thread/started` notification for it before streaming review items.
Example request/response:
```json
{ "method": "review/start", "id": 40, "params": {
"threadId": "thr_123",
"appendToOriginalThread": true,
"delivery": "inline",
"target": { "type": "commit", "sha": "1234567deadbeef", "title": "Polish tui colors" }
} }
{ "id": 40, "result": { "turn": {
"id": "turn_900",
"status": "inProgress",
"items": [
{ "type": "userMessage", "id": "turn_900", "content": [ { "type": "text", "text": "Review commit 1234567: Polish tui colors" } ] }
],
"error": null
} } }
{ "id": 40, "result": {
"turn": {
"id": "turn_900",
"status": "inProgress",
"items": [
{ "type": "userMessage", "id": "turn_900", "content": [ { "type": "text", "text": "Review commit 1234567: Polish tui colors" } ] }
],
"error": null
},
"reviewThreadId": "thr_123"
} }
```
For a detached review, use `"delivery": "detached"`. The response is the same shape, but `reviewThreadId` will be the id of the new review thread (different from the original `threadId`). The server also emits a `thread/started` notification for that new thread before streaming the review turn.
Codex streams the usual `turn/started` notification followed by an `item/started`
with the same `codeReview` item id so clients can show progress:
with an `enteredReviewMode` item so clients can show progress:
```json
{ "method": "item/started", "params": { "item": {
"type": "codeReview",
"type": "enteredReviewMode",
"id": "turn_900",
"review": "current changes"
} } }
```
When the reviewer finishes, the server emits `item/completed` containing the same
`codeReview` item with the final review text:
When the reviewer finishes, the server emits `item/started` and `item/completed`
containing an `exitedReviewMode` item with the final review text:
```json
{ "method": "item/completed", "params": { "item": {
"type": "codeReview",
"type": "exitedReviewMode",
"id": "turn_900",
"review": "Looks solid overall...\n\n- Prefer Stylize helpers — app.rs:10-20\n ..."
} } }
```
The `review` string is plain text that already bundles the overall explanation plus a bullet list for each structured finding (matching `ThreadItem::CodeReview` in the generated schema). Use this notification to render the reviewer output in your client.
The `review` string is plain text that already bundles the overall explanation plus a bullet list for each structured finding (matching `ThreadItem::ExitedReviewMode` in the generated schema). Use this notification to render the reviewer output in your client.
### 7) One-off command execution
Run a standalone command (argv vector) in the servers sandbox without creating a thread or turn:
```json
{ "method": "command/exec", "id": 32, "params": {
"command": ["ls", "-la"],
"cwd": "/Users/me/project", // optional; defaults to server cwd
"sandboxPolicy": { "type": "workspaceWrite" }, // optional; defaults to user config
"timeoutMs": 10000 // optional; ms timeout; defaults to server timeout
} }
{ "id": 32, "result": { "exitCode": 0, "stdout": "...", "stderr": "" } }
```
Notes:
- Empty `command` arrays are rejected.
- `sandboxPolicy` accepts the same shape used by `turn/start` (e.g., `dangerFullAccess`, `readOnly`, `workspaceWrite` with flags).
- When omitted, `timeoutMs` falls back to the server default.
## Events (work-in-progress)
@@ -244,6 +271,7 @@ The app-server streams JSON-RPC notifications while a turn is running. Each turn
- `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/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`.
Today both notifications carry an empty `items` array even when item events were streamed; rely on `item/*` notifications for the canonical item list until this is fixed.

View File

@@ -18,6 +18,7 @@ use codex_app_server_protocol::ContextCompactedNotification;
use codex_app_server_protocol::ErrorNotification;
use codex_app_server_protocol::ExecCommandApprovalParams;
use codex_app_server_protocol::ExecCommandApprovalResponse;
use codex_app_server_protocol::FileChangeOutputDeltaNotification;
use codex_app_server_protocol::FileChangeRequestApprovalParams;
use codex_app_server_protocol::FileChangeRequestApprovalResponse;
use codex_app_server_protocol::FileUpdateChange;
@@ -43,6 +44,8 @@ use codex_app_server_protocol::TurnCompletedNotification;
use codex_app_server_protocol::TurnDiffUpdatedNotification;
use codex_app_server_protocol::TurnError;
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_core::parse_command::shlex_join;
@@ -59,7 +62,9 @@ use codex_core::protocol::ReviewDecision;
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::plan_tool::UpdatePlanArgs;
use codex_protocol::protocol::ReviewOutputEvent;
use std::collections::HashMap;
use std::convert::TryFrom;
@@ -257,6 +262,8 @@ pub(crate) async fn apply_bespoke_event_handling(
}
EventMsg::AgentMessageContentDelta(event) => {
let notification = AgentMessageDeltaNotification {
thread_id: conversation_id.to_string(),
turn_id: event_turn_id.clone(),
item_id: event.item_id,
delta: event.delta,
};
@@ -275,6 +282,8 @@ pub(crate) async fn apply_bespoke_event_handling(
}
EventMsg::ReasoningContentDelta(event) => {
let notification = ReasoningSummaryTextDeltaNotification {
thread_id: conversation_id.to_string(),
turn_id: event_turn_id.clone(),
item_id: event.item_id,
delta: event.delta,
summary_index: event.summary_index,
@@ -287,6 +296,8 @@ pub(crate) async fn apply_bespoke_event_handling(
}
EventMsg::ReasoningRawContentDelta(event) => {
let notification = ReasoningTextDeltaNotification {
thread_id: conversation_id.to_string(),
turn_id: event_turn_id.clone(),
item_id: event.item_id,
delta: event.delta,
content_index: event.content_index,
@@ -297,6 +308,8 @@ pub(crate) async fn apply_bespoke_event_handling(
}
EventMsg::AgentReasoningSectionBreak(event) => {
let notification = ReasoningSummaryPartAddedNotification {
thread_id: conversation_id.to_string(),
turn_id: event_turn_id.clone(),
item_id: event.item_id,
summary_index: event.summary_index,
};
@@ -339,17 +352,51 @@ pub(crate) async fn apply_bespoke_event_handling(
}))
.await;
}
EventMsg::EnteredReviewMode(review_request) => {
let notification = ItemStartedNotification {
EventMsg::ViewImageToolCall(view_image_event) => {
let item = ThreadItem::ImageView {
id: view_image_event.call_id.clone(),
path: view_image_event.path.to_string_lossy().into_owned(),
};
let started = ItemStartedNotification {
thread_id: conversation_id.to_string(),
turn_id: event_turn_id.clone(),
item: ThreadItem::CodeReview {
id: event_turn_id.clone(),
review: review_request.user_facing_hint,
},
item: item.clone(),
};
outgoing
.send_server_notification(ServerNotification::ItemStarted(notification))
.send_server_notification(ServerNotification::ItemStarted(started))
.await;
let completed = ItemCompletedNotification {
thread_id: conversation_id.to_string(),
turn_id: event_turn_id.clone(),
item,
};
outgoing
.send_server_notification(ServerNotification::ItemCompleted(completed))
.await;
}
EventMsg::EnteredReviewMode(review_request) => {
let review = review_request
.user_facing_hint
.unwrap_or_else(|| review_prompts::user_facing_hint(&review_request.target));
let item = ThreadItem::EnteredReviewMode {
id: event_turn_id.clone(),
review,
};
let started = ItemStartedNotification {
thread_id: conversation_id.to_string(),
turn_id: event_turn_id.clone(),
item: item.clone(),
};
outgoing
.send_server_notification(ServerNotification::ItemStarted(started))
.await;
let completed = ItemCompletedNotification {
thread_id: conversation_id.to_string(),
turn_id: event_turn_id.clone(),
item,
};
outgoing
.send_server_notification(ServerNotification::ItemCompleted(completed))
.await;
}
EventMsg::ItemStarted(item_started_event) => {
@@ -375,21 +422,29 @@ pub(crate) async fn apply_bespoke_event_handling(
.await;
}
EventMsg::ExitedReviewMode(review_event) => {
let review_text = match review_event.review_output {
let review = match review_event.review_output {
Some(output) => render_review_output_text(&output),
None => REVIEW_FALLBACK_MESSAGE.to_string(),
};
let review_item_id = event_turn_id.clone();
let notification = ItemCompletedNotification {
let item = ThreadItem::ExitedReviewMode {
id: event_turn_id.clone(),
review,
};
let started = ItemStartedNotification {
thread_id: conversation_id.to_string(),
turn_id: event_turn_id.clone(),
item: ThreadItem::CodeReview {
id: review_item_id,
review: review_text,
},
item: item.clone(),
};
outgoing
.send_server_notification(ServerNotification::ItemCompleted(notification))
.send_server_notification(ServerNotification::ItemStarted(started))
.await;
let completed = ItemCompletedNotification {
thread_id: conversation_id.to_string(),
turn_id: event_turn_id.clone(),
item,
};
outgoing
.send_server_notification(ServerNotification::ItemCompleted(completed))
.await;
}
EventMsg::PatchApplyBegin(patch_begin_event) => {
@@ -449,11 +504,13 @@ pub(crate) async fn apply_bespoke_event_handling(
.collect::<Vec<_>>();
let command = shlex_join(&exec_command_begin_event.command);
let cwd = exec_command_begin_event.cwd;
let process_id = exec_command_begin_event.process_id;
let item = ThreadItem::CommandExecution {
id: item_id,
command,
cwd,
process_id,
status: CommandExecutionStatus::InProgress,
command_actions,
aggregated_output: None,
@@ -470,15 +527,44 @@ pub(crate) async fn apply_bespoke_event_handling(
.await;
}
EventMsg::ExecCommandOutputDelta(exec_command_output_delta_event) => {
let notification = CommandExecutionOutputDeltaNotification {
item_id: exec_command_output_delta_event.call_id.clone(),
delta: String::from_utf8_lossy(&exec_command_output_delta_event.chunk).to_string(),
let item_id = exec_command_output_delta_event.call_id.clone();
let delta = String::from_utf8_lossy(&exec_command_output_delta_event.chunk).to_string();
// The underlying EventMsg::ExecCommandOutputDelta is used for shell, unified_exec,
// and apply_patch tool calls. We represent apply_patch with the FileChange item, and
// everything else with the CommandExecution item.
//
// We need to detect which item type it is so we can emit the right notification.
// We already have state tracking FileChange items on item/started, so let's use that.
let is_file_change = {
let map = turn_summary_store.lock().await;
map.get(&conversation_id)
.is_some_and(|summary| summary.file_change_started.contains(&item_id))
};
outgoing
.send_server_notification(ServerNotification::CommandExecutionOutputDelta(
notification,
))
.await;
if is_file_change {
let notification = FileChangeOutputDeltaNotification {
thread_id: conversation_id.to_string(),
turn_id: event_turn_id.clone(),
item_id,
delta,
};
outgoing
.send_server_notification(ServerNotification::FileChangeOutputDelta(
notification,
))
.await;
} else {
let notification = CommandExecutionOutputDeltaNotification {
thread_id: conversation_id.to_string(),
turn_id: event_turn_id.clone(),
item_id,
delta,
};
outgoing
.send_server_notification(ServerNotification::CommandExecutionOutputDelta(
notification,
))
.await;
}
}
EventMsg::ExecCommandEnd(exec_command_end_event) => {
let ExecCommandEndEvent {
@@ -486,6 +572,7 @@ pub(crate) async fn apply_bespoke_event_handling(
command,
cwd,
parsed_cmd,
process_id,
aggregated_output,
exit_code,
duration,
@@ -514,6 +601,7 @@ pub(crate) async fn apply_bespoke_event_handling(
id: call_id,
command: shlex_join(&command),
cwd,
process_id,
status,
command_actions,
aggregated_output,
@@ -563,6 +651,7 @@ pub(crate) async fn apply_bespoke_event_handling(
}
EventMsg::TurnDiff(turn_diff_event) => {
handle_turn_diff(
conversation_id,
&event_turn_id,
turn_diff_event,
api_version,
@@ -570,12 +659,22 @@ pub(crate) async fn apply_bespoke_event_handling(
)
.await;
}
EventMsg::PlanUpdate(plan_update_event) => {
handle_turn_plan_update(
&event_turn_id,
plan_update_event,
api_version,
outgoing.as_ref(),
)
.await;
}
_ => {}
}
}
async fn handle_turn_diff(
conversation_id: ConversationId,
event_turn_id: &str,
turn_diff_event: TurnDiffEvent,
api_version: ApiVersion,
@@ -583,6 +682,7 @@ async fn handle_turn_diff(
) {
if let ApiVersion::V2 = api_version {
let notification = TurnDiffUpdatedNotification {
thread_id: conversation_id.to_string(),
turn_id: event_turn_id.to_string(),
diff: turn_diff_event.unified_diff,
};
@@ -592,10 +692,33 @@ async fn handle_turn_diff(
}
}
async fn handle_turn_plan_update(
event_turn_id: &str,
plan_update_event: UpdatePlanArgs,
api_version: ApiVersion,
outgoing: &OutgoingMessageSender,
) {
if let ApiVersion::V2 = api_version {
let notification = TurnPlanUpdatedNotification {
turn_id: event_turn_id.to_string(),
explanation: plan_update_event.explanation,
plan: plan_update_event
.plan
.into_iter()
.map(TurnPlanStep::from)
.collect(),
};
outgoing
.send_server_notification(ServerNotification::TurnPlanUpdated(notification))
.await;
}
}
async fn emit_turn_completed_with_status(
conversation_id: ConversationId,
event_turn_id: String,
status: TurnStatus,
error: Option<TurnError>,
outgoing: &OutgoingMessageSender,
) {
let notification = TurnCompletedNotification {
@@ -603,6 +726,7 @@ async fn emit_turn_completed_with_status(
turn: Turn {
id: event_turn_id,
items: vec![],
error,
status,
},
};
@@ -649,6 +773,7 @@ async fn complete_command_execution_item(
item_id: String,
command: String,
cwd: PathBuf,
process_id: Option<String>,
command_actions: Vec<V2ParsedCommand>,
status: CommandExecutionStatus,
outgoing: &OutgoingMessageSender,
@@ -657,6 +782,7 @@ async fn complete_command_execution_item(
id: item_id,
command,
cwd,
process_id,
status,
command_actions,
aggregated_output: None,
@@ -689,13 +815,12 @@ async fn handle_turn_complete(
) {
let turn_summary = find_and_remove_turn_summary(conversation_id, turn_summary_store).await;
let status = if let Some(error) = turn_summary.last_error {
TurnStatus::Failed { error }
} else {
TurnStatus::Completed
let (status, error) = match turn_summary.last_error {
Some(error) => (TurnStatus::Failed, Some(error)),
None => (TurnStatus::Completed, None),
};
emit_turn_completed_with_status(conversation_id, event_turn_id, status, outgoing).await;
emit_turn_completed_with_status(conversation_id, event_turn_id, status, error, outgoing).await;
}
async fn handle_turn_interrupted(
@@ -710,6 +835,7 @@ async fn handle_turn_interrupted(
conversation_id,
event_turn_id,
TurnStatus::Interrupted,
None,
outgoing,
)
.await;
@@ -1015,6 +1141,7 @@ async fn on_command_execution_request_approval_response(
item_id.clone(),
command.clone(),
cwd.clone(),
None,
command_actions.clone(),
status,
outgoing.as_ref(),
@@ -1108,12 +1235,15 @@ mod tests {
use anyhow::Result;
use anyhow::anyhow;
use anyhow::bail;
use codex_app_server_protocol::TurnPlanStepStatus;
use codex_core::protocol::CreditsSnapshot;
use codex_core::protocol::McpInvocation;
use codex_core::protocol::RateLimitSnapshot;
use codex_core::protocol::RateLimitWindow;
use codex_core::protocol::TokenUsage;
use codex_core::protocol::TokenUsageInfo;
use codex_protocol::plan_tool::PlanItemArg;
use codex_protocol::plan_tool::StepStatus;
use mcp_types::CallToolResult;
use mcp_types::ContentBlock;
use mcp_types::TextContent;
@@ -1178,6 +1308,7 @@ mod tests {
OutgoingMessage::AppServerNotification(ServerNotification::TurnCompleted(n)) => {
assert_eq!(n.turn.id, event_turn_id);
assert_eq!(n.turn.status, TurnStatus::Completed);
assert_eq!(n.turn.error, None);
}
other => bail!("unexpected message: {other:?}"),
}
@@ -1218,6 +1349,7 @@ mod tests {
OutgoingMessage::AppServerNotification(ServerNotification::TurnCompleted(n)) => {
assert_eq!(n.turn.id, event_turn_id);
assert_eq!(n.turn.status, TurnStatus::Interrupted);
assert_eq!(n.turn.error, None);
}
other => bail!("unexpected message: {other:?}"),
}
@@ -1257,14 +1389,13 @@ mod tests {
match msg {
OutgoingMessage::AppServerNotification(ServerNotification::TurnCompleted(n)) => {
assert_eq!(n.turn.id, event_turn_id);
assert_eq!(n.turn.status, TurnStatus::Failed);
assert_eq!(
n.turn.status,
TurnStatus::Failed {
error: TurnError {
message: "bad".to_string(),
codex_error_info: Some(V2CodexErrorInfo::Other),
}
}
n.turn.error,
Some(TurnError {
message: "bad".to_string(),
codex_error_info: Some(V2CodexErrorInfo::Other),
})
);
}
other => bail!("unexpected message: {other:?}"),
@@ -1273,6 +1404,46 @@ mod tests {
Ok(())
}
#[tokio::test]
async fn test_handle_turn_plan_update_emits_notification_for_v2() -> Result<()> {
let (tx, mut rx) = mpsc::channel(CHANNEL_CAPACITY);
let outgoing = OutgoingMessageSender::new(tx);
let update = UpdatePlanArgs {
explanation: Some("need plan".to_string()),
plan: vec![
PlanItemArg {
step: "first".to_string(),
status: StepStatus::Pending,
},
PlanItemArg {
step: "second".to_string(),
status: StepStatus::Completed,
},
],
};
handle_turn_plan_update("turn-123", update, ApiVersion::V2, &outgoing).await;
let msg = rx
.recv()
.await
.ok_or_else(|| anyhow!("should send one notification"))?;
match msg {
OutgoingMessage::AppServerNotification(ServerNotification::TurnPlanUpdated(n)) => {
assert_eq!(n.turn_id, "turn-123");
assert_eq!(n.explanation.as_deref(), Some("need plan"));
assert_eq!(n.plan.len(), 2);
assert_eq!(n.plan[0].step, "first");
assert_eq!(n.plan[0].status, TurnPlanStepStatus::Pending);
assert_eq!(n.plan[1].step, "second");
assert_eq!(n.plan[1].status, TurnPlanStepStatus::Completed);
}
other => bail!("unexpected message: {other:?}"),
}
assert!(rx.try_recv().is_err(), "no extra messages expected");
Ok(())
}
#[tokio::test]
async fn test_handle_token_count_event_emits_usage_and_rate_limits() -> Result<()> {
let conversation_id = ConversationId::new();
@@ -1485,14 +1656,13 @@ mod tests {
match msg {
OutgoingMessage::AppServerNotification(ServerNotification::TurnCompleted(n)) => {
assert_eq!(n.turn.id, a_turn1);
assert_eq!(n.turn.status, TurnStatus::Failed);
assert_eq!(
n.turn.status,
TurnStatus::Failed {
error: TurnError {
message: "a1".to_string(),
codex_error_info: Some(V2CodexErrorInfo::BadRequest),
}
}
n.turn.error,
Some(TurnError {
message: "a1".to_string(),
codex_error_info: Some(V2CodexErrorInfo::BadRequest),
})
);
}
other => bail!("unexpected message: {other:?}"),
@@ -1506,14 +1676,13 @@ mod tests {
match msg {
OutgoingMessage::AppServerNotification(ServerNotification::TurnCompleted(n)) => {
assert_eq!(n.turn.id, b_turn1);
assert_eq!(n.turn.status, TurnStatus::Failed);
assert_eq!(
n.turn.status,
TurnStatus::Failed {
error: TurnError {
message: "b1".to_string(),
codex_error_info: None,
}
}
n.turn.error,
Some(TurnError {
message: "b1".to_string(),
codex_error_info: None,
})
);
}
other => bail!("unexpected message: {other:?}"),
@@ -1528,6 +1697,7 @@ mod tests {
OutgoingMessage::AppServerNotification(ServerNotification::TurnCompleted(n)) => {
assert_eq!(n.turn.id, a_turn2);
assert_eq!(n.turn.status, TurnStatus::Completed);
assert_eq!(n.turn.error, None);
}
other => bail!("unexpected message: {other:?}"),
}
@@ -1672,8 +1842,10 @@ 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();
handle_turn_diff(
conversation_id,
"turn-1",
TurnDiffEvent {
unified_diff: unified_diff.clone(),
@@ -1691,6 +1863,7 @@ mod tests {
OutgoingMessage::AppServerNotification(ServerNotification::TurnDiffUpdated(
notification,
)) => {
assert_eq!(notification.thread_id, conversation_id.to_string());
assert_eq!(notification.turn_id, "turn-1");
assert_eq!(notification.diff, unified_diff);
}
@@ -1704,8 +1877,10 @@ 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();
handle_turn_diff(
conversation_id,
"turn-1",
TurnDiffEvent {
unified_diff: "diff".to_string(),

View File

@@ -21,9 +21,9 @@ use codex_app_server_protocol::CancelLoginAccountParams;
use codex_app_server_protocol::CancelLoginAccountResponse;
use codex_app_server_protocol::CancelLoginChatGptResponse;
use codex_app_server_protocol::ClientRequest;
use codex_app_server_protocol::CommandExecParams;
use codex_app_server_protocol::ConversationGitInfo;
use codex_app_server_protocol::ConversationSummary;
use codex_app_server_protocol::ExecOneOffCommandParams;
use codex_app_server_protocol::ExecOneOffCommandResponse;
use codex_app_server_protocol::FeedbackUploadParams;
use codex_app_server_protocol::FeedbackUploadResponse;
@@ -61,8 +61,10 @@ use codex_app_server_protocol::RemoveConversationSubscriptionResponse;
use codex_app_server_protocol::RequestId;
use codex_app_server_protocol::ResumeConversationParams;
use codex_app_server_protocol::ResumeConversationResponse;
use codex_app_server_protocol::ReviewDelivery as ApiReviewDelivery;
use codex_app_server_protocol::ReviewStartParams;
use codex_app_server_protocol::ReviewTarget;
use codex_app_server_protocol::ReviewStartResponse;
use codex_app_server_protocol::ReviewTarget as ApiReviewTarget;
use codex_app_server_protocol::SandboxMode;
use codex_app_server_protocol::SendUserMessageParams;
use codex_app_server_protocol::SendUserMessageResponse;
@@ -120,7 +122,9 @@ use codex_core::git_info::git_diff_to_remote;
use codex_core::parse_cursor;
use codex_core::protocol::EventMsg;
use codex_core::protocol::Op;
use codex_core::protocol::ReviewDelivery as CoreReviewDelivery;
use codex_core::protocol::ReviewRequest;
use codex_core::protocol::ReviewTarget as CoreReviewTarget;
use codex_core::protocol::SessionConfiguredEvent;
use codex_core::read_head_for_summary;
use codex_feedback::CodexFeedback;
@@ -148,7 +152,6 @@ use std::sync::Arc;
use std::sync::atomic::AtomicBool;
use std::sync::atomic::Ordering;
use std::time::Duration;
use tokio::select;
use tokio::sync::Mutex;
use tokio::sync::oneshot;
use tracing::error;
@@ -252,8 +255,7 @@ impl CodexMessageProcessor {
}
fn review_request_from_target(
target: ReviewTarget,
append_to_original_thread: bool,
target: ApiReviewTarget,
) -> Result<(ReviewRequest, String), JSONRPCErrorError> {
fn invalid_request(message: String) -> JSONRPCErrorError {
JSONRPCErrorError {
@@ -263,77 +265,52 @@ impl CodexMessageProcessor {
}
}
match target {
// TODO(jif) those messages will be extracted in a follow-up PR.
ReviewTarget::UncommittedChanges => Ok((
ReviewRequest {
prompt: "Review the current code changes (staged, unstaged, and untracked files) and provide prioritized findings.".to_string(),
user_facing_hint: "current changes".to_string(),
append_to_original_thread,
},
"Review uncommitted changes".to_string(),
)),
ReviewTarget::BaseBranch { branch } => {
let cleaned_target = match target {
ApiReviewTarget::UncommittedChanges => ApiReviewTarget::UncommittedChanges,
ApiReviewTarget::BaseBranch { branch } => {
let branch = branch.trim().to_string();
if branch.is_empty() {
return Err(invalid_request("branch must not be empty".to_string()));
}
let prompt = format!("Review the code changes against the base branch '{branch}'. Start by finding the merge diff between the current branch and {branch}'s upstream e.g. (`git merge-base HEAD \"$(git rev-parse --abbrev-ref \"{branch}@{{upstream}}\")\"`), then run `git diff` against that SHA to see what changes we would merge into the {branch} branch. Provide prioritized, actionable findings.");
let hint = format!("changes against '{branch}'");
let display = format!("Review changes against base branch '{branch}'");
Ok((
ReviewRequest {
prompt,
user_facing_hint: hint,
append_to_original_thread,
},
display,
))
ApiReviewTarget::BaseBranch { branch }
}
ReviewTarget::Commit { sha, title } => {
ApiReviewTarget::Commit { sha, title } => {
let sha = sha.trim().to_string();
if sha.is_empty() {
return Err(invalid_request("sha must not be empty".to_string()));
}
let brief_title = title
let title = title
.map(|t| t.trim().to_string())
.filter(|t| !t.is_empty());
let prompt = if let Some(title) = brief_title.clone() {
format!("Review the code changes introduced by commit {sha} (\"{title}\"). Provide prioritized, actionable findings.")
} else {
format!("Review the code changes introduced by commit {sha}. Provide prioritized, actionable findings.")
};
let short_sha = sha.chars().take(7).collect::<String>();
let hint = format!("commit {short_sha}");
let display = if let Some(title) = brief_title {
format!("Review commit {short_sha}: {title}")
} else {
format!("Review commit {short_sha}")
};
Ok((
ReviewRequest {
prompt,
user_facing_hint: hint,
append_to_original_thread,
},
display,
))
ApiReviewTarget::Commit { sha, title }
}
ReviewTarget::Custom { instructions } => {
ApiReviewTarget::Custom { instructions } => {
let trimmed = instructions.trim().to_string();
if trimmed.is_empty() {
return Err(invalid_request("instructions must not be empty".to_string()));
return Err(invalid_request(
"instructions must not be empty".to_string(),
));
}
ApiReviewTarget::Custom {
instructions: trimmed,
}
Ok((
ReviewRequest {
prompt: trimmed.clone(),
user_facing_hint: trimmed.clone(),
append_to_original_thread,
},
trimmed,
))
}
}
};
let core_target = match cleaned_target {
ApiReviewTarget::UncommittedChanges => CoreReviewTarget::UncommittedChanges,
ApiReviewTarget::BaseBranch { branch } => CoreReviewTarget::BaseBranch { branch },
ApiReviewTarget::Commit { sha, title } => CoreReviewTarget::Commit { sha, title },
ApiReviewTarget::Custom { instructions } => CoreReviewTarget::Custom { instructions },
};
let hint = codex_core::review_prompts::user_facing_hint(&core_target);
let review_request = ReviewRequest {
target: core_target,
user_facing_hint: Some(hint.clone()),
};
Ok((review_request, hint))
}
pub async fn process_request(&mut self, request: ClientRequest) {
@@ -469,9 +446,12 @@ impl CodexMessageProcessor {
ClientRequest::FuzzyFileSearch { request_id, params } => {
self.fuzzy_file_search(request_id, params).await;
}
ClientRequest::ExecOneOffCommand { request_id, params } => {
ClientRequest::OneOffCommandExec { request_id, params } => {
self.exec_one_off_command(request_id, params).await;
}
ClientRequest::ExecOneOffCommand { request_id, params } => {
self.exec_one_off_command(request_id, params.into()).await;
}
ClientRequest::ConfigRead { .. }
| ClientRequest::ConfigValueWrite { .. }
| ClientRequest::ConfigBatchWrite { .. } => {
@@ -1156,7 +1136,7 @@ impl CodexMessageProcessor {
}
}
async fn exec_one_off_command(&self, request_id: RequestId, params: ExecOneOffCommandParams) {
async fn exec_one_off_command(&self, request_id: RequestId, params: CommandExecParams) {
tracing::debug!("ExecOneOffCommand params: {params:?}");
if params.command.is_empty() {
@@ -1171,7 +1151,9 @@ impl CodexMessageProcessor {
let cwd = params.cwd.unwrap_or_else(|| self.config.cwd.clone());
let env = create_env(&self.config.shell_environment_policy);
let timeout_ms = params.timeout_ms;
let timeout_ms = params
.timeout_ms
.and_then(|timeout_ms| u64::try_from(timeout_ms).ok());
let exec_params = ExecParams {
command: params.command,
cwd,
@@ -1184,6 +1166,7 @@ impl CodexMessageProcessor {
let effective_policy = params
.sandbox_policy
.map(|policy| policy.to_core())
.unwrap_or_else(|| self.config.sandbox_policy.clone());
let codex_linux_sandbox_exe = self.config.codex_linux_sandbox_exe.clone();
@@ -2228,51 +2211,25 @@ impl CodexMessageProcessor {
.await
{
info!("conversation {conversation_id} was active; shutting down");
let conversation_clone = conversation.clone();
let notify = Arc::new(tokio::sync::Notify::new());
let notify_clone = notify.clone();
// Do not wait on conversation.next_event(); the listener task already consumes
// the stream. Request shutdown and ensure the rollout file is flushed before moving it.
if let Err(err) = conversation.submit(Op::Shutdown).await {
error!("failed to submit Shutdown to conversation {conversation_id}: {err}");
}
// Establish the listener for ShutdownComplete before submitting
// Shutdown so it is not missed.
let is_shutdown = tokio::spawn(async move {
// Create the notified future outside the loop to avoid losing notifications.
let notified = notify_clone.notified();
tokio::pin!(notified);
loop {
select! {
_ = &mut notified => { break; }
event = conversation_clone.next_event() => {
match event {
Ok(event) => {
if matches!(event.msg, EventMsg::ShutdownComplete) { break; }
}
// Break on errors to avoid tight loops when the agent loop has exited.
Err(_) => { break; }
}
}
}
let flush_result =
tokio::time::timeout(Duration::from_secs(5), conversation.flush_rollout()).await;
match flush_result {
Ok(Ok(())) => {}
Ok(Err(err)) => {
warn!(
"conversation {conversation_id} rollout flush failed before archive: {err}"
);
}
});
// Request shutdown.
match conversation.submit(Op::Shutdown).await {
Ok(_) => {
// Successfully submitted Shutdown; wait before proceeding.
select! {
_ = is_shutdown => {
// Normal shutdown: proceed with archive.
}
_ = tokio::time::sleep(Duration::from_secs(10)) => {
warn!("conversation {conversation_id} shutdown timed out; proceeding with archive");
// Wake any waiter; use notify_waiters to avoid missing the signal.
notify.notify_waiters();
// Perhaps we lost a shutdown race, so let's continue to
// clean up the .jsonl file.
}
}
}
Err(err) => {
error!("failed to submit Shutdown to conversation {conversation_id}: {err}");
notify.notify_waiters();
Err(_) => {
warn!(
"conversation {conversation_id} rollout flush timed out; proceeding with archive"
);
}
}
}
@@ -2284,7 +2241,8 @@ impl CodexMessageProcessor {
.codex_home
.join(codex_core::ARCHIVED_SESSIONS_SUBDIR);
tokio::fs::create_dir_all(&archive_folder).await?;
tokio::fs::rename(&canonical_rollout_path, &archive_folder.join(&file_name)).await?;
let destination = archive_folder.join(&file_name);
tokio::fs::rename(&canonical_rollout_path, &destination).await?;
Ok(())
}
.await;
@@ -2471,6 +2429,7 @@ impl CodexMessageProcessor {
let turn = Turn {
id: turn_id.clone(),
items: vec![],
error: None,
status: TurnStatus::InProgress,
};
@@ -2497,60 +2456,221 @@ impl CodexMessageProcessor {
}
}
async fn review_start(&self, request_id: RequestId, params: ReviewStartParams) {
fn build_review_turn(turn_id: String, display_text: &str) -> Turn {
let items = if display_text.is_empty() {
Vec::new()
} else {
vec![ThreadItem::UserMessage {
id: turn_id.clone(),
content: vec![V2UserInput::Text {
text: display_text.to_string(),
}],
}]
};
Turn {
id: turn_id,
items,
error: None,
status: TurnStatus::InProgress,
}
}
async fn emit_review_started(
&self,
request_id: &RequestId,
turn: Turn,
parent_thread_id: String,
review_thread_id: String,
) {
let response = ReviewStartResponse {
turn: turn.clone(),
review_thread_id,
};
self.outgoing
.send_response(request_id.clone(), response)
.await;
let notif = TurnStartedNotification {
thread_id: parent_thread_id,
turn,
};
self.outgoing
.send_server_notification(ServerNotification::TurnStarted(notif))
.await;
}
async fn start_inline_review(
&self,
request_id: &RequestId,
parent_conversation: Arc<CodexConversation>,
review_request: ReviewRequest,
display_text: &str,
parent_thread_id: String,
) -> std::result::Result<(), JSONRPCErrorError> {
let turn_id = parent_conversation
.submit(Op::Review { review_request })
.await;
match turn_id {
Ok(turn_id) => {
let turn = Self::build_review_turn(turn_id, display_text);
self.emit_review_started(
request_id,
turn,
parent_thread_id.clone(),
parent_thread_id,
)
.await;
Ok(())
}
Err(err) => Err(JSONRPCErrorError {
code: INTERNAL_ERROR_CODE,
message: format!("failed to start review: {err}"),
data: None,
}),
}
}
async fn start_detached_review(
&mut self,
request_id: &RequestId,
parent_conversation_id: ConversationId,
review_request: ReviewRequest,
display_text: &str,
) -> std::result::Result<(), JSONRPCErrorError> {
let rollout_path = find_conversation_path_by_id_str(
&self.config.codex_home,
&parent_conversation_id.to_string(),
)
.await
.map_err(|err| JSONRPCErrorError {
code: INTERNAL_ERROR_CODE,
message: format!("failed to locate conversation id {parent_conversation_id}: {err}"),
data: None,
})?
.ok_or_else(|| JSONRPCErrorError {
code: INVALID_REQUEST_ERROR_CODE,
message: format!("no rollout found for conversation id {parent_conversation_id}"),
data: None,
})?;
let mut config = self.config.as_ref().clone();
config.model = self.config.review_model.clone();
let NewConversation {
conversation_id,
conversation,
session_configured,
..
} = self
.conversation_manager
.fork_conversation(usize::MAX, config, rollout_path)
.await
.map_err(|err| JSONRPCErrorError {
code: INTERNAL_ERROR_CODE,
message: format!("error creating detached review conversation: {err}"),
data: None,
})?;
if let Err(err) = self
.attach_conversation_listener(conversation_id, false, ApiVersion::V2)
.await
{
tracing::warn!(
"failed to attach listener for review conversation {}: {}",
conversation_id,
err.message
);
}
let rollout_path = conversation.rollout_path();
let fallback_provider = self.config.model_provider_id.as_str();
match read_summary_from_rollout(rollout_path.as_path(), fallback_provider).await {
Ok(summary) => {
let thread = summary_to_thread(summary);
let notif = ThreadStartedNotification { thread };
self.outgoing
.send_server_notification(ServerNotification::ThreadStarted(notif))
.await;
}
Err(err) => {
tracing::warn!(
"failed to load summary for review conversation {}: {}",
session_configured.session_id,
err
);
}
}
let turn_id = conversation
.submit(Op::Review { review_request })
.await
.map_err(|err| JSONRPCErrorError {
code: INTERNAL_ERROR_CODE,
message: format!("failed to start detached review turn: {err}"),
data: None,
})?;
let turn = Self::build_review_turn(turn_id, display_text);
let review_thread_id = conversation_id.to_string();
self.emit_review_started(request_id, turn, review_thread_id.clone(), review_thread_id)
.await;
Ok(())
}
async fn review_start(&mut self, request_id: RequestId, params: ReviewStartParams) {
let ReviewStartParams {
thread_id,
target,
append_to_original_thread,
delivery,
} = params;
let (_, conversation) = match self.conversation_from_thread_id(&thread_id).await {
Ok(v) => v,
Err(error) => {
self.outgoing.send_error(request_id, error).await;
return;
}
};
let (review_request, display_text) =
match Self::review_request_from_target(target, append_to_original_thread) {
Ok(value) => value,
Err(err) => {
self.outgoing.send_error(request_id, err).await;
let (parent_conversation_id, parent_conversation) =
match self.conversation_from_thread_id(&thread_id).await {
Ok(v) => v,
Err(error) => {
self.outgoing.send_error(request_id, error).await;
return;
}
};
let turn_id = conversation.submit(Op::Review { review_request }).await;
match turn_id {
Ok(turn_id) => {
let mut items = Vec::new();
if !display_text.is_empty() {
items.push(ThreadItem::UserMessage {
id: turn_id.clone(),
content: vec![V2UserInput::Text { text: display_text }],
});
}
let turn = Turn {
id: turn_id.clone(),
items,
status: TurnStatus::InProgress,
};
let response = TurnStartResponse { turn: turn.clone() };
self.outgoing.send_response(request_id, response).await;
let notif = TurnStartedNotification { thread_id, turn };
self.outgoing
.send_server_notification(ServerNotification::TurnStarted(notif))
.await;
}
let (review_request, display_text) = match Self::review_request_from_target(target) {
Ok(value) => value,
Err(err) => {
let error = JSONRPCErrorError {
code: INTERNAL_ERROR_CODE,
message: format!("failed to start review: {err}"),
data: None,
};
self.outgoing.send_error(request_id, error).await;
self.outgoing.send_error(request_id, err).await;
return;
}
};
let delivery = delivery.unwrap_or(ApiReviewDelivery::Inline).to_core();
match delivery {
CoreReviewDelivery::Inline => {
if let Err(err) = self
.start_inline_review(
&request_id,
parent_conversation,
review_request,
display_text.as_str(),
thread_id.clone(),
)
.await
{
self.outgoing.send_error(request_id, err).await;
}
}
CoreReviewDelivery::Detached => {
if let Err(err) = self
.start_detached_review(
&request_id,
parent_conversation_id,
review_request,
display_text.as_str(),
)
.await
{
self.outgoing.send_error(request_id, err).await;
}
}
}
}

View File

@@ -15,6 +15,7 @@ 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 responses::create_apply_patch_sse_response;
pub use responses::create_exec_command_sse_response;
pub use responses::create_final_assistant_message_sse_response;
pub use responses::create_shell_command_sse_response;
pub use rollout::create_fake_rollout;

View File

@@ -94,3 +94,42 @@ pub fn create_apply_patch_sse_response(
);
Ok(sse)
}
pub fn create_exec_command_sse_response(call_id: &str) -> anyhow::Result<String> {
let (cmd, args) = if cfg!(windows) {
("cmd.exe", vec!["/d", "/c", "echo hi"])
} else {
("/bin/sh", vec!["-c", "echo hi"])
};
let command = std::iter::once(cmd.to_string())
.chain(args.into_iter().map(str::to_string))
.collect::<Vec<_>>();
let tool_call_arguments = serde_json::to_string(&json!({
"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)
}

View File

@@ -49,6 +49,7 @@ pub fn create_fake_rollout(
instructions: None,
source: SessionSource::Cli,
model_provider: model_provider.map(str::to_string),
name: None,
};
let payload = serde_json::to_value(SessionMetaLine {
meta,

View File

@@ -9,12 +9,13 @@ use codex_app_server_protocol::JSONRPCError;
use codex_app_server_protocol::JSONRPCNotification;
use codex_app_server_protocol::JSONRPCResponse;
use codex_app_server_protocol::RequestId;
use codex_app_server_protocol::ReviewDelivery;
use codex_app_server_protocol::ReviewStartParams;
use codex_app_server_protocol::ReviewStartResponse;
use codex_app_server_protocol::ReviewTarget;
use codex_app_server_protocol::ThreadItem;
use codex_app_server_protocol::ThreadStartParams;
use codex_app_server_protocol::ThreadStartResponse;
use codex_app_server_protocol::TurnStartResponse;
use codex_app_server_protocol::TurnStatus;
use serde_json::json;
use tempfile::TempDir;
@@ -59,7 +60,7 @@ async fn review_start_runs_review_turn_and_emits_code_review_item() -> Result<()
let review_req = mcp
.send_review_start_request(ReviewStartParams {
thread_id: thread_id.clone(),
append_to_original_thread: true,
delivery: Some(ReviewDelivery::Inline),
target: ReviewTarget::Commit {
sha: "1234567deadbeef".to_string(),
title: Some("Tidy UI colors".to_string()),
@@ -71,43 +72,43 @@ async fn review_start_runs_review_turn_and_emits_code_review_item() -> Result<()
mcp.read_stream_until_response_message(RequestId::Integer(review_req)),
)
.await??;
let TurnStartResponse { turn } = to_response::<TurnStartResponse>(review_resp)?;
let ReviewStartResponse {
turn,
review_thread_id,
} = to_response::<ReviewStartResponse>(review_resp)?;
assert_eq!(review_thread_id, thread_id.clone());
let turn_id = turn.id.clone();
assert_eq!(turn.status, TurnStatus::InProgress);
assert_eq!(turn.items.len(), 1);
match &turn.items[0] {
ThreadItem::UserMessage { content, .. } => {
assert_eq!(content.len(), 1);
assert!(matches!(
&content[0],
codex_app_server_protocol::UserInput::Text { .. }
));
}
other => panic!("expected user message, got {other:?}"),
}
let _started: JSONRPCNotification = timeout(
DEFAULT_READ_TIMEOUT,
mcp.read_stream_until_notification_message("turn/started"),
)
.await??;
let item_started: JSONRPCNotification = timeout(
DEFAULT_READ_TIMEOUT,
mcp.read_stream_until_notification_message("item/started"),
)
.await??;
let started: ItemStartedNotification =
serde_json::from_value(item_started.params.expect("params must be present"))?;
match started.item {
ThreadItem::CodeReview { id, review } => {
assert_eq!(id, turn_id);
assert_eq!(review, "commit 1234567");
// Confirm we see the EnteredReviewMode marker on the main thread.
let mut saw_entered_review_mode = false;
for _ in 0..10 {
let item_started: JSONRPCNotification = timeout(
DEFAULT_READ_TIMEOUT,
mcp.read_stream_until_notification_message("item/started"),
)
.await??;
let started: ItemStartedNotification =
serde_json::from_value(item_started.params.expect("params must be present"))?;
match started.item {
ThreadItem::EnteredReviewMode { id, review } => {
assert_eq!(id, turn_id);
assert_eq!(review, "commit 1234567: Tidy UI colors");
saw_entered_review_mode = true;
break;
}
_ => continue,
}
other => panic!("expected code review item, got {other:?}"),
}
assert!(
saw_entered_review_mode,
"did not observe enteredReviewMode item"
);
// Confirm we see the ExitedReviewMode marker (with review text)
// on the same turn. Ignore any other items the stream surfaces.
let mut review_body: Option<String> = None;
for _ in 0..5 {
for _ in 0..10 {
let review_notif: JSONRPCNotification = timeout(
DEFAULT_READ_TIMEOUT,
mcp.read_stream_until_notification_message("item/completed"),
@@ -116,13 +117,12 @@ async fn review_start_runs_review_turn_and_emits_code_review_item() -> Result<()
let completed: ItemCompletedNotification =
serde_json::from_value(review_notif.params.expect("params must be present"))?;
match completed.item {
ThreadItem::CodeReview { id, review } => {
ThreadItem::ExitedReviewMode { id, review } => {
assert_eq!(id, turn_id);
review_body = Some(review);
break;
}
ThreadItem::UserMessage { .. } => continue,
other => panic!("unexpected item/completed payload: {other:?}"),
_ => continue,
}
}
@@ -146,7 +146,7 @@ async fn review_start_rejects_empty_base_branch() -> Result<()> {
let request_id = mcp
.send_review_start_request(ReviewStartParams {
thread_id,
append_to_original_thread: true,
delivery: Some(ReviewDelivery::Inline),
target: ReviewTarget::BaseBranch {
branch: " ".to_string(),
},
@@ -167,6 +167,56 @@ async fn review_start_rejects_empty_base_branch() -> Result<()> {
Ok(())
}
#[tokio::test]
async fn review_start_with_detached_delivery_returns_new_thread_id() -> Result<()> {
let review_payload = json!({
"findings": [],
"overall_correctness": "ok",
"overall_explanation": "detached review",
"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 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_default_thread(&mut mcp).await?;
let review_req = mcp
.send_review_start_request(ReviewStartParams {
thread_id: thread_id.clone(),
delivery: Some(ReviewDelivery::Detached),
target: ReviewTarget::Custom {
instructions: "detached review".to_string(),
},
})
.await?;
let review_resp: JSONRPCResponse = timeout(
DEFAULT_READ_TIMEOUT,
mcp.read_stream_until_response_message(RequestId::Integer(review_req)),
)
.await??;
let ReviewStartResponse {
turn,
review_thread_id,
} = to_response::<ReviewStartResponse>(review_resp)?;
assert_eq!(turn.status, TurnStatus::InProgress);
assert_ne!(
review_thread_id, thread_id,
"detached review should run on a different thread"
);
Ok(())
}
#[tokio::test]
async fn review_start_rejects_empty_commit_sha() -> Result<()> {
let server = create_mock_chat_completions_server_unchecked(vec![]).await;
@@ -180,7 +230,7 @@ async fn review_start_rejects_empty_commit_sha() -> Result<()> {
let request_id = mcp
.send_review_start_request(ReviewStartParams {
thread_id,
append_to_original_thread: true,
delivery: Some(ReviewDelivery::Inline),
target: ReviewTarget::Commit {
sha: "\t".to_string(),
title: None,
@@ -215,7 +265,7 @@ async fn review_start_rejects_empty_custom_instructions() -> Result<()> {
let request_id = mcp
.send_review_start_request(ReviewStartParams {
thread_id,
append_to_original_thread: true,
delivery: Some(ReviewDelivery::Inline),
target: ReviewTarget::Custom {
instructions: "\n\n".to_string(),
},

View File

@@ -1,6 +1,7 @@
use anyhow::Result;
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;
@@ -10,6 +11,7 @@ use app_test_support::to_response;
use codex_app_server_protocol::ApprovalDecision;
use codex_app_server_protocol::CommandExecutionRequestApprovalResponse;
use codex_app_server_protocol::CommandExecutionStatus;
use codex_app_server_protocol::FileChangeOutputDeltaNotification;
use codex_app_server_protocol::FileChangeRequestApprovalResponse;
use codex_app_server_protocol::ItemCompletedNotification;
use codex_app_server_protocol::ItemStartedNotification;
@@ -724,6 +726,26 @@ async fn turn_start_file_change_approval_v2() -> Result<()> {
)
.await?;
let output_delta_notif = timeout(
DEFAULT_READ_TIMEOUT,
mcp.read_stream_until_notification_message("item/fileChange/outputDelta"),
)
.await??;
let output_delta: FileChangeOutputDeltaNotification = serde_json::from_value(
output_delta_notif
.params
.clone()
.expect("item/fileChange/outputDelta params"),
)?;
assert_eq!(output_delta.thread_id, thread.id);
assert_eq!(output_delta.turn_id, turn.id);
assert_eq!(output_delta.item_id, "patch-call");
assert!(
!output_delta.delta.is_empty(),
"expected delta to be non-empty, got: {}",
output_delta.delta
);
let completed_file_change = timeout(DEFAULT_READ_TIMEOUT, async {
loop {
let completed_notif = mcp
@@ -907,6 +929,134 @@ async fn turn_start_file_change_approval_decline_v2() -> Result<()> {
Ok(())
}
#[tokio::test]
#[cfg_attr(windows, ignore = "process id reporting differs on Windows")]
async fn command_execution_notifications_include_process_id() -> Result<()> {
skip_if_no_network!(Ok(()));
let responses = vec![
create_exec_command_sse_response("uexec-1")?,
create_final_assistant_message_sse_response("done")?,
];
let server = create_mock_chat_completions_server(responses).await;
let codex_home = TempDir::new()?;
create_config_toml(codex_home.path(), &server.uri(), "never")?;
let config_toml = codex_home.path().join("config.toml");
let mut config_contents = std::fs::read_to_string(&config_toml)?;
config_contents.push_str(
r#"
[features]
unified_exec = true
"#,
);
std::fs::write(&config_toml, config_contents)?;
let mut mcp = McpProcess::new(codex_home.path()).await?;
timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
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)?;
let turn_id = mcp
.send_turn_start_request(TurnStartParams {
thread_id: thread.id.clone(),
input: vec![V2UserInput::Text {
text: "run a command".to_string(),
}],
..Default::default()
})
.await?;
let turn_resp: JSONRPCResponse = timeout(
DEFAULT_READ_TIMEOUT,
mcp.read_stream_until_response_message(RequestId::Integer(turn_id)),
)
.await??;
let TurnStartResponse { turn: _turn } = to_response::<TurnStartResponse>(turn_resp)?;
let started_command = timeout(DEFAULT_READ_TIMEOUT, async {
loop {
let notif = mcp
.read_stream_until_notification_message("item/started")
.await?;
let started: ItemStartedNotification = serde_json::from_value(
notif
.params
.clone()
.expect("item/started should include params"),
)?;
if let ThreadItem::CommandExecution { .. } = started.item {
return Ok::<ThreadItem, anyhow::Error>(started.item);
}
}
})
.await??;
let ThreadItem::CommandExecution {
id,
process_id: started_process_id,
status,
..
} = started_command
else {
unreachable!("loop ensures we break on command execution items");
};
assert_eq!(id, "uexec-1");
assert_eq!(status, CommandExecutionStatus::InProgress);
let started_process_id = started_process_id.expect("process id should be present");
let completed_command = timeout(DEFAULT_READ_TIMEOUT, async {
loop {
let notif = mcp
.read_stream_until_notification_message("item/completed")
.await?;
let completed: ItemCompletedNotification = serde_json::from_value(
notif
.params
.clone()
.expect("item/completed should include params"),
)?;
if let ThreadItem::CommandExecution { .. } = completed.item {
return Ok::<ThreadItem, anyhow::Error>(completed.item);
}
}
})
.await??;
let ThreadItem::CommandExecution {
id: completed_id,
process_id: completed_process_id,
status: completed_status,
exit_code,
..
} = completed_command
else {
unreachable!("loop ensures we break on command execution items");
};
assert_eq!(completed_id, "uexec-1");
assert_eq!(completed_status, CommandExecutionStatus::Completed);
assert_eq!(exit_code, Some(0));
assert_eq!(
completed_process_id.as_deref(),
Some(started_process_id.as_str())
);
timeout(
DEFAULT_READ_TIMEOUT,
mcp.read_stream_until_notification_message("turn/completed"),
)
.await??;
Ok(())
}
// Helper to create a config.toml pointing at the mock model server.
fn create_config_toml(
codex_home: &Path,

View File

@@ -18,6 +18,8 @@ use codex_cli::login::run_logout;
use codex_cloud_tasks::Cli as CloudTasksCli;
use codex_common::CliConfigOverrides;
use codex_exec::Cli as ExecCli;
use codex_exec::Command as ExecCommand;
use codex_exec::ReviewArgs;
use codex_execpolicy::ExecPolicyCheckCommand;
use codex_responses_api_proxy::Args as ResponsesApiProxyArgs;
use codex_tui::AppExitInfo;
@@ -72,6 +74,9 @@ enum Subcommand {
#[clap(visible_alias = "e")]
Exec(ExecCli),
/// Run a code review non-interactively.
Review(ReviewArgs),
/// Manage login.
Login(LoginCommand),
@@ -105,6 +110,9 @@ enum Subcommand {
/// Resume a previous interactive session (picker by default; use --last to continue the most recent).
Resume(ResumeCommand),
/// Fork an existing session into a new conversation.
Fork(ForkCommand),
/// [EXPERIMENTAL] Browse tasks from Codex Cloud and apply changes locally.
#[clap(name = "cloud", alias = "cloud-tasks")]
Cloud(CloudTasksCli),
@@ -147,6 +155,16 @@ struct ResumeCommand {
config_overrides: TuiCli,
}
#[derive(Debug, Parser)]
struct ForkCommand {
/// Resume from a saved session name or rollout id, but start a new conversation.
#[arg(value_name = "ID|NAME")]
target: String,
#[clap(flatten)]
config_overrides: TuiCli,
}
#[derive(Debug, Parser)]
struct SandboxArgs {
#[command(subcommand)]
@@ -449,6 +467,15 @@ async fn cli_main(codex_linux_sandbox_exe: Option<PathBuf>) -> anyhow::Result<()
);
codex_exec::run_main(exec_cli, codex_linux_sandbox_exe).await?;
}
Some(Subcommand::Review(review_args)) => {
let mut exec_cli = ExecCli::try_parse_from(["codex", "exec"])?;
exec_cli.command = Some(ExecCommand::Review(review_args));
prepend_config_flags(
&mut exec_cli.config_overrides,
root_config_overrides.clone(),
);
codex_exec::run_main(exec_cli, codex_linux_sandbox_exe).await?;
}
Some(Subcommand::McpServer) => {
codex_mcp_server::run_main(codex_linux_sandbox_exe, root_config_overrides).await?;
}
@@ -488,6 +515,19 @@ async fn cli_main(codex_linux_sandbox_exe: Option<PathBuf>) -> anyhow::Result<()
let exit_info = codex_tui::run_main(interactive, codex_linux_sandbox_exe).await?;
handle_app_exit(exit_info)?;
}
Some(Subcommand::Fork(ForkCommand {
target,
config_overrides,
})) => {
interactive = finalize_fork_interactive(
interactive,
root_config_overrides.clone(),
target,
config_overrides,
);
let exit_info = codex_tui::run_main(interactive, codex_linux_sandbox_exe).await?;
handle_app_exit(exit_info)?;
}
Some(Subcommand::Login(mut login_cli)) => {
prepend_config_flags(
&mut login_cli.config_overrides,
@@ -652,6 +692,7 @@ fn finalize_resume_interactive(
interactive.resume_last = last;
interactive.resume_session_id = resume_session_id;
interactive.resume_show_all = show_all;
interactive.fork_source = None;
// Merge resume-scoped flags and overrides with highest precedence.
merge_resume_cli_flags(&mut interactive, resume_cli);
@@ -662,6 +703,21 @@ fn finalize_resume_interactive(
interactive
}
fn finalize_fork_interactive(
mut interactive: TuiCli,
root_config_overrides: CliConfigOverrides,
target: String,
fork_cli: TuiCli,
) -> TuiCli {
interactive.resume_picker = false;
interactive.resume_last = false;
interactive.resume_session_id = None;
interactive.fork_source = Some(target);
merge_resume_cli_flags(&mut interactive, fork_cli);
prepend_config_flags(&mut interactive.config_overrides, root_config_overrides);
interactive
}
/// Merge flags provided to `codex resume` so they take precedence over any
/// root-level flags. Only overrides fields explicitly set on the resume-scoped
/// CLI. Also appends `-c key=value` overrides with highest precedence.
@@ -752,6 +808,26 @@ mod tests {
)
}
fn fork_from_args(args: &[&str]) -> TuiCli {
let cli = MultitoolCli::try_parse_from(args).expect("parse");
let MultitoolCli {
interactive,
config_overrides: root_overrides,
subcommand,
feature_toggles: _,
} = cli;
let Subcommand::Fork(ForkCommand {
target,
config_overrides,
}) = subcommand.expect("fork present")
else {
unreachable!()
};
finalize_fork_interactive(interactive, root_overrides, target, config_overrides)
}
fn sample_exit_info(conversation: Option<&str>) -> AppExitInfo {
let token_usage = TokenUsage {
output_tokens: 2,
@@ -844,6 +920,15 @@ mod tests {
assert!(interactive.resume_show_all);
}
#[test]
fn fork_sets_target_and_disables_resume_controls() {
let interactive = fork_from_args(["codex", "fork", "saved"].as_ref());
assert_eq!(interactive.fork_source.as_deref(), Some("saved"));
assert!(!interactive.resume_picker);
assert!(!interactive.resume_last);
assert!(interactive.resume_session_id.is_none());
}
#[test]
fn resume_merges_option_flags_and_full_auto() {
let interactive = finalize_from_args(

View File

@@ -1,206 +0,0 @@
# Client Extraction Plan
## Goals
- Split the HTTP transport/client code out of `codex-core` into a reusable crate that is agnostic of Codex/OpenAI business logic and API schemas.
- Create a separate API library crate that houses typed requests/responses for well-known APIs (Responses, Chat Completions, Compact) and plugs into the transport crate via minimal traits.
- Preserve current behaviour (auth headers, retries, SSE handling, rate-limit parsing, compaction, fixtures) while making the APIs symmetric and avoiding code duplication.
- Keep existing consumers (`codex-core`, tests, and tools) stable by providing a small compatibility layer during the transition.
## Snapshot of Today
- `core/src/client.rs (ModelClient)` owns config/auth/session state, chooses wire API, builds payloads, drives retries, parses SSE, compaction, and rate-limit headers.
- `core/src/chat_completions.rs` implements the Chat Completions call + SSE parser + aggregation helper.
- `core/src/client_common.rs` holds `Prompt`, tool specs, shared request structs (`ResponsesApiRequest`, `TextControls`), and `ResponseEvent`/`ResponseStream`.
- `core/src/default_client.rs` wraps `reqwest` with Codex UA/originator defaults.
- `core/src/model_provider_info.rs` models providers (base URL, headers, env keys, retry/timeout tuning) and builds `CodexRequestBuilder`s.
- Current retry logic is co-located with API handling; streaming SSE parsing is duplicated across Responses/Chat.
## Target Crates (with interfaces)
- `codex-client` (generic transport)
- Owns the generic HTTP machinery: a `CodexHttpClient`/`CodexRequestBuilder`-style wrapper, retry/backoff hooks, streaming connector (SSE framing + idle timeout), header injection, and optional telemetry callbacks.
- Does **not** know about OpenAI/Codex-specific paths, headers, or error codes; it only exposes HTTP-level concepts (status, headers, bodies, connection errors).
- Minimal surface:
```rust
pub trait HttpTransport {
fn execute(&self, req: Request) -> Result<Response, TransportError>;
fn stream(&self, req: Request) -> Result<ByteStream, TransportError>;
}
pub struct Request {
pub method: Method,
pub url: String,
pub headers: HeaderMap,
pub body: Option<serde_json::Value>,
pub timeout: Option<Duration>,
}
```
- Generic client traits (request/response/chunk are abstract over the transport):
```rust
#[async_trait::async_trait]
pub trait UnaryClient<Req, Resp> {
async fn run(&self, req: Req) -> Result<Resp, TransportError>;
}
#[async_trait::async_trait]
pub trait StreamClient<Req, Chunk> {
async fn run(&self, req: Req) -> Result<ResponseStream<Chunk>, TransportError>;
}
pub struct RetryPolicy {
pub max_attempts: u64,
pub base_delay: Duration,
pub retry_on: RetryOn, // e.g., transport errors + 429/5xx
}
```
- `RetryOn` lives in `codex-client` and captures HTTP status classes and transport failures that qualify for retry.
- Implementations in `codex-api` plug in their own request types, parsers, and retry policies while reusing the transports backoff and error types.
- Planned runtime helper:
```rust
pub async fn run_with_retry<T, F, Fut>(
policy: RetryPolicy,
make_req: impl Fn() -> Request,
op: F,
) -> Result<T, TransportError>
where
F: Fn(Request) -> Fut,
Fut: Future<Output = Result<T, TransportError>>,
{
for attempt in 0..=policy.max_attempts {
let req = make_req();
match op(req).await {
Ok(resp) => return Ok(resp),
Err(err) if policy.retry_on.should_retry(&err, attempt) => {
tokio::time::sleep(backoff(policy.base_delay, attempt + 1)).await;
}
Err(err) => return Err(err),
}
}
Err(TransportError::RetryLimit)
}
```
- Unary clients wrap `transport.execute` with this helper and then deserialize.
- Stream clients wrap the **initial** `transport.stream` call with this helper. Mid-stream disconnects are surfaced as `StreamError`s; automatic resume/reconnect can be added later on top of this primitive if we introduce cursor support.
- Common helpers: `retry::backoff(attempt)`, `errors::{TransportError, StreamError}`.
- Streaming utility (SSE framing only):
```rust
pub fn sse_stream<S>(
bytes: S,
idle_timeout: Duration,
tx: mpsc::Sender<Result<String, StreamError>>,
telemetry: Option<Box<dyn Telemetry>>,
)
where
S: Stream<Item = Result<Bytes, TransportError>> + Unpin + Send + 'static;
```
- `sse_stream` is responsible for timeouts, connection-level errors, and emitting raw `data:` chunks as UTF-8 strings; parsing those strings into structured events is done in `codex-api`.
- `codex-api` (OpenAI/Codex API library)
- Owns typed models for Responses/Chat/Compact plus shared helpers (`Prompt`, tool specs, text controls, `ResponsesApiRequest`, etc.).
- Knows about OpenAI/Codex semantics:
- URL shapes (`/v1/responses`, `/v1/chat/completions`, `/responses/compact`).
- Provider configuration (`WireApi`, base URLs, query params, per-provider retry knobs).
- Rate-limit headers (`x-codex-*`) and their mapping into `RateLimitSnapshot` / `CreditsSnapshot`.
- Error body formats (`{ error: { type, code, message, plan_type, resets_at } }`) and how they become API errors (context window exceeded, quota/usage limit, etc.).
- SSE event names (`response.output_item.done`, `response.completed`, `response.failed`, etc.) and their mapping into high-level events.
- Provides a provider abstraction (conceptually similar to `ModelProviderInfo`):
```rust
pub struct Provider {
pub name: String,
pub base_url: String,
pub wire: WireApi, // Responses | Chat
pub headers: HeaderMap,
pub retry: RetryConfig,
pub stream_idle_timeout: Duration,
}
pub trait AuthProvider {
/// Returns a bearer token to use for this request (if any).
/// Implementations are expected to be cheap and to surface already-refreshed tokens;
/// higher layers (`codex-core`) remain responsible for token refresh flows.
fn bearer_token(&self) -> Option<String>;
/// Optional ChatGPT account id header for Chat mode.
fn account_id(&self) -> Option<String>;
}
```
- Ready-made clients built on `HttpTransport`:
```rust
pub struct ResponsesClient<T: HttpTransport, A: AuthProvider> { /* ... */ }
impl<T, A> ResponsesClient<T, A> {
pub async fn stream(&self, prompt: &Prompt) -> ApiResult<ResponseStream<ApiEvent>>;
pub async fn compact(&self, prompt: &Prompt) -> ApiResult<Vec<ResponseItem>>;
}
pub struct ChatClient<T: HttpTransport, A: AuthProvider> { /* ... */ }
impl<T, A> ChatClient<T, A> {
pub async fn stream(&self, prompt: &Prompt) -> ApiResult<ResponseStream<ApiEvent>>;
}
pub struct CompactClient<T: HttpTransport, A: AuthProvider> { /* ... */ }
impl<T, A> CompactClient<T, A> {
pub async fn compact(&self, prompt: &Prompt) -> ApiResult<Vec<ResponseItem>>;
}
```
- Streaming events unified across wire APIs (this can closely mirror `ResponseEvent` today, and we may type-alias one to the other during migration):
```rust
pub enum ApiEvent {
Created,
OutputItemAdded(ResponseItem),
OutputItemDone(ResponseItem),
OutputTextDelta(String),
ReasoningContentDelta { delta: String, content_index: i64 },
ReasoningSummaryDelta { delta: String, summary_index: i64 },
RateLimits(RateLimitSnapshot),
Completed { response_id: String, token_usage: Option<TokenUsage> },
}
```
- Error layering:
- `codex-client`: defines `TransportError` / `StreamError` (status codes, IO, timeouts).
- `codex-api`: defines `ApiError` that wraps `TransportError` plus API-specific errors parsed from bodies and headers.
- `codex-core`: maps `ApiError` into existing `CodexErr` variants so downstream callers remain unchanged.
- Aggregation strategies (todays `AggregateStreamExt`) live here as adapters (`Aggregated`, `Streaming`) that transform `ResponseStream<ApiEvent>` into the higher-level views used by `codex-core`.
## Implementation Steps
1. **Create crates**: add `codex-client` and `codex-api` (names keep the `codex-` prefix). Stub lib files with feature flags/tests wired into the workspace; wire them into `Cargo.toml`.
2. **Extract API-level SSE + rate limits into `codex-api`**:
- Move the Responses SSE parser (`process_sse`), rate-limit parsing, and related tests from `core/src/client.rs` into `codex-api`, keeping the behavior identical.
- Introduce `ApiEvent` (initially equivalent to `ResponseEvent`) and `ApiError`, and adjust the parser to emit those.
- Provide test-only helpers for fixture streams (replacement for `CODEX_RS_SSE_FIXTURE`) in `codex-api`.
3. **Lift transport layer into `codex-client`**:
- Move `CodexHttpClient`/`CodexRequestBuilder`, UA/originator plumbing, and backoff helpers from `core/src/default_client.rs` into `codex-client` (or a thin wrapper on top of it).
- Introduce `HttpTransport`, `Request`, `RetryPolicy`, `RetryOn`, and `run_with_retry` as described above.
- Keep sandbox/no-proxy toggles behind injected configuration so `codex-client` stays generic and does not depend on Codex-specific env vars.
4. **Model provider abstraction in `codex-api`**:
- Relocate `ModelProviderInfo` (base URL, env/header resolution, retry knobs, wire API enum) into `codex-api`, expressed in terms of `Provider` and `AuthProvider`.
- Ensure provider logic handles:
- URL building for Responses/Chat/Compact (including Azure special cases).
- Static and env-based headers.
- Per-provider retry and idle-timeout settings that map cleanly into `RetryPolicy`/`RetryOn`.
5. **API crate wiring**:
- Move `Prompt`, tool specs, `ResponsesApiRequest`, `TextControls`, and `ResponseEvent/ResponseStream` into `codex-api` under modules (`common`, `responses`, `chat`, `compact`), keeping public types stable or re-exported through `codex-core` as needed.
- Rebuild Responses and Chat clients on top of `HttpTransport` + `StreamClient`, reusing shared retry + SSE helpers; keep aggregation adapters as reusable strategies instead of `ModelClient`-local logic.
- Implement Compact on top of `UnaryClient` and the unary `execute` path with JSON deserialization, sharing the same retry policy.
- Keep request builders symmetric: each client prepares a `Request<serde_json::Value>`, attaches headers/auth via `AuthProvider`, and plugs in its parser (streaming clients) or deserializer (unary) while sharing retry/backoff configuration derived from `Provider`.
6. **Core integration layer**:
- Replace `core::ModelClient` internals with thin adapters that construct `codex-api` clients using `Config`, `AuthManager`, and `OtelEventManager`.
- Keep the public `ModelClient` API and `ResponseEvent`/`ResponseStream` types stable by re-exporting `codex-api` types or providing type aliases.
- Preserve existing auth flows (including ChatGPT token refresh) inside `codex-core` or a thin adapter, using `AuthProvider` to surface bearer tokens to `codex-api` and handling 401/refresh semantics at this layer.
7. **Tests/migration**:
- Move unit tests for SSE parsing, retry/backoff decisions, and provider/header behavior into the new crates; keep integration tests in `core` using the compatibility layer.
- Update fixtures to be consumed via test-only adapters in `codex-api`.
- Run targeted `just fmt`, `just fix -p` for the touched crates, and scoped `cargo test -p codex-client`, `-p codex-api`, and existing `codex-core` suites.
## Design Decisions
- **UA construction**
- `codex-client` exposes an optional UA suffix/provider hook (tiny feature) and remains unaware of the CLI; `codex-core` / the CLI compute the full UA (including `terminal::user_agent()`) and pass the suffix or builder down.
- **Config vs provider**
- Most configuration stays in `codex-core`. `codex-api::Provider` only contains what is strictly required for HTTP (base URLs, query params, retry/timeout knobs, wire API), while higher-level knobs (reasoning defaults, verbosity flags, etc.) remain core concerns.
- **Auth flow ownership**
- Auth flows (including ChatGPT token refresh) remain in `codex-core`. `AuthProvider` simply exposes already-fresh tokens/account IDs; 401 handling and refresh retries stay in the existing auth layer.
- **Error enums**
- `codex-client` continues to define `TransportError` / `StreamError`. `codex-api` defines an `ApiError` (deriving `thiserror::Error`) that wraps `TransportError` and API-specific failures, and `codex-core` maps `ApiError` into existing `CodexErr` variants for callers.
- **Streaming reconnection semantics**
- For now, mid-stream SSE failures are surfaced as errors and only the initial connection is retried via `run_with_retry`. We will revisit mid-stream reconnect/resume once the underlying APIs support cursor/idempotent event semantics.

View File

@@ -1,21 +1,22 @@
[package]
name = "codex-client"
version.workspace = true
edition.workspace = true
license.workspace = true
name = "codex-client"
version.workspace = true
[dependencies]
async-trait = { workspace = true }
bytes = { workspace = true }
eventsource-stream = { workspace = true }
futures = { workspace = true }
http = { workspace = true }
rand = { workspace = true }
reqwest = { workspace = true, features = ["json", "stream"] }
serde = { workspace = true, features = ["derive"] }
serde_json = { workspace = true }
thiserror = { workspace = true }
tokio = { workspace = true, features = ["macros", "rt", "time", "sync"] }
rand = { workspace = true }
eventsource-stream = { workspace = true }
tracing = { workspace = true }
[lints]
workspace = true

View File

@@ -8,6 +8,9 @@ use futures::stream::BoxStream;
use http::HeaderMap;
use http::Method;
use http::StatusCode;
use tracing::Level;
use tracing::enabled;
use tracing::trace;
pub type ByteStream = BoxStream<'static, Result<Bytes, TransportError>>;
@@ -83,6 +86,15 @@ impl HttpTransport for ReqwestTransport {
}
async fn stream(&self, req: Request) -> Result<StreamResponse, TransportError> {
if enabled!(Level::TRACE) {
trace!(
"{} to {}: {}",
req.method,
req.url,
req.body.as_ref().unwrap_or_default()
);
}
let builder = self.build(req)?;
let resp = builder.send().await.map_err(Self::map_error)?;
let status = resp.status();

View File

@@ -52,6 +52,7 @@ regex-lite = { workspace = true }
reqwest = { workspace = true, features = ["json", "stream"] }
serde = { workspace = true, features = ["derive"] }
serde_json = { workspace = true }
serde_yaml = { workspace = true }
sha1 = { workspace = true }
sha2 = { workspace = true }
shlex = { workspace = true }
@@ -87,6 +88,9 @@ uuid = { workspace = true, features = ["serde", "v4", "v5"] }
which = { workspace = true }
wildmatch = { workspace = true }
[features]
deterministic_process_ids = []
[target.'cfg(target_os = "linux")'.dependencies]
landlock = { workspace = true }
@@ -115,6 +119,7 @@ keyring = { workspace = true, features = ["sync-secret-service"] }
assert_cmd = { workspace = true }
assert_matches = { workspace = true }
codex-arg0 = { workspace = true }
codex-core = { path = ".", features = ["deterministic_process_ids"] }
core_test_support = { workspace = true }
ctor = { workspace = true }
escargot = { workspace = true }

View File

@@ -33,12 +33,20 @@ pub(crate) fn map_api_error(err: ApiError) -> CodexErr {
headers,
body,
} => {
if status == http::StatusCode::INTERNAL_SERVER_ERROR {
let body_text = body.unwrap_or_default();
if status == http::StatusCode::BAD_REQUEST {
if body_text
.contains("The image data you provided does not represent a valid image")
{
CodexErr::InvalidImageRequest()
} else {
CodexErr::InvalidRequest(body_text)
}
} else if status == http::StatusCode::INTERNAL_SERVER_ERROR {
CodexErr::InternalServerError
} else if status == http::StatusCode::TOO_MANY_REQUESTS {
if let Some(body) = body
&& let Ok(err) = serde_json::from_str::<UsageErrorResponse>(&body)
{
if let Ok(err) = serde_json::from_str::<UsageErrorResponse>(&body_text) {
if err.error.error_type.as_deref() == Some("usage_limit_reached") {
let rate_limits = headers.as_ref().and_then(parse_rate_limit);
let resets_at = err
@@ -62,7 +70,7 @@ pub(crate) fn map_api_error(err: ApiError) -> CodexErr {
} else {
CodexErr::UnexpectedStatus(UnexpectedResponseError {
status,
body: body.unwrap_or_default(),
body: body_text,
request_id: extract_request_id(headers.as_ref()),
})
}

View File

@@ -102,6 +102,9 @@ use crate::protocol::TurnDiffEvent;
use crate::protocol::WarningEvent;
use crate::rollout::RolloutRecorder;
use crate::rollout::RolloutRecorderParams;
use crate::rollout::map_session_init_error;
use crate::saved_sessions::build_saved_session_entry;
use crate::saved_sessions::upsert_saved_session;
use crate::shell;
use crate::state::ActiveTurn;
use crate::state::SessionServices;
@@ -136,6 +139,7 @@ use codex_protocol::protocol::InitialHistory;
use codex_protocol::user_input::UserInput;
use codex_utils_readiness::Readiness;
use codex_utils_readiness::ReadinessFlag;
use std::path::Path;
/// The high-level interface to the Codex system.
/// It operates as a queue pair where you send submissions and receive events.
@@ -151,6 +155,7 @@ pub struct Codex {
pub struct CodexSpawnOk {
pub codex: Codex,
pub conversation_id: ConversationId,
pub(crate) session: Arc<Session>,
}
pub(crate) const INITIAL_SUBMIT_ID: &str = "";
@@ -206,12 +211,13 @@ impl Codex {
.await
.map_err(|e| {
error!("Failed to create session: {e:#}");
CodexErr::InternalAgentDied
map_session_init_error(&e, &config.codex_home)
})?;
let conversation_id = session.conversation_id;
// This task will run until Op::Shutdown is received.
tokio::spawn(submission_loop(session, config, rx_sub));
let submission_session = Arc::clone(&session);
tokio::spawn(submission_loop(submission_session, config, rx_sub));
let codex = Codex {
next_id: AtomicU64::new(0),
tx_sub,
@@ -221,6 +227,7 @@ impl Codex {
Ok(CodexSpawnOk {
codex,
conversation_id,
session,
})
}
@@ -508,7 +515,7 @@ impl Session {
let rollout_recorder = rollout_recorder.map_err(|e| {
error!("failed to initialize rollout recorder: {e:#}");
anyhow::anyhow!("failed to initialize rollout recorder: {e:#}")
anyhow::Error::from(e)
})?;
let rollout_path = rollout_recorder.rollout_path.clone();
@@ -642,18 +649,68 @@ impl Session {
}
/// Ensure all rollout writes are durably flushed.
pub(crate) async fn flush_rollout(&self) {
pub(crate) async fn flush_rollout(&self) -> std::io::Result<()> {
let recorder = {
let guard = self.services.rollout.lock().await;
guard.clone()
};
if let Some(rec) = recorder
&& let Err(e) = rec.flush().await
{
warn!("failed to flush rollout recorder: {e}");
if let Some(rec) = recorder {
rec.flush().await
} else {
Ok(())
}
}
pub(crate) async fn set_session_name(&self, name: Option<String>) -> std::io::Result<()> {
let recorder = {
let guard = self.services.rollout.lock().await;
guard.clone()
};
if let Some(rec) = recorder {
rec.set_session_name(name).await
} else {
Ok(())
}
}
pub(crate) async fn rollout_path(&self) -> CodexResult<PathBuf> {
let guard = self.services.rollout.lock().await;
let Some(rec) = guard.as_ref() else {
return Err(CodexErr::Fatal(
"Rollout recorder is not initialized; cannot save session.".to_string(),
));
};
Ok(rec.rollout_path.clone())
}
pub(crate) async fn model(&self) -> String {
let state = self.state.lock().await;
state.session_configuration.model.clone()
}
pub(crate) async fn save_session(
&self,
codex_home: &Path,
name: &str,
) -> CodexResult<crate::SavedSessionEntry> {
let trimmed = name.trim();
if trimmed.is_empty() {
return Err(CodexErr::Fatal("Usage: /save <name>".to_string()));
}
let rollout_path = self.rollout_path().await?;
self.flush_rollout()
.await
.map_err(|e| CodexErr::Fatal(format!("failed to flush rollout recorder: {e}")))?;
self.set_session_name(Some(trimmed.to_string()))
.await
.map_err(|e| CodexErr::Fatal(format!("failed to write session name: {e}")))?;
let entry =
build_saved_session_entry(trimmed.to_string(), rollout_path, self.model().await)
.await?;
upsert_saved_session(codex_home, entry.clone()).await?;
Ok(entry)
}
fn next_internal_sub_id(&self) -> String {
let id = self
.next_internal_sub_id
@@ -674,7 +731,9 @@ impl Session {
let items = self.build_initial_context(&turn_context);
self.record_conversation_items(&turn_context, &items).await;
// Ensure initial items are visible to immediate readers (e.g., tests, forks).
self.flush_rollout().await;
if let Err(e) = self.flush_rollout().await {
warn!("failed to flush rollout recorder: {e}");
}
}
InitialHistory::Resumed(_) | InitialHistory::Forked(_) => {
let rollout_items = conversation_history.get_rollout_items();
@@ -718,10 +777,18 @@ impl Session {
// If persisting, persist all rollout items as-is (recorder filters)
if persist && !rollout_items.is_empty() {
self.persist_rollout_items(&rollout_items).await;
// Drop legacy SessionMeta lines from the source rollout so the forked
// session only contains its own SessionMeta written by the recorder.
let filtered =
InitialHistory::Forked(rollout_items.clone()).without_session_meta();
if !filtered.is_empty() {
self.persist_rollout_items(&filtered).await;
}
}
// Flush after seeding history and any persisted rollout copy.
self.flush_rollout().await;
if let Err(e) = self.flush_rollout().await {
warn!("failed to flush rollout recorder: {e}");
}
}
}
}
@@ -975,6 +1042,7 @@ impl Session {
.read()
.await
.resolve_elicitation(server_name, id, response)
.await
}
/// Records input items: always append to conversation history and
@@ -1034,6 +1102,22 @@ impl Session {
state.record_items(items.iter(), turn_context.truncation_policy);
}
pub(crate) async fn record_model_warning(&self, message: impl Into<String>, ctx: &TurnContext) {
if !self.enabled(Feature::ModelWarnings).await {
return;
}
let item = ResponseItem::Message {
id: None,
role: "user".to_string(),
content: vec![ContentItem::InputText {
text: format!("Warning: {}", message.into()),
}],
};
self.record_conversation_items(ctx, &[item]).await;
}
pub(crate) async fn replace_history(&self, items: Vec<ResponseItem>) {
let mut state = self.state.lock().await;
state.replace_history(items);
@@ -1189,22 +1273,17 @@ impl Session {
}
}
/// Record a user input item to conversation history and also persist a
/// corresponding UserMessage EventMsg to rollout.
async fn record_input_and_rollout_usermsg(
pub(crate) async fn record_response_item_and_emit_turn_item(
&self,
turn_context: &TurnContext,
response_input: &ResponseInputItem,
response_item: ResponseItem,
) {
let response_item: ResponseItem = response_input.clone().into();
// Add to conversation history and persist response item to rollout
// Add to conversation history and persist response item to rollout.
self.record_conversation_items(turn_context, std::slice::from_ref(&response_item))
.await;
// Derive user message events and persist only UserMessage to rollout
let turn_item = parse_turn_item(&response_item);
if let Some(item @ TurnItem::UserMessage(_)) = turn_item {
// Derive a turn item and emit lifecycle events if applicable.
if let Some(item) = parse_turn_item(&response_item) {
self.emit_turn_item_started(turn_context, &item).await;
self.emit_turn_item_completed(turn_context, item).await;
}
@@ -1465,6 +1544,9 @@ async fn submission_loop(sess: Arc<Session>, config: Arc<Config>, rx_sub: Receiv
Op::Review { review_request } => {
handlers::review(&sess, &config, sub.id.clone(), review_request).await;
}
Op::SaveSession { name } => {
handlers::save_session(&sess, &config, sub.id.clone(), name).await;
}
_ => {} // Ignore unknown ops; enum is non_exhaustive to allow extensions.
}
}
@@ -1480,6 +1562,7 @@ mod handlers {
use crate::codex::spawn_review_thread;
use crate::config::Config;
use crate::mcp::auth::compute_auth_statuses;
use crate::review_prompts::resolve_review_request;
use crate::tasks::CompactTask;
use crate::tasks::RegularTask;
use crate::tasks::UndoTask;
@@ -1493,6 +1576,7 @@ mod handlers {
use codex_protocol::protocol::Op;
use codex_protocol::protocol::ReviewDecision;
use codex_protocol::protocol::ReviewRequest;
use codex_protocol::protocol::SaveSessionResponseEvent;
use codex_protocol::protocol::TurnAbortReason;
use codex_protocol::user_input::UserInput;
@@ -1743,6 +1827,38 @@ mod handlers {
.await;
}
pub async fn save_session(
sess: &Arc<Session>,
config: &Arc<crate::config::Config>,
sub_id: String,
name: String,
) {
match sess.save_session(&config.codex_home, &name).await {
Ok(entry) => {
let event = Event {
id: sub_id,
msg: EventMsg::SaveSessionResponse(SaveSessionResponseEvent {
name: entry.name,
rollout_path: entry.rollout_path,
conversation_id: entry.conversation_id,
}),
};
sess.send_event_raw(event).await;
}
Err(err) => {
let message = format!("Failed to save session '{name}': {err}");
let event = Event {
id: sub_id,
msg: EventMsg::Error(ErrorEvent {
message,
codex_error_info: None,
}),
};
sess.send_event_raw(event).await;
}
}
}
pub async fn shutdown(sess: &Arc<Session>, sub_id: String) -> bool {
sess.abort_all_tasks(TurnAbortReason::Interrupted).await;
sess.services
@@ -1788,14 +1904,28 @@ mod handlers {
let turn_context = sess
.new_turn_with_sub_id(sub_id.clone(), SessionSettingsUpdate::default())
.await;
spawn_review_thread(
Arc::clone(sess),
Arc::clone(config),
turn_context.clone(),
sub_id,
review_request,
)
.await;
match resolve_review_request(review_request, config.cwd.as_path()) {
Ok(resolved) => {
spawn_review_thread(
Arc::clone(sess),
Arc::clone(config),
turn_context.clone(),
sub_id,
resolved,
)
.await;
}
Err(err) => {
let event = Event {
id: sub_id,
msg: EventMsg::Error(ErrorEvent {
message: err.to_string(),
codex_error_info: Some(CodexErrorInfo::Other),
}),
};
sess.send_event(&turn_context, event.msg).await;
}
}
}
}
@@ -1805,7 +1935,7 @@ async fn spawn_review_thread(
config: Arc<Config>,
parent_turn_context: Arc<TurnContext>,
sub_id: String,
review_request: ReviewRequest,
resolved: crate::review_prompts::ResolvedReviewRequest,
) {
let model = config.review_model.clone();
let review_model_family = find_family_for_model(&model)
@@ -1821,7 +1951,7 @@ async fn spawn_review_thread(
});
let base_instructions = REVIEW_PROMPT.to_string();
let review_prompt = review_request.prompt.clone();
let review_prompt = resolved.prompt.clone();
let provider = parent_turn_context.client.get_provider();
let auth_manager = parent_turn_context.client.get_auth_manager();
let model_family = review_model_family.clone();
@@ -1880,14 +2010,13 @@ async fn spawn_review_thread(
text: review_prompt,
}];
let tc = Arc::new(review_turn_context);
sess.spawn_task(
tc.clone(),
input,
ReviewTask::new(review_request.append_to_original_thread),
)
.await;
sess.spawn_task(tc.clone(), input, ReviewTask::new()).await;
// Announce entering review mode so UIs can switch modes.
let review_request = ReviewRequest {
target: resolved.target,
user_facing_hint: Some(resolved.user_facing_hint),
};
sess.send_event(&tc, EventMsg::EnteredReviewMode(review_request))
.await;
}
@@ -1921,7 +2050,8 @@ pub(crate) async fn run_task(
sess.send_event(&turn_context, event).await;
let initial_input_for_turn: ResponseInputItem = ResponseInputItem::from(input);
sess.record_input_and_rollout_usermsg(turn_context.as_ref(), &initial_input_for_turn)
let response_item: ResponseItem = initial_input_for_turn.clone().into();
sess.record_response_item_and_emit_turn_item(turn_context.as_ref(), response_item)
.await;
sess.maybe_start_ghost_snapshot(Arc::clone(&turn_context), cancellation_token.child_token())
@@ -2011,6 +2141,13 @@ pub(crate) async fn run_task(
// Aborted turn is reported via a different event.
break;
}
Err(CodexErr::InvalidImageRequest()) => {
let mut state = sess.state.lock().await;
error_or_panic(
"Invalid image detected, replacing it in the last turn to prevent poisoning",
);
state.history.replace_last_turn_images("Invalid image");
}
Err(e) => {
info!("Turn error: {e:#}");
let event = EventMsg::Error(e.to_error_event(None));
@@ -2118,6 +2255,8 @@ async fn run_turn(
}
Err(CodexErr::UsageNotIncluded) => return Err(CodexErr::UsageNotIncluded),
Err(e @ CodexErr::QuotaExceeded) => return Err(e),
Err(e @ CodexErr::InvalidImageRequest()) => return Err(e),
Err(e @ CodexErr::InvalidRequest(_)) => return Err(e),
Err(e @ CodexErr::RefreshTokenFailed(_)) => return Err(e),
Err(e) => {
// Use the configured provider-specific stream retry budget.
@@ -2453,7 +2592,10 @@ mod tests {
use crate::tools::format_exec_output_str;
use crate::protocol::CompactedItem;
use crate::protocol::CreditsSnapshot;
use crate::protocol::InitialHistory;
use crate::protocol::RateLimitSnapshot;
use crate::protocol::RateLimitWindow;
use crate::protocol::ResumedHistory;
use crate::state::TaskKind;
use crate::tasks::SessionTask;
@@ -2523,6 +2665,75 @@ mod tests {
assert_eq!(expected, actual);
}
#[test]
fn set_rate_limits_retains_previous_credits() {
let codex_home = tempfile::tempdir().expect("create temp dir");
let config = Config::load_from_base_config_with_overrides(
ConfigToml::default(),
ConfigOverrides::default(),
codex_home.path().to_path_buf(),
)
.expect("load default test config");
let config = Arc::new(config);
let session_configuration = SessionConfiguration {
provider: config.model_provider.clone(),
model: config.model.clone(),
model_reasoning_effort: config.model_reasoning_effort,
model_reasoning_summary: config.model_reasoning_summary,
developer_instructions: config.developer_instructions.clone(),
user_instructions: config.user_instructions.clone(),
base_instructions: config.base_instructions.clone(),
compact_prompt: config.compact_prompt.clone(),
approval_policy: config.approval_policy,
sandbox_policy: config.sandbox_policy.clone(),
cwd: config.cwd.clone(),
original_config_do_not_use: Arc::clone(&config),
features: Features::default(),
exec_policy: Arc::new(ExecPolicy::empty()),
session_source: SessionSource::Exec,
};
let mut state = SessionState::new(session_configuration);
let initial = RateLimitSnapshot {
primary: Some(RateLimitWindow {
used_percent: 10.0,
window_minutes: Some(15),
resets_at: Some(1_700),
}),
secondary: None,
credits: Some(CreditsSnapshot {
has_credits: true,
unlimited: false,
balance: Some("10.00".to_string()),
}),
};
state.set_rate_limits(initial.clone());
let update = RateLimitSnapshot {
primary: Some(RateLimitWindow {
used_percent: 40.0,
window_minutes: Some(30),
resets_at: Some(1_800),
}),
secondary: Some(RateLimitWindow {
used_percent: 5.0,
window_minutes: Some(60),
resets_at: Some(1_900),
}),
credits: None,
};
state.set_rate_limits(update.clone());
assert_eq!(
state.latest_rate_limits,
Some(RateLimitSnapshot {
primary: update.primary.clone(),
secondary: update.secondary,
credits: initial.credits,
})
);
}
#[test]
fn prefers_structured_content_when_present() {
let ctr = CallToolResult {
@@ -2795,6 +3006,40 @@ mod tests {
(session, turn_context, rx_event)
}
#[tokio::test]
async fn record_model_warning_appends_user_message() {
let (session, turn_context) = make_session_and_context();
session
.state
.lock()
.await
.session_configuration
.features
.enable(Feature::ModelWarnings);
session
.record_model_warning("too many unified exec sessions", &turn_context)
.await;
let mut history = session.clone_history().await;
let history_items = history.get_history();
let last = history_items.last().expect("warning recorded");
match last {
ResponseItem::Message { role, content, .. } => {
assert_eq!(role, "user");
assert_eq!(
content,
&vec![ContentItem::InputText {
text: "Warning: too many unified exec sessions".to_string(),
}]
);
}
other => panic!("expected user message, got {other:?}"),
}
}
#[derive(Clone, Copy)]
struct NeverEndingTask {
kind: TaskKind,
@@ -2886,7 +3131,7 @@ mod tests {
let input = vec![UserInput::Text {
text: "start review".to_string(),
}];
sess.spawn_task(Arc::clone(&tc), input, ReviewTask::new(true))
sess.spawn_task(Arc::clone(&tc), input, ReviewTask::new())
.await;
sess.abort_all_tasks(TurnAbortReason::Interrupted).await;
@@ -2914,6 +3159,8 @@ mod tests {
.expect("event");
match evt.msg {
EventMsg::RawResponseItem(_) => continue,
EventMsg::ItemStarted(_) | EventMsg::ItemCompleted(_) => continue,
EventMsg::AgentMessage(_) => continue,
EventMsg::TurnAborted(e) => {
assert_eq!(TurnAbortReason::Interrupted, e.reason);
break;
@@ -2923,23 +3170,7 @@ mod tests {
}
let history = sess.clone_history().await.get_history();
let found = history.iter().any(|item| match item {
ResponseItem::Message { role, content, .. } if role == "user" => {
content.iter().any(|ci| match ci {
ContentItem::InputText { text } => {
text.contains("<user_action>")
&& text.contains("review")
&& text.contains("interrupted")
}
_ => false,
})
}
_ => false,
});
assert!(
found,
"synthetic review interruption not recorded in history"
);
let _ = history;
}
#[tokio::test]

View File

@@ -1,22 +1,26 @@
use crate::codex::Codex;
use crate::codex::Session;
use crate::error::Result as CodexResult;
use crate::protocol::Event;
use crate::protocol::Op;
use crate::protocol::Submission;
use std::path::PathBuf;
use std::sync::Arc;
pub struct CodexConversation {
codex: Codex,
rollout_path: PathBuf,
session: Arc<Session>,
}
/// Conduit for the bidirectional stream of messages that compose a conversation
/// in Codex.
impl CodexConversation {
pub(crate) fn new(codex: Codex, rollout_path: PathBuf) -> Self {
pub(crate) fn new(codex: Codex, rollout_path: PathBuf, session: Arc<Session>) -> Self {
Self {
codex,
rollout_path,
session,
}
}
@@ -36,4 +40,24 @@ impl CodexConversation {
pub fn rollout_path(&self) -> PathBuf {
self.rollout_path.clone()
}
pub async fn flush_rollout(&self) -> CodexResult<()> {
Ok(self.session.flush_rollout().await?)
}
pub async fn set_session_name(&self, name: Option<String>) -> CodexResult<()> {
Ok(self.session.set_session_name(name).await?)
}
pub async fn model(&self) -> String {
self.session.model().await
}
pub async fn save_session(
&self,
codex_home: &std::path::Path,
name: &str,
) -> CodexResult<crate::SavedSessionEntry> {
self.session.save_session(codex_home, name).await
}
}

View File

@@ -34,6 +34,7 @@ use crate::project_doc::DEFAULT_PROJECT_DOC_FILENAME;
use crate::project_doc::LOCAL_PROJECT_DOC_FILENAME;
use crate::protocol::AskForApproval;
use crate::protocol::SandboxPolicy;
use crate::util::resolve_path;
use codex_app_server_protocol::Tools;
use codex_app_server_protocol::UserSavedConfig;
use codex_protocol::config_types::ForcedLoginMethod;
@@ -160,6 +161,9 @@ pub struct Config {
/// Enable ASCII animations and shimmer effects in the TUI.
pub animations: bool,
/// Show startup tooltips in the TUI welcome screen.
pub show_tooltips: bool,
/// The directory that should be treated as the current working directory
/// for the session. All relative paths inside the business-logic layer are
/// resolved against this path.
@@ -1016,15 +1020,8 @@ impl Config {
let additional_writable_roots: Vec<PathBuf> = additional_writable_roots
.into_iter()
.map(|path| {
let absolute = if path.is_absolute() {
path
} else {
resolved_cwd.join(path)
};
match canonicalize(&absolute) {
Ok(canonical) => canonical,
Err(_) => absolute,
}
let absolute = resolve_path(&resolved_cwd, &path);
canonicalize(&absolute).unwrap_or(absolute)
})
.collect();
let active_project = cfg
@@ -1258,6 +1255,7 @@ impl Config {
.map(|t| t.notifications.clone())
.unwrap_or_default(),
animations: cfg.tui.as_ref().map(|t| t.animations).unwrap_or(true),
show_tooltips: cfg.tui.as_ref().map(|t| t.show_tooltips).unwrap_or(true),
otel: {
let t: OtelConfigToml = cfg.otel.unwrap_or_default();
let log_user_prompt = t.log_user_prompt.unwrap_or(false);
@@ -1299,11 +1297,7 @@ impl Config {
return Ok(None);
};
let full_path = if p.is_relative() {
cwd.join(p)
} else {
p.to_path_buf()
};
let full_path = resolve_path(cwd, p);
let contents = std::fs::read_to_string(&full_path).map_err(|e| {
std::io::Error::new(
@@ -1436,6 +1430,7 @@ persistence = "none"
let tui = parsed.tui.expect("config should include tui section");
assert_eq!(tui.notifications, Notifications::Enabled(true));
assert!(tui.show_tooltips);
}
#[test]
@@ -3009,6 +3004,7 @@ model_verbosity = "high"
disable_paste_burst: false,
tui_notifications: Default::default(),
animations: true,
show_tooltips: true,
otel: OtelConfig::default(),
},
o3_profile_config
@@ -3082,6 +3078,7 @@ model_verbosity = "high"
disable_paste_burst: false,
tui_notifications: Default::default(),
animations: true,
show_tooltips: true,
otel: OtelConfig::default(),
};
@@ -3170,6 +3167,7 @@ model_verbosity = "high"
disable_paste_burst: false,
tui_notifications: Default::default(),
animations: true,
show_tooltips: true,
otel: OtelConfig::default(),
};
@@ -3244,6 +3242,7 @@ model_verbosity = "high"
disable_paste_burst: false,
tui_notifications: Default::default(),
animations: true,
show_tooltips: true,
otel: OtelConfig::default(),
};

View File

@@ -256,8 +256,8 @@ pub struct History {
/// If true, history entries will not be written to disk.
pub persistence: HistoryPersistence,
/// If set, the maximum size of the history file in bytes.
/// TODO(mbolin): Not currently honored.
/// If set, the maximum size of the history file in bytes. The oldest entries
/// are dropped once the file exceeds this limit.
pub max_bytes: Option<usize>,
}
@@ -368,6 +368,11 @@ pub struct Tui {
/// Defaults to `true`.
#[serde(default = "default_true")]
pub animations: bool,
/// Show startup tooltips in the TUI welcome screen.
/// Defaults to `true`.
#[serde(default = "default_true")]
pub show_tooltips: bool,
}
const fn default_true() -> bool {

View File

@@ -5,6 +5,8 @@ use crate::truncate::approx_token_count;
use crate::truncate::approx_tokens_from_byte_count;
use crate::truncate::truncate_function_output_items_with_policy;
use crate::truncate::truncate_text;
use codex_protocol::models::ContentItem;
use codex_protocol::models::FunctionCallOutputContentItem;
use codex_protocol::models::FunctionCallOutputPayload;
use codex_protocol::models::ResponseItem;
use codex_protocol::protocol::TokenUsage;
@@ -118,6 +120,37 @@ impl ContextManager {
self.items = items;
}
pub(crate) fn replace_last_turn_images(&mut self, placeholder: &str) {
let Some(last_item) = self.items.last_mut() else {
return;
};
match last_item {
ResponseItem::Message { role, content, .. } if role == "user" => {
for item in content.iter_mut() {
if matches!(item, ContentItem::InputImage { .. }) {
*item = ContentItem::InputText {
text: placeholder.to_string(),
};
}
}
}
ResponseItem::FunctionCallOutput { output, .. } => {
let Some(content_items) = output.content_items.as_mut() else {
return;
};
for item in content_items.iter_mut() {
if matches!(item, FunctionCallOutputContentItem::InputImage { .. }) {
*item = FunctionCallOutputContentItem::InputText {
text: placeholder.to_string(),
};
}
}
}
_ => {}
}
}
pub(crate) fn update_token_info(
&mut self,
usage: &TokenUsage,

View File

@@ -3,6 +3,7 @@ use crate::CodexAuth;
use crate::codex::Codex;
use crate::codex::CodexSpawnOk;
use crate::codex::INITIAL_SUBMIT_ID;
use crate::codex::Session;
use crate::codex_conversation::CodexConversation;
use crate::config::Config;
use crate::error::CodexErr;
@@ -11,6 +12,7 @@ use crate::protocol::Event;
use crate::protocol::EventMsg;
use crate::protocol::SessionConfiguredEvent;
use crate::rollout::RolloutRecorder;
use crate::saved_sessions::resolve_rollout_path;
use codex_protocol::ConversationId;
use codex_protocol::items::TurnItem;
use codex_protocol::models::ResponseItem;
@@ -18,6 +20,7 @@ use codex_protocol::protocol::InitialHistory;
use codex_protocol::protocol::RolloutItem;
use codex_protocol::protocol::SessionSource;
use std::collections::HashMap;
use std::path::Path;
use std::path::PathBuf;
use std::sync::Arc;
use tokio::sync::RwLock;
@@ -60,6 +63,7 @@ impl ConversationManager {
self.session_source.clone()
}
/// Start a brand new conversation with default initial history.
pub async fn new_conversation(&self, config: Config) -> CodexResult<NewConversation> {
self.spawn_conversation(config, self.auth_manager.clone())
.await
@@ -73,6 +77,7 @@ impl ConversationManager {
let CodexSpawnOk {
codex,
conversation_id,
session,
} = Codex::spawn(
config,
auth_manager,
@@ -80,13 +85,14 @@ impl ConversationManager {
self.session_source.clone(),
)
.await?;
self.finalize_spawn(codex, conversation_id).await
self.finalize_spawn(codex, conversation_id, session).await
}
async fn finalize_spawn(
&self,
codex: Codex,
conversation_id: ConversationId,
session: Arc<Session>,
) -> CodexResult<NewConversation> {
// The first event must be `SessionInitialized`. Validate and forward it
// to the caller so that they can display it in the conversation
@@ -105,6 +111,7 @@ impl ConversationManager {
let conversation = Arc::new(CodexConversation::new(
codex,
session_configured.rollout_path.clone(),
session,
));
self.conversations
.write()
@@ -129,6 +136,7 @@ impl ConversationManager {
.ok_or_else(|| CodexErr::ConversationNotFound(conversation_id))
}
/// Resume a conversation from an on-disk rollout file.
pub async fn resume_conversation_from_rollout(
&self,
config: Config,
@@ -140,6 +148,23 @@ impl ConversationManager {
.await
}
/// Resume a conversation by saved-session name or rollout id string.
pub async fn resume_conversation_from_identifier(
&self,
config: Config,
identifier: &str,
auth_manager: Arc<AuthManager>,
) -> CodexResult<NewConversation> {
let Some(path) = resolve_rollout_path(&config.codex_home, identifier).await? else {
return Err(CodexErr::Fatal(format!(
"No saved session or rollout found for '{identifier}'"
)));
};
self.resume_conversation_from_rollout(config, path, auth_manager)
.await
}
/// Resume a conversation from provided rollout history items.
pub async fn resume_conversation_with_history(
&self,
config: Config,
@@ -149,6 +174,7 @@ impl ConversationManager {
let CodexSpawnOk {
codex,
conversation_id,
session,
} = Codex::spawn(
config,
auth_manager,
@@ -156,7 +182,54 @@ impl ConversationManager {
self.session_source.clone(),
)
.await?;
self.finalize_spawn(codex, conversation_id).await
self.finalize_spawn(codex, conversation_id, session).await
}
/// Fork a new conversation from the given rollout path.
pub async fn fork_from_rollout(
&self,
config: Config,
path: PathBuf,
auth_manager: Arc<AuthManager>,
) -> CodexResult<NewConversation> {
let initial_history = RolloutRecorder::get_rollout_history(&path).await?;
let forked = match initial_history {
InitialHistory::Resumed(resumed) => InitialHistory::Forked(resumed.history),
InitialHistory::Forked(items) => InitialHistory::Forked(items),
InitialHistory::New => InitialHistory::New,
};
self.resume_conversation_with_history(config, forked, auth_manager)
.await
}
/// Fork a new conversation from a saved-session name or rollout id string.
pub async fn fork_from_identifier(
&self,
config: Config,
identifier: &str,
auth_manager: Arc<AuthManager>,
) -> CodexResult<NewConversation> {
let Some(path) = resolve_rollout_path(&config.codex_home, identifier).await? else {
return Err(CodexErr::Fatal(format!(
"No saved session or rollout found for '{identifier}'"
)));
};
self.fork_from_rollout(config, path, auth_manager).await
}
/// Persist a human-friendly session name and record it in saved_sessions.json.
pub async fn save_session(
&self,
conversation_id: ConversationId,
codex_home: &Path,
name: &str,
) -> CodexResult<crate::SavedSessionEntry> {
let trimmed = name.trim();
if trimmed.is_empty() {
return Err(CodexErr::Fatal("Usage: /save <name>".to_string()));
}
let conversation = self.get_conversation(conversation_id).await?;
conversation.save_session(codex_home, trimmed).await
}
/// Removes the conversation from the manager's internal map, though the
@@ -189,9 +262,10 @@ impl ConversationManager {
let CodexSpawnOk {
codex,
conversation_id,
session,
} = Codex::spawn(config, auth_manager, history, self.session_source.clone()).await?;
self.finalize_spawn(codex, conversation_id).await
self.finalize_spawn(codex, conversation_id, session).await
}
}

View File

@@ -103,6 +103,14 @@ pub enum CodexErr {
#[error("{0}")]
UnexpectedStatus(UnexpectedResponseError),
/// Invalid request.
#[error("{0}")]
InvalidRequest(String),
/// Invalid image.
#[error("Image poisoning")]
InvalidImageRequest(),
#[error("{0}")]
UsageLimitReached(UsageLimitReachedError),

View File

@@ -115,7 +115,7 @@ fn evaluate_with_policy(
}
}
pub(crate) fn create_approval_requirement_for_command(
pub(crate) async fn create_approval_requirement_for_command(
policy: &Policy,
command: &[String],
approval_policy: AskForApproval,
@@ -296,8 +296,8 @@ prefix_rule(pattern=["rm"], decision="forbidden")
);
}
#[test]
fn approval_requirement_prefers_execpolicy_match() {
#[tokio::test]
async fn approval_requirement_prefers_execpolicy_match() {
let policy_src = r#"prefix_rule(pattern=["rm"], decision="prompt")"#;
let mut parser = PolicyParser::new();
parser
@@ -312,7 +312,8 @@ prefix_rule(pattern=["rm"], decision="forbidden")
AskForApproval::OnRequest,
&SandboxPolicy::DangerFullAccess,
SandboxPermissions::UseDefault,
);
)
.await;
assert_eq!(
requirement,
@@ -322,8 +323,8 @@ prefix_rule(pattern=["rm"], decision="forbidden")
);
}
#[test]
fn approval_requirement_respects_approval_policy() {
#[tokio::test]
async fn approval_requirement_respects_approval_policy() {
let policy_src = r#"prefix_rule(pattern=["rm"], decision="prompt")"#;
let mut parser = PolicyParser::new();
parser
@@ -338,7 +339,8 @@ prefix_rule(pattern=["rm"], decision="forbidden")
AskForApproval::Never,
&SandboxPolicy::DangerFullAccess,
SandboxPermissions::UseDefault,
);
)
.await;
assert_eq!(
requirement,
@@ -348,8 +350,8 @@ prefix_rule(pattern=["rm"], decision="forbidden")
);
}
#[test]
fn approval_requirement_falls_back_to_heuristics() {
#[tokio::test]
async fn approval_requirement_falls_back_to_heuristics() {
let command = vec!["python".to_string()];
let empty_policy = Policy::empty();
@@ -359,7 +361,8 @@ prefix_rule(pattern=["rm"], decision="forbidden")
AskForApproval::UnlessTrusted,
&SandboxPolicy::ReadOnly,
SandboxPermissions::UseDefault,
);
)
.await;
assert_eq!(
requirement,

View File

@@ -51,6 +51,10 @@ pub enum Feature {
ShellTool,
/// Allow model to call multiple tools in parallel (only for models supporting it).
ParallelToolCalls,
/// Experimental skills injection (CLI flag-driven).
Skills,
/// Send warnings to the model to correct it on the tool usage.
ModelWarnings,
}
impl Feature {
@@ -265,6 +269,12 @@ pub const FEATURES: &[FeatureSpec] = &[
stage: Stage::Stable,
default_enabled: true,
},
FeatureSpec {
id: Feature::ShellTool,
key: "shell_tool",
stage: Stage::Stable,
default_enabled: true,
},
// Unstable features.
FeatureSpec {
id: Feature::UnifiedExec,
@@ -321,9 +331,15 @@ pub const FEATURES: &[FeatureSpec] = &[
default_enabled: false,
},
FeatureSpec {
id: Feature::ShellTool,
key: "shell_tool",
stage: Stage::Stable,
default_enabled: true,
id: Feature::ModelWarnings,
key: "warnings",
stage: Stage::Experimental,
default_enabled: false,
},
FeatureSpec {
id: Feature::Skills,
key: "skills",
stage: Stage::Experimental,
default_enabled: false,
},
];

View File

@@ -2,6 +2,7 @@ use std::collections::HashSet;
use std::path::Path;
use std::path::PathBuf;
use crate::util::resolve_path;
use codex_app_server_protocol::GitSha;
use codex_protocol::protocol::GitInfo;
use futures::future::join_all;
@@ -131,11 +132,15 @@ pub async fn recent_commits(cwd: &Path, limit: usize) -> Vec<CommitLogEntry> {
}
let fmt = "%H%x1f%ct%x1f%s"; // <sha> <US> <commit_time> <US> <subject>
let n = limit.max(1).to_string();
let Some(log_out) =
run_git_command_with_timeout(&["log", "-n", &n, &format!("--pretty=format:{fmt}")], cwd)
.await
else {
let limit_arg = (limit > 0).then(|| limit.to_string());
let mut args: Vec<String> = vec!["log".to_string()];
if let Some(n) = &limit_arg {
args.push("-n".to_string());
args.push(n.clone());
}
args.push(format!("--pretty=format:{fmt}"));
let arg_refs: Vec<&str> = args.iter().map(String::as_str).collect();
let Some(log_out) = run_git_command_with_timeout(&arg_refs, cwd).await else {
return Vec::new();
};
if !log_out.status.success() {
@@ -544,11 +549,7 @@ pub fn resolve_root_git_project_for_trust(cwd: &Path) -> Option<PathBuf> {
.trim()
.to_string();
let git_dir_path_raw = if Path::new(&git_dir_s).is_absolute() {
PathBuf::from(&git_dir_s)
} else {
base.join(&git_dir_s)
};
let git_dir_path_raw = resolve_path(base, &PathBuf::from(&git_dir_s));
// Normalize to handle macOS /var vs /private/var and resolve ".." segments.
let git_dir_path = std::fs::canonicalize(&git_dir_path_raw).unwrap_or(git_dir_path_raw);

View File

@@ -58,6 +58,7 @@ pub use model_provider_info::create_oss_provider_with_base_url;
mod conversation_manager;
mod event_mapping;
pub mod review_format;
pub mod review_prompts;
pub use codex_protocol::protocol::InitialHistory;
pub use conversation_manager::ConversationManager;
pub use conversation_manager::NewConversation;
@@ -70,8 +71,10 @@ mod openai_model_info;
pub mod project_doc;
mod rollout;
pub(crate) mod safety;
pub mod saved_sessions;
pub mod seatbelt;
pub mod shell;
pub mod skills;
pub mod spawn;
pub mod terminal;
mod tools;
@@ -87,6 +90,12 @@ pub use rollout::list::ConversationsPage;
pub use rollout::list::Cursor;
pub use rollout::list::parse_cursor;
pub use rollout::list::read_head_for_summary;
pub use saved_sessions::SavedSessionEntry;
pub use saved_sessions::build_saved_session_entry;
pub use saved_sessions::list_saved_sessions;
pub use saved_sessions::resolve_rollout_path;
pub use saved_sessions::resolve_saved_session;
pub use saved_sessions::upsert_saved_session;
mod function_tool;
mod state;
mod tasks;

View File

@@ -12,7 +12,6 @@ use std::env;
use std::ffi::OsString;
use std::path::PathBuf;
use std::sync::Arc;
use std::sync::Mutex;
use std::time::Duration;
use crate::mcp::auth::McpAuthStatusEntry;
@@ -55,6 +54,7 @@ use serde::Serialize;
use serde_json::json;
use sha1::Digest;
use sha1::Sha1;
use tokio::sync::Mutex;
use tokio::sync::oneshot;
use tokio::task::JoinSet;
use tokio_util::sync::CancellationToken;
@@ -128,7 +128,7 @@ struct ElicitationRequestManager {
}
impl ElicitationRequestManager {
fn resolve(
async fn resolve(
&self,
server_name: String,
id: RequestId,
@@ -136,7 +136,7 @@ impl ElicitationRequestManager {
) -> Result<()> {
self.requests
.lock()
.map_err(|e| anyhow!("failed to lock elicitation requests: {e:?}"))?
.await
.remove(&(server_name, id))
.ok_or_else(|| anyhow!("elicitation request not found"))?
.send(response)
@@ -151,7 +151,8 @@ impl ElicitationRequestManager {
let server_name = server_name.clone();
async move {
let (tx, rx) = oneshot::channel();
if let Ok(mut lock) = elicitation_requests.lock() {
{
let mut lock = elicitation_requests.lock().await;
lock.insert((server_name.clone(), id.clone()), tx);
}
let _ = tx_event
@@ -365,13 +366,15 @@ impl McpConnectionManager {
.context("failed to get client")
}
pub fn resolve_elicitation(
pub async fn resolve_elicitation(
&self,
server_name: String,
id: RequestId,
response: ElicitationResponse,
) -> Result<()> {
self.elicitation_requests.resolve(server_name, id, response)
self.elicitation_requests
.resolve(server_name, id, response)
.await
}
/// Returns a single map that contains all tools. Each key is the

View File

@@ -16,8 +16,14 @@
use std::fs::File;
use std::fs::OpenOptions;
use std::io::BufRead;
use std::io::BufReader;
use std::io::Read;
use std::io::Result;
use std::io::Seek;
use std::io::SeekFrom;
use std::io::Write;
use std::path::Path;
use std::path::PathBuf;
use serde::Deserialize;
@@ -39,10 +45,13 @@ use std::os::unix::fs::PermissionsExt;
/// Filename that stores the message history inside `~/.codex`.
const HISTORY_FILENAME: &str = "history.jsonl";
/// When history exceeds the hard cap, trim it down to this fraction of `max_bytes`.
const HISTORY_SOFT_CAP_RATIO: f64 = 0.8;
const MAX_RETRIES: usize = 10;
const RETRY_SLEEP: Duration = Duration::from_millis(100);
#[derive(Serialize, Deserialize, Debug, Clone)]
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq)]
pub struct HistoryEntry {
pub session_id: String,
pub ts: u64,
@@ -97,11 +106,12 @@ pub(crate) async fn append_entry(
.map_err(|e| std::io::Error::other(format!("failed to serialise history entry: {e}")))?;
line.push('\n');
// Open in append-only mode.
// Open the history file for read/write access (append-only on Unix).
let mut options = OpenOptions::new();
options.append(true).read(true).create(true);
options.read(true).write(true).create(true);
#[cfg(unix)]
{
options.append(true);
options.mode(0o600);
}
@@ -110,6 +120,8 @@ pub(crate) async fn append_entry(
// Ensure permissions.
ensure_owner_only_permissions(&history_file).await?;
let history_max_bytes = config.history.max_bytes;
// Perform a blocking write under an advisory write lock using std::fs.
tokio::task::spawn_blocking(move || -> Result<()> {
// Retry a few times to avoid indefinite blocking when contended.
@@ -117,8 +129,12 @@ pub(crate) async fn append_entry(
match history_file.try_lock() {
Ok(()) => {
// While holding the exclusive lock, write the full line.
// We do not open the file with `append(true)` on Windows, so ensure the
// cursor is positioned at the end before writing.
history_file.seek(SeekFrom::End(0))?;
history_file.write_all(line.as_bytes())?;
history_file.flush()?;
enforce_history_limit(&mut history_file, history_max_bytes)?;
return Ok(());
}
Err(std::fs::TryLockError::WouldBlock) => {
@@ -138,27 +154,144 @@ pub(crate) async fn append_entry(
Ok(())
}
/// Trim the history file to honor `max_bytes`, dropping the oldest lines while holding
/// the write lock so the newest entry is always retained. When the file exceeds the
/// hard cap, it rewrites the remaining tail to a soft cap to avoid trimming again
/// immediately on the next write.
fn enforce_history_limit(file: &mut File, max_bytes: Option<usize>) -> Result<()> {
let Some(max_bytes) = max_bytes else {
return Ok(());
};
if max_bytes == 0 {
return Ok(());
}
let max_bytes = match u64::try_from(max_bytes) {
Ok(value) => value,
Err(_) => return Ok(()),
};
let mut current_len = file.metadata()?.len();
if current_len <= max_bytes {
return Ok(());
}
let mut reader_file = file.try_clone()?;
reader_file.seek(SeekFrom::Start(0))?;
let mut buf_reader = BufReader::new(reader_file);
let mut line_lengths = Vec::new();
let mut line_buf = String::new();
loop {
line_buf.clear();
let bytes = buf_reader.read_line(&mut line_buf)?;
if bytes == 0 {
break;
}
line_lengths.push(bytes as u64);
}
if line_lengths.is_empty() {
return Ok(());
}
let last_index = line_lengths.len() - 1;
let trim_target = trim_target_bytes(max_bytes, line_lengths[last_index]);
let mut drop_bytes = 0u64;
let mut idx = 0usize;
while current_len > trim_target && idx < last_index {
current_len = current_len.saturating_sub(line_lengths[idx]);
drop_bytes += line_lengths[idx];
idx += 1;
}
if drop_bytes == 0 {
return Ok(());
}
let mut reader = buf_reader.into_inner();
reader.seek(SeekFrom::Start(drop_bytes))?;
let capacity = usize::try_from(current_len).unwrap_or(0);
let mut tail = Vec::with_capacity(capacity);
reader.read_to_end(&mut tail)?;
file.set_len(0)?;
file.seek(SeekFrom::Start(0))?;
file.write_all(&tail)?;
file.flush()?;
Ok(())
}
fn trim_target_bytes(max_bytes: u64, newest_entry_len: u64) -> u64 {
let soft_cap_bytes = ((max_bytes as f64) * HISTORY_SOFT_CAP_RATIO)
.floor()
.clamp(1.0, max_bytes as f64) as u64;
soft_cap_bytes.max(newest_entry_len)
}
/// Asynchronously fetch the history file's *identifier* (inode on Unix) and
/// the current number of entries by counting newline characters.
pub(crate) async fn history_metadata(config: &Config) -> (u64, usize) {
let path = history_filepath(config);
history_metadata_for_file(&path).await
}
#[cfg(unix)]
let log_id = {
use std::os::unix::fs::MetadataExt;
// Obtain metadata (async) to get the identifier.
let meta = match fs::metadata(&path).await {
Ok(m) => m,
Err(e) if e.kind() == std::io::ErrorKind::NotFound => return (0, 0),
Err(_) => return (0, 0),
};
meta.ino()
/// Given a `log_id` (on Unix this is the file's inode number,
/// on Windows this is the file's creation time) and a zero-based
/// `offset`, return the corresponding `HistoryEntry` if the identifier matches
/// the current history file **and** the requested offset exists. Any I/O or
/// parsing errors are logged and result in `None`.
///
/// Note this function is not async because it uses a sync advisory file
/// locking API.
pub(crate) fn lookup(log_id: u64, offset: usize, config: &Config) -> Option<HistoryEntry> {
let path = history_filepath(config);
lookup_history_entry(&path, log_id, offset)
}
/// On Unix systems, ensure the file permissions are `0o600` (rw-------). If the
/// permissions cannot be changed the error is propagated to the caller.
#[cfg(unix)]
async fn ensure_owner_only_permissions(file: &File) -> Result<()> {
let metadata = file.metadata()?;
let current_mode = metadata.permissions().mode() & 0o777;
if current_mode != 0o600 {
let mut perms = metadata.permissions();
perms.set_mode(0o600);
let perms_clone = perms.clone();
let file_clone = file.try_clone()?;
tokio::task::spawn_blocking(move || file_clone.set_permissions(perms_clone)).await??;
}
Ok(())
}
#[cfg(windows)]
// On Windows, simply succeed.
async fn ensure_owner_only_permissions(_file: &File) -> Result<()> {
Ok(())
}
async fn history_metadata_for_file(path: &Path) -> (u64, usize) {
let log_id = match fs::metadata(path).await {
Ok(metadata) => history_log_id(&metadata).unwrap_or(0),
Err(e) if e.kind() == std::io::ErrorKind::NotFound => return (0, 0),
Err(_) => return (0, 0),
};
#[cfg(not(unix))]
let log_id = 0u64;
// Open the file.
let mut file = match fs::File::open(&path).await {
let mut file = match fs::File::open(path).await {
Ok(f) => f,
Err(_) => return (log_id, 0),
};
@@ -179,21 +312,11 @@ pub(crate) async fn history_metadata(config: &Config) -> (u64, usize) {
(log_id, count)
}
/// Given a `log_id` (on Unix this is the file's inode number) and a zero-based
/// `offset`, return the corresponding `HistoryEntry` if the identifier matches
/// the current history file **and** the requested offset exists. Any I/O or
/// parsing errors are logged and result in `None`.
///
/// Note this function is not async because it uses a sync advisory file
/// locking API.
#[cfg(unix)]
pub(crate) fn lookup(log_id: u64, offset: usize, config: &Config) -> Option<HistoryEntry> {
fn lookup_history_entry(path: &Path, log_id: u64, offset: usize) -> Option<HistoryEntry> {
use std::io::BufRead;
use std::io::BufReader;
use std::os::unix::fs::MetadataExt;
let path = history_filepath(config);
let file: File = match OpenOptions::new().read(true).open(&path) {
let file: File = match OpenOptions::new().read(true).open(path) {
Ok(f) => f,
Err(e) => {
tracing::warn!(error = %e, "failed to open history file");
@@ -209,7 +332,9 @@ pub(crate) fn lookup(log_id: u64, offset: usize, config: &Config) -> Option<Hist
}
};
if metadata.ino() != log_id {
let current_log_id = history_log_id(&metadata)?;
if log_id != 0 && current_log_id != log_id {
return None;
}
@@ -256,31 +381,238 @@ pub(crate) fn lookup(log_id: u64, offset: usize, config: &Config) -> Option<Hist
None
}
/// Fallback stub for non-Unix systems: currently always returns `None`.
#[cfg(not(unix))]
pub(crate) fn lookup(log_id: u64, offset: usize, config: &Config) -> Option<HistoryEntry> {
let _ = (log_id, offset, config);
#[cfg(unix)]
fn history_log_id(metadata: &std::fs::Metadata) -> Option<u64> {
use std::os::unix::fs::MetadataExt;
Some(metadata.ino())
}
#[cfg(windows)]
fn history_log_id(metadata: &std::fs::Metadata) -> Option<u64> {
use std::os::windows::fs::MetadataExt;
Some(metadata.creation_time())
}
#[cfg(not(any(unix, windows)))]
fn history_log_id(_metadata: &std::fs::Metadata) -> Option<u64> {
None
}
/// On Unix systems ensure the file permissions are `0o600` (rw-------). If the
/// permissions cannot be changed the error is propagated to the caller.
#[cfg(unix)]
async fn ensure_owner_only_permissions(file: &File) -> Result<()> {
let metadata = file.metadata()?;
let current_mode = metadata.permissions().mode() & 0o777;
if current_mode != 0o600 {
let mut perms = metadata.permissions();
perms.set_mode(0o600);
let perms_clone = perms.clone();
let file_clone = file.try_clone()?;
tokio::task::spawn_blocking(move || file_clone.set_permissions(perms_clone)).await??;
}
Ok(())
}
#[cfg(test)]
mod tests {
use super::*;
use crate::config::Config;
use crate::config::ConfigOverrides;
use crate::config::ConfigToml;
use codex_protocol::ConversationId;
use pretty_assertions::assert_eq;
use std::fs::File;
use std::io::Write;
use tempfile::TempDir;
#[cfg(not(unix))]
async fn ensure_owner_only_permissions(_file: &File) -> Result<()> {
// For now, on non-Unix, simply succeed.
Ok(())
#[tokio::test]
async fn lookup_reads_history_entries() {
let temp_dir = TempDir::new().expect("create temp dir");
let history_path = temp_dir.path().join(HISTORY_FILENAME);
let entries = vec![
HistoryEntry {
session_id: "first-session".to_string(),
ts: 1,
text: "first".to_string(),
},
HistoryEntry {
session_id: "second-session".to_string(),
ts: 2,
text: "second".to_string(),
},
];
let mut file = File::create(&history_path).expect("create history file");
for entry in &entries {
writeln!(
file,
"{}",
serde_json::to_string(entry).expect("serialize history entry")
)
.expect("write history entry");
}
let (log_id, count) = history_metadata_for_file(&history_path).await;
assert_eq!(count, entries.len());
let second_entry =
lookup_history_entry(&history_path, log_id, 1).expect("fetch second history entry");
assert_eq!(second_entry, entries[1]);
}
#[tokio::test]
async fn lookup_uses_stable_log_id_after_appends() {
let temp_dir = TempDir::new().expect("create temp dir");
let history_path = temp_dir.path().join(HISTORY_FILENAME);
let initial = HistoryEntry {
session_id: "first-session".to_string(),
ts: 1,
text: "first".to_string(),
};
let appended = HistoryEntry {
session_id: "second-session".to_string(),
ts: 2,
text: "second".to_string(),
};
let mut file = File::create(&history_path).expect("create history file");
writeln!(
file,
"{}",
serde_json::to_string(&initial).expect("serialize initial entry")
)
.expect("write initial entry");
let (log_id, count) = history_metadata_for_file(&history_path).await;
assert_eq!(count, 1);
let mut append = std::fs::OpenOptions::new()
.append(true)
.open(&history_path)
.expect("open history file for append");
writeln!(
append,
"{}",
serde_json::to_string(&appended).expect("serialize appended entry")
)
.expect("append history entry");
let fetched =
lookup_history_entry(&history_path, log_id, 1).expect("lookup appended history entry");
assert_eq!(fetched, appended);
}
#[tokio::test]
async fn append_entry_trims_history_when_beyond_max_bytes() {
let codex_home = TempDir::new().expect("create temp dir");
let mut config = Config::load_from_base_config_with_overrides(
ConfigToml::default(),
ConfigOverrides::default(),
codex_home.path().to_path_buf(),
)
.expect("load config");
let conversation_id = ConversationId::new();
let entry_one = "a".repeat(200);
let entry_two = "b".repeat(200);
let history_path = codex_home.path().join("history.jsonl");
append_entry(&entry_one, &conversation_id, &config)
.await
.expect("write first entry");
let first_len = std::fs::metadata(&history_path).expect("metadata").len();
let limit_bytes = first_len + 10;
config.history.max_bytes =
Some(usize::try_from(limit_bytes).expect("limit should fit into usize"));
append_entry(&entry_two, &conversation_id, &config)
.await
.expect("write second entry");
let contents = std::fs::read_to_string(&history_path).expect("read history");
let entries = contents
.lines()
.map(|line| serde_json::from_str::<HistoryEntry>(line).expect("parse entry"))
.collect::<Vec<HistoryEntry>>();
assert_eq!(
entries.len(),
1,
"only one entry left because entry_one should be evicted"
);
assert_eq!(entries[0].text, entry_two);
assert!(std::fs::metadata(&history_path).expect("metadata").len() <= limit_bytes);
}
#[tokio::test]
async fn append_entry_trims_history_to_soft_cap() {
let codex_home = TempDir::new().expect("create temp dir");
let mut config = Config::load_from_base_config_with_overrides(
ConfigToml::default(),
ConfigOverrides::default(),
codex_home.path().to_path_buf(),
)
.expect("load config");
let conversation_id = ConversationId::new();
let short_entry = "a".repeat(200);
let long_entry = "b".repeat(400);
let history_path = codex_home.path().join("history.jsonl");
append_entry(&short_entry, &conversation_id, &config)
.await
.expect("write first entry");
let short_entry_len = std::fs::metadata(&history_path).expect("metadata").len();
append_entry(&long_entry, &conversation_id, &config)
.await
.expect("write second entry");
let two_entry_len = std::fs::metadata(&history_path).expect("metadata").len();
let long_entry_len = two_entry_len
.checked_sub(short_entry_len)
.expect("second entry length should be larger than first entry length");
config.history.max_bytes = Some(
usize::try_from((2 * long_entry_len) + (short_entry_len / 2))
.expect("max bytes should fit into usize"),
);
append_entry(&long_entry, &conversation_id, &config)
.await
.expect("write third entry");
let contents = std::fs::read_to_string(&history_path).expect("read history");
let entries = contents
.lines()
.map(|line| serde_json::from_str::<HistoryEntry>(line).expect("parse entry"))
.collect::<Vec<HistoryEntry>>();
assert_eq!(entries.len(), 1);
assert_eq!(entries[0].text, long_entry);
let pruned_len = std::fs::metadata(&history_path).expect("metadata").len();
let max_bytes = config
.history
.max_bytes
.expect("max bytes should be configured") as u64;
assert!(pruned_len <= max_bytes);
let soft_cap_bytes = ((max_bytes as f64) * HISTORY_SOFT_CAP_RATIO)
.floor()
.clamp(1.0, max_bytes as f64) as u64;
let len_without_first = 2 * long_entry_len;
assert!(
len_without_first <= max_bytes,
"dropping only the first entry would satisfy the hard cap"
);
assert!(
len_without_first > soft_cap_bytes,
"soft cap should require more aggressive trimming than the hard cap"
);
assert_eq!(pruned_len, long_entry_len);
assert!(pruned_len <= soft_cap_bytes.max(long_entry_len));
}
}

View File

@@ -76,6 +76,8 @@ pub(crate) fn get_model_info(model_family: &ModelFamily) -> Option<ModelInfo> {
_ if slug.starts_with("codex-") => Some(ModelInfo::new(CONTEXT_WINDOW_272K)),
_ if slug.starts_with("exp-") => Some(ModelInfo::new(CONTEXT_WINDOW_272K)),
_ => None,
}
}

View File

@@ -14,6 +14,9 @@
//! 3. We do **not** walk past the Git root.
use crate::config::Config;
use crate::features::Feature;
use crate::skills::load_skills;
use crate::skills::render_skills_section;
use dunce::canonicalize as normalize_path;
use std::path::PathBuf;
use tokio::io::AsyncReadExt;
@@ -31,18 +34,47 @@ const PROJECT_DOC_SEPARATOR: &str = "\n\n--- project-doc ---\n\n";
/// Combines `Config::instructions` and `AGENTS.md` (if present) into a single
/// string of instructions.
pub(crate) async fn get_user_instructions(config: &Config) -> Option<String> {
match read_project_docs(config).await {
Ok(Some(project_doc)) => match &config.user_instructions {
Some(original_instructions) => Some(format!(
"{original_instructions}{PROJECT_DOC_SEPARATOR}{project_doc}"
)),
None => Some(project_doc),
},
Ok(None) => config.user_instructions.clone(),
let skills_section = if config.features.enabled(Feature::Skills) {
let skills_outcome = load_skills(config);
for err in &skills_outcome.errors {
error!(
"failed to load skill {}: {}",
err.path.display(),
err.message
);
}
render_skills_section(&skills_outcome.skills)
} else {
None
};
let project_docs = match read_project_docs(config).await {
Ok(docs) => docs,
Err(e) => {
error!("error trying to find project doc: {e:#}");
config.user_instructions.clone()
return config.user_instructions.clone();
}
};
let combined_project_docs = merge_project_docs_with_skills(project_docs, skills_section);
let mut parts: Vec<String> = Vec::new();
if let Some(instructions) = config.user_instructions.clone() {
parts.push(instructions);
}
if let Some(project_doc) = combined_project_docs {
if !parts.is_empty() {
parts.push(PROJECT_DOC_SEPARATOR.to_string());
}
parts.push(project_doc);
}
if parts.is_empty() {
None
} else {
Some(parts.concat())
}
}
@@ -195,12 +227,25 @@ fn candidate_filenames<'a>(config: &'a Config) -> Vec<&'a str> {
names
}
fn merge_project_docs_with_skills(
project_doc: Option<String>,
skills_section: Option<String>,
) -> Option<String> {
match (project_doc, skills_section) {
(Some(doc), Some(skills)) => Some(format!("{doc}\n\n{skills}")),
(Some(doc), None) => Some(doc),
(None, Some(skills)) => Some(skills),
(None, None) => None,
}
}
#[cfg(test)]
mod tests {
use super::*;
use crate::config::ConfigOverrides;
use crate::config::ConfigToml;
use std::fs;
use std::path::PathBuf;
use tempfile::TempDir;
/// Helper that returns a `Config` pointing at `root` and using `limit` as
@@ -219,6 +264,7 @@ mod tests {
config.cwd = root.path().to_path_buf();
config.project_doc_max_bytes = limit;
config.features.enable(Feature::Skills);
config.user_instructions = instructions.map(ToOwned::to_owned);
config
@@ -447,4 +493,58 @@ mod tests {
.eq(DEFAULT_PROJECT_DOC_FILENAME)
);
}
#[tokio::test]
async fn skills_are_appended_to_project_doc() {
let tmp = tempfile::tempdir().expect("tempdir");
fs::write(tmp.path().join("AGENTS.md"), "base doc").unwrap();
let cfg = make_config(&tmp, 4096, None);
create_skill(
cfg.codex_home.clone(),
"pdf-processing",
"extract from pdfs",
);
let res = get_user_instructions(&cfg)
.await
.expect("instructions expected");
let expected_path = dunce::canonicalize(
cfg.codex_home
.join("skills/pdf-processing/SKILL.md")
.as_path(),
)
.unwrap_or_else(|_| cfg.codex_home.join("skills/pdf-processing/SKILL.md"));
let expected_path_str = expected_path.to_string_lossy().replace('\\', "/");
let expected = format!(
"base doc\n\n## Skills\nThese skills are discovered at startup from ~/.codex/skills; each entry shows name, description, and file path so you can open the source for full instructions. Content is not inlined to keep context lean.\n- pdf-processing: extract from pdfs (file: {expected_path_str})"
);
assert_eq!(res, expected);
}
#[tokio::test]
async fn skills_render_without_project_doc() {
let tmp = tempfile::tempdir().expect("tempdir");
let cfg = make_config(&tmp, 4096, None);
create_skill(cfg.codex_home.clone(), "linting", "run clippy");
let res = get_user_instructions(&cfg)
.await
.expect("instructions expected");
let expected_path =
dunce::canonicalize(cfg.codex_home.join("skills/linting/SKILL.md").as_path())
.unwrap_or_else(|_| cfg.codex_home.join("skills/linting/SKILL.md"));
let expected_path_str = expected_path.to_string_lossy().replace('\\', "/");
let expected = format!(
"## Skills\nThese skills are discovered at startup from ~/.codex/skills; each entry shows name, description, and file path so you can open the source for full instructions. Content is not inlined to keep context lean.\n- linting: run clippy (file: {expected_path_str})"
);
assert_eq!(res, expected);
}
fn create_skill(codex_home: PathBuf, name: &str, description: &str) {
let skill_dir = codex_home.join(format!("skills/{name}"));
fs::create_dir_all(&skill_dir).unwrap();
let content = format!("---\nname: {name}\ndescription: {description}\n---\n\n# Body\n");
fs::write(skill_dir.join("SKILL.md"), content).unwrap();
}
}

View File

@@ -1,4 +1,5 @@
use crate::protocol::ReviewFinding;
use crate::protocol::ReviewOutputEvent;
// Note: We keep this module UI-agnostic. It returns plain strings that
// higher layers (e.g., TUI) may style as needed.
@@ -10,6 +11,8 @@ fn format_location(item: &ReviewFinding) -> String {
format!("{path}:{start}-{end}")
}
const REVIEW_FALLBACK_MESSAGE: &str = "Reviewer failed to output a response.";
/// Format a full review findings block as plain text lines.
///
/// - When `selection` is `Some`, each item line includes a checkbox marker:
@@ -53,3 +56,27 @@ pub fn format_review_findings_block(
lines.join("\n")
}
/// Render a human-readable review summary suitable for a user-facing message.
///
/// Returns either the explanation, the formatted findings block, or both
/// separated by a blank line. If neither is present, emits a fallback message.
pub fn render_review_output_text(output: &ReviewOutputEvent) -> String {
let mut sections = Vec::new();
let explanation = output.overall_explanation.trim();
if !explanation.is_empty() {
sections.push(explanation.to_string());
}
if !output.findings.is_empty() {
let findings = format_review_findings_block(&output.findings, None);
let trimmed = findings.trim();
if !trimmed.is_empty() {
sections.push(trimmed.to_string());
}
}
if sections.is_empty() {
REVIEW_FALLBACK_MESSAGE.to_string()
} else {
sections.join("\n\n")
}
}

View File

@@ -0,0 +1,93 @@
use codex_git::merge_base_with_head;
use codex_protocol::protocol::ReviewRequest;
use codex_protocol::protocol::ReviewTarget;
use std::path::Path;
#[derive(Clone, Debug, PartialEq)]
pub struct ResolvedReviewRequest {
pub target: ReviewTarget,
pub prompt: String,
pub user_facing_hint: String,
}
const UNCOMMITTED_PROMPT: &str = "Review the current code changes (staged, unstaged, and untracked files) and provide prioritized findings.";
const BASE_BRANCH_PROMPT_BACKUP: &str = "Review the code changes against the base branch '{branch}'. Start by finding the merge diff between the current branch and {branch}'s upstream e.g. (`git merge-base HEAD \"$(git rev-parse --abbrev-ref \"{branch}@{upstream}\")\"`), then run `git diff` against that SHA to see what changes we would merge into the {branch} branch. Provide prioritized, actionable findings.";
const BASE_BRANCH_PROMPT: &str = "Review the code changes against the base branch '{baseBranch}'. The merge base commit for this comparison is {mergeBaseSha}. Run `git diff {mergeBaseSha}` to inspect the changes relative to {baseBranch}. Provide prioritized, actionable findings.";
const COMMIT_PROMPT_WITH_TITLE: &str = "Review the code changes introduced by commit {sha} (\"{title}\"). Provide prioritized, actionable findings.";
const COMMIT_PROMPT: &str =
"Review the code changes introduced by commit {sha}. Provide prioritized, actionable findings.";
pub fn resolve_review_request(
request: ReviewRequest,
cwd: &Path,
) -> anyhow::Result<ResolvedReviewRequest> {
let target = request.target;
let prompt = review_prompt(&target, cwd)?;
let user_facing_hint = request
.user_facing_hint
.unwrap_or_else(|| user_facing_hint(&target));
Ok(ResolvedReviewRequest {
target,
prompt,
user_facing_hint,
})
}
pub fn review_prompt(target: &ReviewTarget, cwd: &Path) -> anyhow::Result<String> {
match target {
ReviewTarget::UncommittedChanges => Ok(UNCOMMITTED_PROMPT.to_string()),
ReviewTarget::BaseBranch { branch } => {
if let Some(commit) = merge_base_with_head(cwd, branch)? {
Ok(BASE_BRANCH_PROMPT
.replace("{baseBranch}", branch)
.replace("{mergeBaseSha}", &commit))
} else {
Ok(BASE_BRANCH_PROMPT_BACKUP.replace("{branch}", branch))
}
}
ReviewTarget::Commit { sha, title } => {
if let Some(title) = title {
Ok(COMMIT_PROMPT_WITH_TITLE
.replace("{sha}", sha)
.replace("{title}", title))
} else {
Ok(COMMIT_PROMPT.replace("{sha}", sha))
}
}
ReviewTarget::Custom { instructions } => {
let prompt = instructions.trim();
if prompt.is_empty() {
anyhow::bail!("Review prompt cannot be empty");
}
Ok(prompt.to_string())
}
}
}
pub fn user_facing_hint(target: &ReviewTarget) -> String {
match target {
ReviewTarget::UncommittedChanges => "current changes".to_string(),
ReviewTarget::BaseBranch { branch } => format!("changes against '{branch}'"),
ReviewTarget::Commit { sha, title } => {
let short_sha: String = sha.chars().take(7).collect();
if let Some(title) = title {
format!("commit {short_sha}: {title}")
} else {
format!("commit {short_sha}")
}
}
ReviewTarget::Custom { instructions } => instructions.trim().to_string(),
}
}
impl From<ResolvedReviewRequest> for ReviewRequest {
fn from(resolved: ResolvedReviewRequest) -> Self {
ReviewRequest {
target: resolved.target,
user_facing_hint: Some(resolved.user_facing_hint),
}
}
}

View File

@@ -0,0 +1,49 @@
use std::io::ErrorKind;
use std::path::Path;
use crate::error::CodexErr;
use crate::rollout::SESSIONS_SUBDIR;
pub(crate) fn map_session_init_error(err: &anyhow::Error, codex_home: &Path) -> CodexErr {
if let Some(mapped) = err
.chain()
.filter_map(|cause| cause.downcast_ref::<std::io::Error>())
.find_map(|io_err| map_rollout_io_error(io_err, codex_home))
{
return mapped;
}
CodexErr::Fatal(format!("Failed to initialize session: {err:#}"))
}
fn map_rollout_io_error(io_err: &std::io::Error, codex_home: &Path) -> Option<CodexErr> {
let sessions_dir = codex_home.join(SESSIONS_SUBDIR);
let hint = match io_err.kind() {
ErrorKind::PermissionDenied => format!(
"Codex cannot access session files at {} (permission denied). If sessions were created using sudo, fix ownership: sudo chown -R $(whoami) {}",
sessions_dir.display(),
codex_home.display()
),
ErrorKind::NotFound => format!(
"Session storage missing at {}. Create the directory or choose a different Codex home.",
sessions_dir.display()
),
ErrorKind::AlreadyExists => format!(
"Session storage path {} is blocked by an existing file. Remove or rename it so Codex can create sessions.",
sessions_dir.display()
),
ErrorKind::InvalidData | ErrorKind::InvalidInput => format!(
"Session data under {} looks corrupt or unreadable. Clearing the sessions directory may help (this will remove saved conversations).",
sessions_dir.display()
),
ErrorKind::IsADirectory | ErrorKind::NotADirectory => format!(
"Session storage path {} has an unexpected type. Ensure it is a directory Codex can use for session files.",
sessions_dir.display()
),
_ => return None,
};
Some(CodexErr::Fatal(format!(
"{hint} (underlying error: {io_err})"
)))
}

View File

@@ -9,6 +9,7 @@ use std::sync::atomic::AtomicBool;
use time::OffsetDateTime;
use time::PrimitiveDateTime;
use time::format_description::FormatItem;
use time::format_description::well_known::Rfc3339;
use time::macros::format_description;
use uuid::Uuid;
@@ -39,18 +40,15 @@ pub struct ConversationItem {
pub path: PathBuf,
/// First up to `HEAD_RECORD_LIMIT` JSONL records parsed as JSON (includes meta line).
pub head: Vec<serde_json::Value>,
/// Last up to `TAIL_RECORD_LIMIT` JSONL response records parsed as JSON.
pub tail: Vec<serde_json::Value>,
/// RFC3339 timestamp string for when the session was created, if available.
pub created_at: Option<String>,
/// RFC3339 timestamp string for the most recent response in the tail, if available.
/// RFC3339 timestamp string for the most recent update (from file mtime).
pub updated_at: Option<String>,
}
#[derive(Default)]
struct HeadTailSummary {
head: Vec<serde_json::Value>,
tail: Vec<serde_json::Value>,
saw_session_meta: bool,
saw_user_event: bool,
source: Option<SessionSource>,
@@ -62,7 +60,6 @@ struct HeadTailSummary {
/// Hard cap to bound worstcase work per request.
const MAX_SCAN_FILES: usize = 10000;
const HEAD_RECORD_LIMIT: usize = 10;
const TAIL_RECORD_LIMIT: usize = 10;
/// Pagination cursor identifying a file by timestamp and UUID.
#[derive(Debug, Clone, PartialEq, Eq)]
@@ -141,13 +138,6 @@ pub(crate) async fn get_conversations(
Ok(result)
}
/// Load the full contents of a single conversation session file at `path`.
/// Returns the entire file contents as a String.
#[allow(dead_code)]
pub(crate) async fn get_conversation(path: &Path) -> io::Result<String> {
tokio::fs::read_to_string(path).await
}
/// Load conversation file paths from disk using directory traversal.
///
/// Directory layout: `~/.codex/sessions/YYYY/MM/DD/rollout-YYYY-MM-DDThh-mm-ss-<uuid>.jsonl`
@@ -212,9 +202,8 @@ async fn traverse_directories_for_paths(
more_matches_available = true;
break 'outer;
}
// Read head and simultaneously detect message events within the same
// first N JSONL records to avoid a second file read.
let summary = read_head_and_tail(&path, HEAD_RECORD_LIMIT, TAIL_RECORD_LIMIT)
// Read head and detect message events; stop once meta + user are found.
let summary = read_head_summary(&path, HEAD_RECORD_LIMIT)
.await
.unwrap_or_default();
if !allowed_sources.is_empty()
@@ -233,16 +222,19 @@ async fn traverse_directories_for_paths(
if summary.saw_session_meta && summary.saw_user_event {
let HeadTailSummary {
head,
tail,
created_at,
mut updated_at,
..
} = summary;
updated_at = updated_at.or_else(|| created_at.clone());
if updated_at.is_none() {
updated_at = file_modified_rfc3339(&path)
.await
.unwrap_or(None)
.or_else(|| created_at.clone());
}
items.push(ConversationItem {
path,
head,
tail,
created_at,
updated_at,
});
@@ -384,11 +376,7 @@ impl<'a> ProviderMatcher<'a> {
}
}
async fn read_head_and_tail(
path: &Path,
head_limit: usize,
tail_limit: usize,
) -> io::Result<HeadTailSummary> {
async fn read_head_summary(path: &Path, head_limit: usize) -> io::Result<HeadTailSummary> {
use tokio::io::AsyncBufReadExt;
let file = tokio::fs::File::open(path).await?;
@@ -441,107 +429,30 @@ async fn read_head_and_tail(
}
}
}
if summary.saw_session_meta && summary.saw_user_event {
break;
}
}
if tail_limit != 0 {
let (tail, updated_at) = read_tail_records(path, tail_limit).await?;
summary.tail = tail;
summary.updated_at = updated_at;
}
Ok(summary)
}
/// Read up to `HEAD_RECORD_LIMIT` records from the start of the rollout file at `path`.
/// This should be enough to produce a summary including the session meta line.
pub async fn read_head_for_summary(path: &Path) -> io::Result<Vec<serde_json::Value>> {
let summary = read_head_and_tail(path, HEAD_RECORD_LIMIT, 0).await?;
let summary = read_head_summary(path, HEAD_RECORD_LIMIT).await?;
Ok(summary.head)
}
async fn read_tail_records(
path: &Path,
max_records: usize,
) -> io::Result<(Vec<serde_json::Value>, Option<String>)> {
use std::io::SeekFrom;
use tokio::io::AsyncReadExt;
use tokio::io::AsyncSeekExt;
if max_records == 0 {
return Ok((Vec::new(), None));
}
const CHUNK_SIZE: usize = 8192;
let mut file = tokio::fs::File::open(path).await?;
let mut pos = file.seek(SeekFrom::End(0)).await?;
if pos == 0 {
return Ok((Vec::new(), None));
}
let mut buffer: Vec<u8> = Vec::new();
let mut latest_timestamp: Option<String> = None;
loop {
let slice_start = match (pos > 0, buffer.iter().position(|&b| b == b'\n')) {
(true, Some(idx)) => idx + 1,
_ => 0,
};
let (tail, newest_ts) = collect_last_response_values(&buffer[slice_start..], max_records);
if latest_timestamp.is_none() {
latest_timestamp = newest_ts.clone();
}
if tail.len() >= max_records || pos == 0 {
return Ok((tail, latest_timestamp.or(newest_ts)));
}
let read_size = CHUNK_SIZE.min(pos as usize);
if read_size == 0 {
return Ok((tail, latest_timestamp.or(newest_ts)));
}
pos -= read_size as u64;
file.seek(SeekFrom::Start(pos)).await?;
let mut chunk = vec![0; read_size];
file.read_exact(&mut chunk).await?;
chunk.extend_from_slice(&buffer);
buffer = chunk;
}
}
fn collect_last_response_values(
buffer: &[u8],
max_records: usize,
) -> (Vec<serde_json::Value>, Option<String>) {
use std::borrow::Cow;
if buffer.is_empty() || max_records == 0 {
return (Vec::new(), None);
}
let text: Cow<'_, str> = String::from_utf8_lossy(buffer);
let mut collected_rev: Vec<serde_json::Value> = Vec::new();
let mut latest_timestamp: Option<String> = None;
for line in text.lines().rev() {
let trimmed = line.trim();
if trimmed.is_empty() {
continue;
}
let parsed: serde_json::Result<RolloutLine> = serde_json::from_str(trimmed);
let Ok(rollout_line) = parsed else { continue };
let RolloutLine { timestamp, item } = rollout_line;
if let RolloutItem::ResponseItem(item) = item
&& let Ok(val) = serde_json::to_value(&item)
{
if latest_timestamp.is_none() {
latest_timestamp = Some(timestamp.clone());
}
collected_rev.push(val);
if collected_rev.len() == max_records {
break;
}
}
}
collected_rev.reverse();
(collected_rev, latest_timestamp)
async fn file_modified_rfc3339(path: &Path) -> io::Result<Option<String>> {
let meta = tokio::fs::metadata(path).await?;
let modified = meta.modified().ok();
let Some(modified) = modified else {
return Ok(None);
};
let dt = OffsetDateTime::from(modified);
Ok(dt.format(&Rfc3339).ok())
}
/// Locate a recorded conversation rollout file by its UUID string using the existing

View File

@@ -7,11 +7,13 @@ pub const ARCHIVED_SESSIONS_SUBDIR: &str = "archived_sessions";
pub const INTERACTIVE_SESSION_SOURCES: &[SessionSource] =
&[SessionSource::Cli, SessionSource::VSCode];
pub(crate) mod error;
pub mod list;
pub(crate) mod policy;
pub mod recorder;
pub use codex_protocol::protocol::SessionMeta;
pub(crate) use error::map_session_init_error;
pub use list::find_conversation_path_by_id_str;
pub use recorder::RolloutRecorder;
pub use recorder::RolloutRecorderParams;

View File

@@ -68,6 +68,7 @@ pub(crate) fn should_persist_event_msg(ev: &EventMsg) -> bool {
| EventMsg::ElicitationRequest(_)
| EventMsg::ApplyPatchApprovalRequest(_)
| EventMsg::BackgroundEvent(_)
| EventMsg::SaveSessionResponse(_)
| EventMsg::StreamError(_)
| EventMsg::PatchApplyBegin(_)
| EventMsg::PatchApplyEnd(_)

View File

@@ -11,6 +11,8 @@ use serde_json::Value;
use time::OffsetDateTime;
use time::format_description::FormatItem;
use time::macros::format_description;
use tokio::io::AsyncReadExt;
use tokio::io::AsyncSeekExt;
use tokio::io::AsyncWriteExt;
use tokio::sync::mpsc::Sender;
use tokio::sync::mpsc::{self};
@@ -70,6 +72,10 @@ enum RolloutCmd {
Shutdown {
ack: oneshot::Sender<()>,
},
SetName {
name: Option<String>,
ack: oneshot::Sender<std::io::Result<()>>,
},
}
impl RolloutRecorderParams {
@@ -148,11 +154,14 @@ impl RolloutRecorder {
instructions,
source,
model_provider: Some(config.model_provider_id.clone()),
name: None,
}),
)
}
RolloutRecorderParams::Resume { path } => (
tokio::fs::OpenOptions::new()
.read(true)
.write(true)
.append(true)
.open(&path)
.await?,
@@ -196,6 +205,21 @@ impl RolloutRecorder {
.map_err(|e| IoError::other(format!("failed to queue rollout items: {e}")))
}
/// Update the session name stored in the rollout's SessionMeta line.
pub async fn set_session_name(&self, name: Option<String>) -> std::io::Result<()> {
let (tx, rx) = oneshot::channel();
self.tx
.send(RolloutCmd::SetName { name, ack: tx })
.await
.map_err(|e| IoError::other(format!("failed to queue session name update: {e}")))?;
match rx.await {
Ok(result) => result,
Err(e) => Err(IoError::other(format!(
"failed waiting for session name update: {e}"
))),
}
}
/// Flush all queued writes and wait until they are committed by the writer task.
pub async fn flush(&self) -> std::io::Result<()> {
let (tx, rx) = oneshot::channel();
@@ -334,6 +358,7 @@ fn create_log_file(
let path = dir.join(filename);
let file = std::fs::OpenOptions::new()
.read(true)
.append(true)
.create(true)
.open(&path)?;
@@ -389,6 +414,10 @@ async fn rollout_writer(
RolloutCmd::Shutdown { ack } => {
let _ = ack.send(());
}
RolloutCmd::SetName { name, ack } => {
let result = rewrite_session_meta_name(&mut writer.file, name).await;
let _ = ack.send(result);
}
}
}
@@ -422,3 +451,232 @@ impl JsonlWriter {
Ok(())
}
}
async fn rewrite_session_meta_name(
file: &mut tokio::fs::File,
name: Option<String>,
) -> std::io::Result<()> {
use std::io::SeekFrom;
file.flush().await?;
file.seek(SeekFrom::Start(0)).await?;
let mut contents = Vec::new();
file.read_to_end(&mut contents).await?;
if contents.is_empty() {
return Err(IoError::other("empty rollout file"));
}
let newline_idx = contents
.iter()
.position(|&b| b == b'\n')
.ok_or_else(|| IoError::other("rollout missing newline after SessionMeta"))?;
let first_line = &contents[..newline_idx];
let mut rollout_line: RolloutLine = serde_json::from_slice(first_line)
.map_err(|e| IoError::other(format!("failed to parse SessionMeta: {e}")))?;
let RolloutItem::SessionMeta(ref mut session_meta_line) = rollout_line.item else {
return Err(IoError::other("first rollout item is not SessionMeta"));
};
session_meta_line.meta.name = name;
let mut updated = serde_json::to_vec(&rollout_line)?;
updated.push(b'\n');
updated.extend_from_slice(&contents[newline_idx + 1..]);
file.set_len(0).await?;
file.seek(SeekFrom::Start(0)).await?;
file.write_all(&updated).await?;
file.flush().await?;
file.seek(SeekFrom::End(0)).await?;
Ok(())
}
#[cfg(test)]
mod tests {
use super::rewrite_session_meta_name;
use codex_protocol::ConversationId;
use codex_protocol::models::ContentItem;
use codex_protocol::models::ResponseItem;
use codex_protocol::protocol::RolloutItem;
use codex_protocol::protocol::RolloutLine;
use codex_protocol::protocol::SessionMeta;
use codex_protocol::protocol::SessionMetaLine;
use tempfile::NamedTempFile;
use tokio::fs::OpenOptions;
use tokio::io::AsyncReadExt;
use tokio::io::AsyncSeekExt;
use tokio::io::AsyncWriteExt;
fn sample_meta(name: Option<&str>) -> RolloutItem {
RolloutItem::SessionMeta(SessionMetaLine {
meta: SessionMeta {
id: ConversationId::from_string("00000000-0000-4000-8000-000000000001")
.expect("conversation id"),
timestamp: "2025-01-01T00:00:00.000Z".to_string(),
cwd: "/tmp".into(),
originator: "tester".to_string(),
cli_version: "1.0.0".to_string(),
instructions: None,
source: codex_protocol::protocol::SessionSource::Cli,
model_provider: Some("provider".to_string()),
name: name.map(str::to_string),
},
git: None,
})
}
fn sample_line() -> RolloutLine {
RolloutLine {
timestamp: "2025-01-01T00:00:00.000Z".to_string(),
item: sample_meta(None),
}
}
async fn write_rollout(lines: &[RolloutLine]) -> (NamedTempFile, tokio::fs::File) {
let temp = NamedTempFile::new().expect("temp file");
let mut file = OpenOptions::new()
.read(true)
.write(true)
.truncate(true)
.create(true)
.open(temp.path())
.await
.expect("open temp file");
for line in lines {
let mut json = serde_json::to_vec(line).expect("serialize line");
json.push(b'\n');
file.write_all(&json).await.expect("write line");
}
file.seek(std::io::SeekFrom::Start(0))
.await
.expect("rewind");
(temp, file)
}
async fn read_first_line(path: &std::path::Path) -> RolloutLine {
let mut contents = String::new();
let mut file = OpenOptions::new()
.read(true)
.open(path)
.await
.expect("open for read");
file.read_to_string(&mut contents).await.expect("read file");
let first = contents.lines().next().expect("first line");
serde_json::from_str(first).expect("parse first line")
}
#[tokio::test]
async fn updates_meta_name_and_preserves_rest() {
let events = vec![
sample_line(),
RolloutLine {
timestamp: "2025-01-01T00:00:01.000Z".to_string(),
item: RolloutItem::ResponseItem(ResponseItem::Message {
id: None,
role: "assistant".to_string(),
content: vec![ContentItem::OutputText {
text: "hello".to_string(),
}],
}),
},
];
let (temp, mut file) = write_rollout(&events).await;
rewrite_session_meta_name(&mut file, Some("renamed".to_string()))
.await
.expect("rewrite ok");
let first = read_first_line(temp.path()).await;
let RolloutItem::SessionMeta(meta_line) = first.item else {
panic!("expected SessionMeta line");
};
assert_eq!(meta_line.meta.name.as_deref(), Some("renamed"));
let contents = tokio::fs::read_to_string(temp.path())
.await
.expect("read file");
let lines: Vec<_> = contents.lines().collect();
assert_eq!(lines.len(), 2);
let parsed: RolloutLine = serde_json::from_str(lines[1]).expect("parse second line");
let RolloutItem::ResponseItem(ResponseItem::Message { role, content, .. }) = parsed.item
else {
panic!("expected response item");
};
assert_eq!(role, "assistant");
assert_eq!(
content,
vec![ContentItem::OutputText {
text: "hello".to_string()
}]
);
}
#[tokio::test]
async fn clearing_name_sets_none() {
let mut first = sample_line();
first.item = sample_meta(Some("existing"));
let (temp, mut file) = write_rollout(&[first]).await;
rewrite_session_meta_name(&mut file, None)
.await
.expect("rewrite ok");
let first = read_first_line(temp.path()).await;
let RolloutItem::SessionMeta(meta_line) = first.item else {
panic!("expected SessionMeta line");
};
assert_eq!(meta_line.meta.name, None);
}
#[tokio::test]
async fn errors_on_empty_file() {
let temp = NamedTempFile::new().expect("temp file");
let mut file = OpenOptions::new()
.read(true)
.write(true)
.open(temp.path())
.await
.expect("open temp file");
let err = rewrite_session_meta_name(&mut file, Some("x".to_string()))
.await
.expect_err("expected error");
assert!(format!("{err}").contains("empty rollout file"));
}
#[tokio::test]
async fn errors_when_first_line_not_session_meta() {
let wrong = RolloutLine {
timestamp: "t".to_string(),
item: RolloutItem::ResponseItem(ResponseItem::Message {
id: None,
role: "assistant".to_string(),
content: vec![ContentItem::OutputText {
text: "hello".to_string(),
}],
}),
};
let (_temp, mut file) = write_rollout(&[wrong]).await;
let err = rewrite_session_meta_name(&mut file, Some("x".to_string()))
.await
.expect_err("expected error");
assert!(format!("{err}").contains("first rollout item is not SessionMeta"));
// ensure file pointer is rewound to end after failure paths
let pos = file
.seek(std::io::SeekFrom::Current(0))
.await
.expect("seek");
assert!(pos > 0);
}
#[tokio::test]
async fn errors_when_missing_newline() {
let temp = NamedTempFile::new().expect("temp file");
let mut file = OpenOptions::new()
.read(true)
.write(true)
.open(temp.path())
.await
.expect("open temp file");
file.write_all(b"no newline").await.expect("write");
let err = rewrite_session_meta_name(&mut file, Some("x".to_string()))
.await
.expect_err("expected error");
assert!(format!("{err}").contains("rollout missing newline after SessionMeta"));
}
}

View File

@@ -16,13 +16,11 @@ use crate::rollout::INTERACTIVE_SESSION_SOURCES;
use crate::rollout::list::ConversationItem;
use crate::rollout::list::ConversationsPage;
use crate::rollout::list::Cursor;
use crate::rollout::list::get_conversation;
use crate::rollout::list::get_conversations;
use anyhow::Result;
use codex_protocol::ConversationId;
use codex_protocol::models::ContentItem;
use codex_protocol::models::ResponseItem;
use codex_protocol::protocol::CompactedItem;
use codex_protocol::protocol::EventMsg;
use codex_protocol::protocol::RolloutItem;
use codex_protocol::protocol::RolloutLine;
@@ -226,28 +224,28 @@ async fn test_list_conversations_latest_first() {
"model_provider": "test-provider",
})];
let updated_times: Vec<Option<String>> =
page.items.iter().map(|i| i.updated_at.clone()).collect();
let expected = ConversationsPage {
items: vec![
ConversationItem {
path: p1,
head: head_3,
tail: Vec::new(),
created_at: Some("2025-01-03T12-00-00".into()),
updated_at: Some("2025-01-03T12-00-00".into()),
updated_at: updated_times.first().cloned().flatten(),
},
ConversationItem {
path: p2,
head: head_2,
tail: Vec::new(),
created_at: Some("2025-01-02T12-00-00".into()),
updated_at: Some("2025-01-02T12-00-00".into()),
updated_at: updated_times.get(1).cloned().flatten(),
},
ConversationItem {
path: p3,
head: head_1,
tail: Vec::new(),
created_at: Some("2025-01-01T12-00-00".into()),
updated_at: Some("2025-01-01T12-00-00".into()),
updated_at: updated_times.get(2).cloned().flatten(),
},
],
next_cursor: None,
@@ -355,6 +353,8 @@ async fn test_pagination_cursor() {
"source": "vscode",
"model_provider": "test-provider",
})];
let updated_page1: Vec<Option<String>> =
page1.items.iter().map(|i| i.updated_at.clone()).collect();
let expected_cursor1: Cursor =
serde_json::from_str(&format!("\"2025-03-04T09-00-00|{u4}\"")).unwrap();
let expected_page1 = ConversationsPage {
@@ -362,16 +362,14 @@ async fn test_pagination_cursor() {
ConversationItem {
path: p5,
head: head_5,
tail: Vec::new(),
created_at: Some("2025-03-05T09-00-00".into()),
updated_at: Some("2025-03-05T09-00-00".into()),
updated_at: updated_page1.first().cloned().flatten(),
},
ConversationItem {
path: p4,
head: head_4,
tail: Vec::new(),
created_at: Some("2025-03-04T09-00-00".into()),
updated_at: Some("2025-03-04T09-00-00".into()),
updated_at: updated_page1.get(1).cloned().flatten(),
},
],
next_cursor: Some(expected_cursor1.clone()),
@@ -422,6 +420,8 @@ async fn test_pagination_cursor() {
"source": "vscode",
"model_provider": "test-provider",
})];
let updated_page2: Vec<Option<String>> =
page2.items.iter().map(|i| i.updated_at.clone()).collect();
let expected_cursor2: Cursor =
serde_json::from_str(&format!("\"2025-03-02T09-00-00|{u2}\"")).unwrap();
let expected_page2 = ConversationsPage {
@@ -429,16 +429,14 @@ async fn test_pagination_cursor() {
ConversationItem {
path: p3,
head: head_3,
tail: Vec::new(),
created_at: Some("2025-03-03T09-00-00".into()),
updated_at: Some("2025-03-03T09-00-00".into()),
updated_at: updated_page2.first().cloned().flatten(),
},
ConversationItem {
path: p2,
head: head_2,
tail: Vec::new(),
created_at: Some("2025-03-02T09-00-00".into()),
updated_at: Some("2025-03-02T09-00-00".into()),
updated_at: updated_page2.get(1).cloned().flatten(),
},
],
next_cursor: Some(expected_cursor2.clone()),
@@ -473,13 +471,14 @@ async fn test_pagination_cursor() {
"source": "vscode",
"model_provider": "test-provider",
})];
let updated_page3: Vec<Option<String>> =
page3.items.iter().map(|i| i.updated_at.clone()).collect();
let expected_page3 = ConversationsPage {
items: vec![ConversationItem {
path: p1,
head: head_1,
tail: Vec::new(),
created_at: Some("2025-03-01T09-00-00".into()),
updated_at: Some("2025-03-01T09-00-00".into()),
updated_at: updated_page3.first().cloned().flatten(),
}],
next_cursor: None,
num_scanned_files: 5, // scanned 05, 04 (anchor), 03, 02 (anchor), 01
@@ -510,7 +509,7 @@ async fn test_get_conversation_contents() {
.unwrap();
let path = &page.items[0].path;
let content = get_conversation(path).await.unwrap();
let content = tokio::fs::read_to_string(path).await.unwrap();
// Page equality (single item)
let expected_path = home
@@ -533,9 +532,8 @@ async fn test_get_conversation_contents() {
items: vec![ConversationItem {
path: expected_path,
head: expected_head,
tail: Vec::new(),
created_at: Some(ts.into()),
updated_at: Some(ts.into()),
updated_at: page.items[0].updated_at.clone(),
}],
next_cursor: None,
num_scanned_files: 1,
@@ -570,7 +568,7 @@ async fn test_get_conversation_contents() {
}
#[tokio::test]
async fn test_tail_includes_last_response_items() -> Result<()> {
async fn test_updated_at_uses_file_mtime() -> Result<()> {
let temp = TempDir::new().unwrap();
let home = temp.path();
@@ -594,6 +592,7 @@ async fn test_tail_includes_last_response_items() -> Result<()> {
cli_version: "test_version".into(),
source: SessionSource::VSCode,
model_provider: Some("test-provider".into()),
name: None,
},
git: None,
}),
@@ -636,229 +635,16 @@ async fn test_tail_includes_last_response_items() -> Result<()> {
)
.await?;
let item = page.items.first().expect("conversation item");
let tail_len = item.tail.len();
assert_eq!(tail_len, 10usize.min(total_messages));
let expected: Vec<serde_json::Value> = (total_messages - tail_len..total_messages)
.map(|idx| {
serde_json::json!({
"type": "message",
"role": "assistant",
"content": [
{
"type": "output_text",
"text": format!("reply-{idx}"),
}
],
})
})
.collect();
assert_eq!(item.tail, expected);
assert_eq!(item.created_at.as_deref(), Some(ts));
let expected_updated = format!("{ts}-{last:02}", last = total_messages - 1);
assert_eq!(item.updated_at.as_deref(), Some(expected_updated.as_str()));
Ok(())
}
#[tokio::test]
async fn test_tail_handles_short_sessions() -> Result<()> {
let temp = TempDir::new().unwrap();
let home = temp.path();
let ts = "2025-06-02T08-30-00";
let uuid = Uuid::from_u128(7);
let day_dir = home.join("sessions").join("2025").join("06").join("02");
fs::create_dir_all(&day_dir)?;
let file_path = day_dir.join(format!("rollout-{ts}-{uuid}.jsonl"));
let mut file = File::create(&file_path)?;
let conversation_id = ConversationId::from_string(&uuid.to_string())?;
let meta_line = RolloutLine {
timestamp: ts.to_string(),
item: RolloutItem::SessionMeta(SessionMetaLine {
meta: SessionMeta {
id: conversation_id,
timestamp: ts.to_string(),
instructions: None,
cwd: ".".into(),
originator: "test_originator".into(),
cli_version: "test_version".into(),
source: SessionSource::VSCode,
model_provider: Some("test-provider".into()),
},
git: None,
}),
};
writeln!(file, "{}", serde_json::to_string(&meta_line)?)?;
let user_event_line = RolloutLine {
timestamp: ts.to_string(),
item: RolloutItem::EventMsg(EventMsg::UserMessage(UserMessageEvent {
message: "hi".into(),
images: None,
})),
};
writeln!(file, "{}", serde_json::to_string(&user_event_line)?)?;
for idx in 0..3 {
let response_line = RolloutLine {
timestamp: format!("{ts}-{idx:02}"),
item: RolloutItem::ResponseItem(ResponseItem::Message {
id: None,
role: "assistant".into(),
content: vec![ContentItem::OutputText {
text: format!("short-{idx}"),
}],
}),
};
writeln!(file, "{}", serde_json::to_string(&response_line)?)?;
}
drop(file);
let provider_filter = provider_vec(&[TEST_PROVIDER]);
let page = get_conversations(
home,
1,
None,
INTERACTIVE_SESSION_SOURCES,
Some(provider_filter.as_slice()),
TEST_PROVIDER,
)
.await?;
let tail = &page.items.first().expect("conversation item").tail;
assert_eq!(tail.len(), 3);
let expected: Vec<serde_json::Value> = (0..3)
.map(|idx| {
serde_json::json!({
"type": "message",
"role": "assistant",
"content": [
{
"type": "output_text",
"text": format!("short-{idx}"),
}
],
})
})
.collect();
assert_eq!(tail, &expected);
let expected_updated = format!("{ts}-{last:02}", last = 2);
assert_eq!(
page.items[0].updated_at.as_deref(),
Some(expected_updated.as_str())
);
Ok(())
}
#[tokio::test]
async fn test_tail_skips_trailing_non_responses() -> Result<()> {
let temp = TempDir::new().unwrap();
let home = temp.path();
let ts = "2025-06-03T10-00-00";
let uuid = Uuid::from_u128(11);
let day_dir = home.join("sessions").join("2025").join("06").join("03");
fs::create_dir_all(&day_dir)?;
let file_path = day_dir.join(format!("rollout-{ts}-{uuid}.jsonl"));
let mut file = File::create(&file_path)?;
let conversation_id = ConversationId::from_string(&uuid.to_string())?;
let meta_line = RolloutLine {
timestamp: ts.to_string(),
item: RolloutItem::SessionMeta(SessionMetaLine {
meta: SessionMeta {
id: conversation_id,
timestamp: ts.to_string(),
instructions: None,
cwd: ".".into(),
originator: "test_originator".into(),
cli_version: "test_version".into(),
source: SessionSource::VSCode,
model_provider: Some("test-provider".into()),
},
git: None,
}),
};
writeln!(file, "{}", serde_json::to_string(&meta_line)?)?;
let user_event_line = RolloutLine {
timestamp: ts.to_string(),
item: RolloutItem::EventMsg(EventMsg::UserMessage(UserMessageEvent {
message: "hello".into(),
images: None,
})),
};
writeln!(file, "{}", serde_json::to_string(&user_event_line)?)?;
for idx in 0..4 {
let response_line = RolloutLine {
timestamp: format!("{ts}-{idx:02}"),
item: RolloutItem::ResponseItem(ResponseItem::Message {
id: None,
role: "assistant".into(),
content: vec![ContentItem::OutputText {
text: format!("response-{idx}"),
}],
}),
};
writeln!(file, "{}", serde_json::to_string(&response_line)?)?;
}
let compacted_line = RolloutLine {
timestamp: format!("{ts}-compacted"),
item: RolloutItem::Compacted(CompactedItem {
message: "compacted".into(),
replacement_history: None,
}),
};
writeln!(file, "{}", serde_json::to_string(&compacted_line)?)?;
let shutdown_event = RolloutLine {
timestamp: format!("{ts}-shutdown"),
item: RolloutItem::EventMsg(EventMsg::ShutdownComplete),
};
writeln!(file, "{}", serde_json::to_string(&shutdown_event)?)?;
drop(file);
let provider_filter = provider_vec(&[TEST_PROVIDER]);
let page = get_conversations(
home,
1,
None,
INTERACTIVE_SESSION_SOURCES,
Some(provider_filter.as_slice()),
TEST_PROVIDER,
)
.await?;
let tail = &page.items.first().expect("conversation item").tail;
let expected: Vec<serde_json::Value> = (0..4)
.map(|idx| {
serde_json::json!({
"type": "message",
"role": "assistant",
"content": [
{
"type": "output_text",
"text": format!("response-{idx}"),
}
],
})
})
.collect();
assert_eq!(tail, &expected);
let expected_updated = format!("{ts}-{last:02}", last = 3);
assert_eq!(
page.items[0].updated_at.as_deref(),
Some(expected_updated.as_str())
);
let updated = item
.updated_at
.as_deref()
.and_then(|s| chrono::DateTime::parse_from_rfc3339(s).ok())
.map(|dt| dt.with_timezone(&chrono::Utc))
.expect("updated_at set from file mtime");
let now = chrono::Utc::now();
let age = now - updated;
assert!(age.num_seconds().abs() < 30);
Ok(())
}
@@ -913,22 +699,22 @@ async fn test_stable_ordering_same_second_pagination() {
"model_provider": "test-provider",
})]
};
let updated_page1: Vec<Option<String>> =
page1.items.iter().map(|i| i.updated_at.clone()).collect();
let expected_cursor1: Cursor = serde_json::from_str(&format!("\"{ts}|{u2}\"")).unwrap();
let expected_page1 = ConversationsPage {
items: vec![
ConversationItem {
path: p3,
head: head(u3),
tail: Vec::new(),
created_at: Some(ts.to_string()),
updated_at: Some(ts.to_string()),
updated_at: updated_page1.first().cloned().flatten(),
},
ConversationItem {
path: p2,
head: head(u2),
tail: Vec::new(),
created_at: Some(ts.to_string()),
updated_at: Some(ts.to_string()),
updated_at: updated_page1.get(1).cloned().flatten(),
},
],
next_cursor: Some(expected_cursor1.clone()),
@@ -953,13 +739,14 @@ async fn test_stable_ordering_same_second_pagination() {
.join("07")
.join("01")
.join(format!("rollout-2025-07-01T00-00-00-{u1}.jsonl"));
let updated_page2: Vec<Option<String>> =
page2.items.iter().map(|i| i.updated_at.clone()).collect();
let expected_page2 = ConversationsPage {
items: vec![ConversationItem {
path: p1,
head: head(u1),
tail: Vec::new(),
created_at: Some(ts.to_string()),
updated_at: Some(ts.to_string()),
updated_at: updated_page2.first().cloned().flatten(),
}],
next_cursor: None,
num_scanned_files: 3, // scanned u3, u2 (anchor), u1

View File

@@ -6,6 +6,7 @@ use codex_apply_patch::ApplyPatchAction;
use codex_apply_patch::ApplyPatchFileChange;
use crate::exec::SandboxType;
use crate::util::resolve_path;
use crate::protocol::AskForApproval;
use crate::protocol::SandboxPolicy;
@@ -150,11 +151,7 @@ fn is_write_patch_constrained_to_writable_paths(
// and roots are converted to absolute, normalized forms before the
// prefix check.
let is_path_writable = |p: &PathBuf| {
let abs = if p.is_absolute() {
p.clone()
} else {
cwd.join(p)
};
let abs = resolve_path(cwd, p);
let abs = match normalize(&abs) {
Some(v) => v,
None => return false,

View File

@@ -14,6 +14,7 @@ use crate::protocol::SandboxPolicy;
use askama::Template;
use codex_otel::otel_event_manager::OtelEventManager;
use codex_protocol::ConversationId;
use codex_protocol::config_types::ReasoningEffort as ReasoningEffortConfig;
use codex_protocol::models::ContentItem;
use codex_protocol::models::ResponseItem;
use codex_protocol::protocol::SandboxCommandAssessment;
@@ -23,7 +24,8 @@ use serde_json::json;
use tokio::time::timeout;
use tracing::warn;
const SANDBOX_ASSESSMENT_TIMEOUT: Duration = Duration::from_secs(5);
const SANDBOX_ASSESSMENT_TIMEOUT: Duration = Duration::from_secs(15);
const SANDBOX_ASSESSMENT_REASONING_EFFORT: ReasoningEffortConfig = ReasoningEffortConfig::Medium;
#[derive(Template)]
#[template(path = "sandboxing/assessment_prompt.md", escape = "none")]
@@ -130,7 +132,7 @@ pub(crate) async fn assess_command(
Some(auth_manager),
child_otel,
provider,
config.model_reasoning_effort,
Some(SANDBOX_ASSESSMENT_REASONING_EFFORT),
config.model_reasoning_summary,
conversation_id,
session_source,

View File

@@ -0,0 +1,144 @@
use crate::error::Result;
use crate::find_conversation_path_by_id_str;
use crate::rollout::list::read_head_for_summary;
use codex_protocol::ConversationId;
use codex_protocol::protocol::SessionMetaLine;
use serde::Deserialize;
use serde::Serialize;
use std::collections::BTreeMap;
use std::io::Error as IoError;
use std::io::ErrorKind;
use std::path::Path;
use std::path::PathBuf;
use time::OffsetDateTime;
use time::format_description::well_known::Rfc3339;
use tracing::warn;
#[derive(Debug, Clone, Serialize, Deserialize, PartialEq, Eq)]
#[serde(rename_all = "camelCase")]
pub struct SavedSessionEntry {
pub name: String,
pub conversation_id: ConversationId,
pub rollout_path: PathBuf,
pub cwd: PathBuf,
pub model: String,
#[serde(skip_serializing_if = "Option::is_none")]
pub model_provider: Option<String>,
pub saved_at: String,
#[serde(skip_serializing_if = "Option::is_none")]
pub created_at: Option<String>,
}
#[derive(Debug, Default, Serialize, Deserialize)]
#[serde(rename_all = "camelCase")]
struct SavedSessionsFile {
#[serde(default)]
entries: BTreeMap<String, SavedSessionEntry>,
}
fn saved_sessions_path(codex_home: &Path) -> PathBuf {
codex_home.join("saved_sessions.json")
}
async fn load_saved_sessions_file(path: &Path) -> Result<SavedSessionsFile> {
match tokio::fs::read_to_string(path).await {
Ok(text) => {
let parsed = serde_json::from_str(&text)
.map_err(|e| IoError::other(format!("failed to parse saved sessions: {e}")))?;
Ok(parsed)
}
Err(err) if err.kind() == ErrorKind::NotFound => Ok(SavedSessionsFile::default()),
Err(err) => Err(err.into()),
}
}
async fn write_saved_sessions_file(path: &Path, file: &SavedSessionsFile) -> Result<()> {
if let Some(parent) = path.parent() {
tokio::fs::create_dir_all(parent).await?;
}
let json = serde_json::to_string_pretty(file)
.map_err(|e| IoError::other(format!("failed to serialize saved sessions: {e}")))?;
let tmp_path = path.with_extension("json.tmp");
tokio::fs::write(&tmp_path, json).await?;
tokio::fs::rename(tmp_path, path).await?;
Ok(())
}
/// Create a new entry from the rollout's SessionMeta line.
pub async fn build_saved_session_entry(
name: String,
rollout_path: PathBuf,
model: String,
) -> Result<SavedSessionEntry> {
let head = read_head_for_summary(&rollout_path).await?;
let first = head.first().ok_or_else(|| {
IoError::other(format!(
"rollout at {} has no SessionMeta",
rollout_path.display()
))
})?;
let SessionMetaLine { mut meta, .. } = serde_json::from_value::<SessionMetaLine>(first.clone())
.map_err(|e| IoError::other(format!("failed to parse SessionMeta: {e}")))?;
meta.name = Some(name.clone());
let saved_at = OffsetDateTime::now_utc()
.format(&Rfc3339)
.map_err(|e| IoError::other(format!("failed to format timestamp: {e}")))?;
let created_at = if meta.timestamp.is_empty() {
None
} else {
Some(meta.timestamp.clone())
};
Ok(SavedSessionEntry {
name,
conversation_id: meta.id,
rollout_path,
cwd: meta.cwd,
model,
model_provider: meta.model_provider,
saved_at,
created_at,
})
}
/// Insert or replace a saved session entry in `saved_sessions.json`.
pub async fn upsert_saved_session(codex_home: &Path, entry: SavedSessionEntry) -> Result<()> {
let path = saved_sessions_path(codex_home);
let mut file = load_saved_sessions_file(&path).await?;
file.entries.insert(entry.name.clone(), entry);
write_saved_sessions_file(&path, &file).await
}
/// Lookup a saved session by name, if present.
pub async fn resolve_saved_session(
codex_home: &Path,
name: &str,
) -> Result<Option<SavedSessionEntry>> {
let path = saved_sessions_path(codex_home);
let file = load_saved_sessions_file(&path).await?;
Ok(file.entries.get(name).cloned())
}
/// Return all saved sessions ordered by newest `saved_at` first.
pub async fn list_saved_sessions(codex_home: &Path) -> Result<Vec<SavedSessionEntry>> {
let path = saved_sessions_path(codex_home);
let file = load_saved_sessions_file(&path).await?;
let mut entries: Vec<SavedSessionEntry> = file.entries.values().cloned().collect();
entries.sort_by(|a, b| b.saved_at.cmp(&a.saved_at));
Ok(entries)
}
/// Resolve a rollout path from either a saved-session name or rollout id string.
/// Returns `Ok(None)` when nothing matches.
pub async fn resolve_rollout_path(codex_home: &Path, identifier: &str) -> Result<Option<PathBuf>> {
if let Some(entry) = resolve_saved_session(codex_home, identifier).await? {
if entry.rollout_path.exists() {
return Ok(Some(entry.rollout_path));
}
warn!(
"saved session '{}' points to missing rollout at {}",
identifier,
entry.rollout_path.display()
);
}
Ok(find_conversation_path_by_id_str(codex_home, identifier).await?)
}

View File

@@ -0,0 +1,291 @@
use crate::config::Config;
use crate::skills::model::SkillError;
use crate::skills::model::SkillLoadOutcome;
use crate::skills::model::SkillMetadata;
use dunce::canonicalize as normalize_path;
use serde::Deserialize;
use std::collections::VecDeque;
use std::error::Error;
use std::fmt;
use std::fs;
use std::path::Path;
use std::path::PathBuf;
use tracing::error;
#[derive(Debug, Deserialize)]
struct SkillFrontmatter {
name: String,
description: String,
}
const SKILLS_FILENAME: &str = "SKILL.md";
const SKILLS_DIR_NAME: &str = "skills";
const MAX_NAME_LEN: usize = 100;
const MAX_DESCRIPTION_LEN: usize = 500;
#[derive(Debug)]
enum SkillParseError {
Read(std::io::Error),
MissingFrontmatter,
InvalidYaml(serde_yaml::Error),
MissingField(&'static str),
InvalidField { field: &'static str, reason: String },
}
impl fmt::Display for SkillParseError {
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
match self {
SkillParseError::Read(e) => write!(f, "failed to read file: {e}"),
SkillParseError::MissingFrontmatter => {
write!(f, "missing YAML frontmatter delimited by ---")
}
SkillParseError::InvalidYaml(e) => write!(f, "invalid YAML: {e}"),
SkillParseError::MissingField(field) => write!(f, "missing field `{field}`"),
SkillParseError::InvalidField { field, reason } => {
write!(f, "invalid {field}: {reason}")
}
}
}
}
impl Error for SkillParseError {}
pub fn load_skills(config: &Config) -> SkillLoadOutcome {
let mut outcome = SkillLoadOutcome::default();
let roots = skill_roots(config);
for root in roots {
discover_skills_under_root(&root, &mut outcome);
}
outcome
.skills
.sort_by(|a, b| a.name.cmp(&b.name).then_with(|| a.path.cmp(&b.path)));
outcome
}
fn skill_roots(config: &Config) -> Vec<PathBuf> {
vec![config.codex_home.join(SKILLS_DIR_NAME)]
}
fn discover_skills_under_root(root: &Path, outcome: &mut SkillLoadOutcome) {
let Ok(root) = normalize_path(root) else {
return;
};
if !root.is_dir() {
return;
}
let mut queue: VecDeque<PathBuf> = VecDeque::from([root]);
while let Some(dir) = queue.pop_front() {
let entries = match fs::read_dir(&dir) {
Ok(entries) => entries,
Err(e) => {
error!("failed to read skills dir {}: {e:#}", dir.display());
continue;
}
};
for entry in entries.flatten() {
let path = entry.path();
let file_name = match path.file_name().and_then(|f| f.to_str()) {
Some(name) => name,
None => continue,
};
if file_name.starts_with('.') {
continue;
}
let Ok(file_type) = entry.file_type() else {
continue;
};
if file_type.is_symlink() {
continue;
}
if file_type.is_dir() {
queue.push_back(path);
continue;
}
if file_type.is_file() && file_name == SKILLS_FILENAME {
match parse_skill_file(&path) {
Ok(skill) => outcome.skills.push(skill),
Err(err) => outcome.errors.push(SkillError {
path,
message: err.to_string(),
}),
}
}
}
}
}
fn parse_skill_file(path: &Path) -> Result<SkillMetadata, SkillParseError> {
let contents = fs::read_to_string(path).map_err(SkillParseError::Read)?;
let frontmatter = extract_frontmatter(&contents).ok_or(SkillParseError::MissingFrontmatter)?;
let parsed: SkillFrontmatter =
serde_yaml::from_str(&frontmatter).map_err(SkillParseError::InvalidYaml)?;
let name = sanitize_single_line(&parsed.name);
let description = sanitize_single_line(&parsed.description);
validate_field(&name, MAX_NAME_LEN, "name")?;
validate_field(&description, MAX_DESCRIPTION_LEN, "description")?;
let resolved_path = normalize_path(path).unwrap_or_else(|_| path.to_path_buf());
Ok(SkillMetadata {
name,
description,
path: resolved_path,
})
}
fn sanitize_single_line(raw: &str) -> String {
raw.split_whitespace().collect::<Vec<_>>().join(" ")
}
fn validate_field(
value: &str,
max_len: usize,
field_name: &'static str,
) -> Result<(), SkillParseError> {
if value.is_empty() {
return Err(SkillParseError::MissingField(field_name));
}
if value.len() > max_len {
return Err(SkillParseError::InvalidField {
field: field_name,
reason: format!("exceeds maximum length of {max_len} characters"),
});
}
Ok(())
}
fn extract_frontmatter(contents: &str) -> Option<String> {
let mut lines = contents.lines();
if !matches!(lines.next(), Some(line) if line.trim() == "---") {
return None;
}
let mut frontmatter_lines: Vec<&str> = Vec::new();
let mut found_closing = false;
for line in lines.by_ref() {
if line.trim() == "---" {
found_closing = true;
break;
}
frontmatter_lines.push(line);
}
if frontmatter_lines.is_empty() || !found_closing {
return None;
}
Some(frontmatter_lines.join("\n"))
}
#[cfg(test)]
mod tests {
use super::*;
use crate::config::ConfigOverrides;
use crate::config::ConfigToml;
use tempfile::TempDir;
fn make_config(codex_home: &TempDir) -> Config {
let mut config = Config::load_from_base_config_with_overrides(
ConfigToml::default(),
ConfigOverrides::default(),
codex_home.path().to_path_buf(),
)
.expect("defaults for test should always succeed");
config.cwd = codex_home.path().to_path_buf();
config
}
fn write_skill(codex_home: &TempDir, dir: &str, name: &str, description: &str) -> PathBuf {
let skill_dir = codex_home.path().join(format!("skills/{dir}"));
fs::create_dir_all(&skill_dir).unwrap();
let indented_description = description.replace('\n', "\n ");
let content = format!(
"---\nname: {name}\ndescription: |-\n {indented_description}\n---\n\n# Body\n"
);
let path = skill_dir.join(SKILLS_FILENAME);
fs::write(&path, content).unwrap();
path
}
#[test]
fn loads_valid_skill() {
let codex_home = tempfile::tempdir().expect("tempdir");
write_skill(&codex_home, "demo", "demo-skill", "does things\ncarefully");
let cfg = make_config(&codex_home);
let outcome = load_skills(&cfg);
assert!(
outcome.errors.is_empty(),
"unexpected errors: {:?}",
outcome.errors
);
assert_eq!(outcome.skills.len(), 1);
let skill = &outcome.skills[0];
assert_eq!(skill.name, "demo-skill");
assert_eq!(skill.description, "does things carefully");
let path_str = skill.path.to_string_lossy().replace('\\', "/");
assert!(
path_str.ends_with("skills/demo/SKILL.md"),
"unexpected path {path_str}"
);
}
#[test]
fn skips_hidden_and_invalid() {
let codex_home = tempfile::tempdir().expect("tempdir");
let hidden_dir = codex_home.path().join("skills/.hidden");
fs::create_dir_all(&hidden_dir).unwrap();
fs::write(
hidden_dir.join(SKILLS_FILENAME),
"---\nname: hidden\ndescription: hidden\n---\n",
)
.unwrap();
// Invalid because missing closing frontmatter.
let invalid_dir = codex_home.path().join("skills/invalid");
fs::create_dir_all(&invalid_dir).unwrap();
fs::write(invalid_dir.join(SKILLS_FILENAME), "---\nname: bad").unwrap();
let cfg = make_config(&codex_home);
let outcome = load_skills(&cfg);
assert_eq!(outcome.skills.len(), 0);
assert_eq!(outcome.errors.len(), 1);
assert!(
outcome.errors[0]
.message
.contains("missing YAML frontmatter"),
"expected frontmatter error"
);
}
#[test]
fn enforces_length_limits() {
let codex_home = tempfile::tempdir().expect("tempdir");
let long_desc = "a".repeat(MAX_DESCRIPTION_LEN + 1);
write_skill(&codex_home, "too-long", "toolong", &long_desc);
let cfg = make_config(&codex_home);
let outcome = load_skills(&cfg);
assert_eq!(outcome.skills.len(), 0);
assert_eq!(outcome.errors.len(), 1);
assert!(
outcome.errors[0].message.contains("invalid description"),
"expected length error"
);
}
}

View File

@@ -0,0 +1,9 @@
pub mod loader;
pub mod model;
pub mod render;
pub use loader::load_skills;
pub use model::SkillError;
pub use model::SkillLoadOutcome;
pub use model::SkillMetadata;
pub use render::render_skills_section;

View File

@@ -0,0 +1,20 @@
use std::path::PathBuf;
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct SkillMetadata {
pub name: String,
pub description: String,
pub path: PathBuf,
}
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct SkillError {
pub path: PathBuf,
pub message: String,
}
#[derive(Debug, Clone, Default)]
pub struct SkillLoadOutcome {
pub skills: Vec<SkillMetadata>,
pub errors: Vec<SkillError>,
}

View File

@@ -0,0 +1,21 @@
use crate::skills::model::SkillMetadata;
pub fn render_skills_section(skills: &[SkillMetadata]) -> Option<String> {
if skills.is_empty() {
return None;
}
let mut lines: Vec<String> = Vec::new();
lines.push("## Skills".to_string());
lines.push("These skills are discovered at startup from ~/.codex/skills; each entry shows name, description, and file path so you can open the source for full instructions. Content is not inlined to keep context lean.".to_string());
for skill in skills {
let path_str = skill.path.to_string_lossy().replace('\\', "/");
lines.push(format!(
"- {}: {} (file: {})",
skill.name, skill.description, path_str
));
}
Some(lines.join("\n"))
}

View File

@@ -62,7 +62,10 @@ impl SessionState {
}
pub(crate) fn set_rate_limits(&mut self, snapshot: RateLimitSnapshot) {
self.latest_rate_limits = Some(snapshot);
self.latest_rate_limits = Some(merge_rate_limit_credits(
self.latest_rate_limits.as_ref(),
snapshot,
));
}
pub(crate) fn token_info_and_rate_limits(
@@ -79,3 +82,14 @@ impl SessionState {
self.history.get_total_token_usage()
}
}
// Sometimes new snapshots don't include credits
fn merge_rate_limit_credits(
previous: Option<&RateLimitSnapshot>,
mut snapshot: RateLimitSnapshot,
) -> RateLimitSnapshot {
if snapshot.credits.is_none() {
snapshot.credits = previous.and_then(|prior| prior.credits.clone());
}
snapshot
}

View File

@@ -128,7 +128,9 @@ impl Session {
task_cancellation_token.child_token(),
)
.await;
session_ctx.clone_session().flush_rollout().await;
if let Err(e) = session_ctx.clone_session().flush_rollout().await {
tracing::warn!("failed to flush rollout recorder: {e}");
}
if !task_cancellation_token.is_cancelled() {
// Emit completion uniformly from spawn site so all tasks share the same lifecycle.
let sess = session_ctx.clone_session();

View File

@@ -17,6 +17,7 @@ use crate::codex::Session;
use crate::codex::TurnContext;
use crate::codex_delegate::run_codex_conversation_one_shot;
use crate::review_format::format_review_findings_block;
use crate::review_format::render_review_output_text;
use crate::state::TaskKind;
use codex_protocol::user_input::UserInput;
@@ -24,15 +25,11 @@ use super::SessionTask;
use super::SessionTaskContext;
#[derive(Clone, Copy)]
pub(crate) struct ReviewTask {
append_to_original_thread: bool,
}
pub(crate) struct ReviewTask;
impl ReviewTask {
pub(crate) fn new(append_to_original_thread: bool) -> Self {
Self {
append_to_original_thread,
}
pub(crate) fn new() -> Self {
Self
}
}
@@ -62,25 +59,13 @@ impl SessionTask for ReviewTask {
None => None,
};
if !cancellation_token.is_cancelled() {
exit_review_mode(
session.clone_session(),
output.clone(),
ctx.clone(),
self.append_to_original_thread,
)
.await;
exit_review_mode(session.clone_session(), output.clone(), ctx.clone()).await;
}
None
}
async fn abort(&self, session: Arc<SessionTaskContext>, ctx: Arc<TurnContext>) {
exit_review_mode(
session.clone_session(),
None,
ctx,
self.append_to_original_thread,
)
.await;
exit_review_mode(session.clone_session(), None, ctx).await;
}
}
@@ -197,39 +182,57 @@ pub(crate) async fn exit_review_mode(
session: Arc<Session>,
review_output: Option<ReviewOutputEvent>,
ctx: Arc<TurnContext>,
append_to_original_thread: bool,
) {
if append_to_original_thread {
let user_message = if let Some(out) = review_output.clone() {
let mut findings_str = String::new();
let text = out.overall_explanation.trim();
if !text.is_empty() {
findings_str.push_str(text);
}
if !out.findings.is_empty() {
let block = format_review_findings_block(&out.findings, None);
findings_str.push_str(&format!("\n{block}"));
}
crate::client_common::REVIEW_EXIT_SUCCESS_TMPL.replace("{results}", &findings_str)
} else {
crate::client_common::REVIEW_EXIT_INTERRUPTED_TMPL.to_string()
};
const REVIEW_USER_MESSAGE_ID: &str = "review:rollout:user";
const REVIEW_ASSISTANT_MESSAGE_ID: &str = "review:rollout:assistant";
let (user_message, assistant_message) = if let Some(out) = review_output.clone() {
let mut findings_str = String::new();
let text = out.overall_explanation.trim();
if !text.is_empty() {
findings_str.push_str(text);
}
if !out.findings.is_empty() {
let block = format_review_findings_block(&out.findings, None);
findings_str.push_str(&format!("\n{block}"));
}
let rendered =
crate::client_common::REVIEW_EXIT_SUCCESS_TMPL.replace("{results}", &findings_str);
let assistant_message = render_review_output_text(&out);
(rendered, assistant_message)
} else {
let rendered = crate::client_common::REVIEW_EXIT_INTERRUPTED_TMPL.to_string();
let assistant_message =
"Review was interrupted. Please re-run /review and wait for it to complete."
.to_string();
(rendered, assistant_message)
};
session
.record_conversation_items(
&ctx,
&[ResponseItem::Message {
id: None,
role: "user".to_string(),
content: vec![ContentItem::InputText { text: user_message }],
}],
)
.await;
}
session
.record_conversation_items(
&ctx,
&[ResponseItem::Message {
id: Some(REVIEW_USER_MESSAGE_ID.to_string()),
role: "user".to_string(),
content: vec![ContentItem::InputText { text: user_message }],
}],
)
.await;
session
.send_event(
ctx.as_ref(),
EventMsg::ExitedReviewMode(ExitedReviewModeEvent { review_output }),
)
.await;
session
.record_response_item_and_emit_turn_item(
ctx.as_ref(),
ResponseItem::Message {
id: Some(REVIEW_ASSISTANT_MESSAGE_ID.to_string()),
role: "assistant".to_string(),
content: vec![ContentItem::OutputText {
text: assistant_message,
}],
},
)
.await;
}

View File

@@ -81,6 +81,7 @@ impl SessionTask for UserShellCommandTask {
turn_context.as_ref(),
EventMsg::ExecCommandBegin(ExecCommandBeginEvent {
call_id: call_id.clone(),
process_id: None,
turn_id: turn_context.sub_id.clone(),
command: command.clone(),
cwd: cwd.clone(),
@@ -139,6 +140,7 @@ impl SessionTask for UserShellCommandTask {
turn_context.as_ref(),
EventMsg::ExecCommandEnd(ExecCommandEndEvent {
call_id,
process_id: None,
turn_id: turn_context.sub_id.clone(),
command: command.clone(),
cwd: cwd.clone(),
@@ -161,6 +163,7 @@ impl SessionTask for UserShellCommandTask {
turn_context.as_ref(),
EventMsg::ExecCommandEnd(ExecCommandEndEvent {
call_id: call_id.clone(),
process_id: None,
turn_id: turn_context.sub_id.clone(),
command: command.clone(),
cwd: cwd.clone(),
@@ -205,6 +208,7 @@ impl SessionTask for UserShellCommandTask {
turn_context.as_ref(),
EventMsg::ExecCommandEnd(ExecCommandEndEvent {
call_id,
process_id: None,
turn_id: turn_context.sub_id.clone(),
command,
cwd,

View File

@@ -65,12 +65,14 @@ pub(crate) async fn emit_exec_command_begin(
parsed_cmd: &[ParsedCommand],
source: ExecCommandSource,
interaction_input: Option<String>,
process_id: Option<&str>,
) {
ctx.session
.send_event(
ctx.turn,
EventMsg::ExecCommandBegin(ExecCommandBeginEvent {
call_id: ctx.call_id.to_string(),
process_id: process_id.map(str::to_owned),
turn_id: ctx.turn.sub_id.clone(),
command: command.to_vec(),
cwd: cwd.to_path_buf(),
@@ -100,6 +102,7 @@ pub(crate) enum ToolEmitter {
source: ExecCommandSource,
interaction_input: Option<String>,
parsed_cmd: Vec<ParsedCommand>,
process_id: Option<String>,
},
}
@@ -132,6 +135,7 @@ impl ToolEmitter {
cwd: PathBuf,
source: ExecCommandSource,
interaction_input: Option<String>,
process_id: Option<String>,
) -> Self {
let parsed_cmd = parse_command(command);
Self::UnifiedExec {
@@ -140,6 +144,7 @@ impl ToolEmitter {
source,
interaction_input,
parsed_cmd,
process_id,
}
}
@@ -157,7 +162,7 @@ impl ToolEmitter {
) => {
emit_exec_stage(
ctx,
ExecCommandInput::new(command, cwd.as_path(), parsed_cmd, *source, None),
ExecCommandInput::new(command, cwd.as_path(), parsed_cmd, *source, None, None),
stage,
)
.await;
@@ -229,6 +234,7 @@ impl ToolEmitter {
source,
interaction_input,
parsed_cmd,
process_id,
},
stage,
) => {
@@ -240,6 +246,7 @@ impl ToolEmitter {
parsed_cmd,
*source,
interaction_input.as_deref(),
process_id.as_deref(),
),
stage,
)
@@ -319,6 +326,7 @@ struct ExecCommandInput<'a> {
parsed_cmd: &'a [ParsedCommand],
source: ExecCommandSource,
interaction_input: Option<&'a str>,
process_id: Option<&'a str>,
}
impl<'a> ExecCommandInput<'a> {
@@ -328,6 +336,7 @@ impl<'a> ExecCommandInput<'a> {
parsed_cmd: &'a [ParsedCommand],
source: ExecCommandSource,
interaction_input: Option<&'a str>,
process_id: Option<&'a str>,
) -> Self {
Self {
command,
@@ -335,6 +344,7 @@ impl<'a> ExecCommandInput<'a> {
parsed_cmd,
source,
interaction_input,
process_id,
}
}
}
@@ -362,6 +372,7 @@ async fn emit_exec_stage(
exec_input.parsed_cmd,
exec_input.source,
exec_input.interaction_input.map(str::to_owned),
exec_input.process_id,
)
.await;
}
@@ -402,6 +413,7 @@ async fn emit_exec_end(
ctx.turn,
EventMsg::ExecCommandEnd(ExecCommandEndEvent {
call_id: ctx.call_id.to_string(),
process_id: exec_input.process_id.map(str::to_owned),
turn_id: ctx.turn.sub_id.clone(),
command: exec_input.command.to_vec(),
cwd: exec_input.cwd.to_path_buf(),

View File

@@ -1,4 +1,5 @@
use std::collections::BTreeMap;
use std::path::Path;
use crate::apply_patch;
use crate::apply_patch::InternalApplyPatchInvocation;
@@ -7,7 +8,10 @@ use crate::client_common::tools::FreeformTool;
use crate::client_common::tools::FreeformToolFormat;
use crate::client_common::tools::ResponsesApiTool;
use crate::client_common::tools::ToolSpec;
use crate::codex::Session;
use crate::codex::TurnContext;
use crate::function_tool::FunctionCallError;
use crate::tools::context::SharedTurnDiffTracker;
use crate::tools::context::ToolInvocation;
use crate::tools::context::ToolOutput;
use crate::tools::context::ToolPayload;
@@ -164,6 +168,86 @@ pub enum ApplyPatchToolType {
Function,
}
#[allow(clippy::too_many_arguments)]
pub(crate) async fn intercept_apply_patch(
command: &[String],
cwd: &Path,
timeout_ms: Option<u64>,
session: &Session,
turn: &TurnContext,
tracker: Option<&SharedTurnDiffTracker>,
call_id: &str,
tool_name: &str,
) -> Result<Option<ToolOutput>, FunctionCallError> {
match codex_apply_patch::maybe_parse_apply_patch_verified(command, cwd) {
codex_apply_patch::MaybeApplyPatchVerified::Body(changes) => {
session
.record_model_warning(
format!("apply_patch was requested via {tool_name}. Use the apply_patch tool instead of exec_command."),
turn,
)
.await;
match apply_patch::apply_patch(session, turn, call_id, changes).await {
InternalApplyPatchInvocation::Output(item) => {
let content = item?;
Ok(Some(ToolOutput::Function {
content,
content_items: None,
success: Some(true),
}))
}
InternalApplyPatchInvocation::DelegateToExec(apply) => {
let emitter = ToolEmitter::apply_patch(
convert_apply_patch_to_protocol(&apply.action),
!apply.user_explicitly_approved_this_action,
);
let event_ctx =
ToolEventCtx::new(session, turn, call_id, tracker.as_ref().copied());
emitter.begin(event_ctx).await;
let req = ApplyPatchRequest {
patch: apply.action.patch.clone(),
cwd: apply.action.cwd.clone(),
timeout_ms,
user_explicitly_approved: apply.user_explicitly_approved_this_action,
codex_exe: turn.codex_linux_sandbox_exe.clone(),
};
let mut orchestrator = ToolOrchestrator::new();
let mut runtime = ApplyPatchRuntime::new();
let tool_ctx = ToolCtx {
session,
turn,
call_id: call_id.to_string(),
tool_name: tool_name.to_string(),
};
let out = orchestrator
.run(&mut runtime, &req, &tool_ctx, turn, turn.approval_policy)
.await;
let event_ctx =
ToolEventCtx::new(session, turn, call_id, tracker.as_ref().copied());
let content = emitter.finish(event_ctx, out).await?;
Ok(Some(ToolOutput::Function {
content,
content_items: None,
success: Some(true),
}))
}
}
}
codex_apply_patch::MaybeApplyPatchVerified::CorrectnessError(parse_error) => {
Err(FunctionCallError::RespondToModel(format!(
"apply_patch verification failed: {parse_error}"
)))
}
codex_apply_patch::MaybeApplyPatchVerified::ShellParseError(error) => {
tracing::trace!("Failed to parse apply_patch input, {error:?}");
Ok(None)
}
codex_apply_patch::MaybeApplyPatchVerified::NotApplyPatch => Ok(None),
}
}
/// Returns a custom tool that can be used to edit files. Well-suited for GPT-5 models
/// https://platform.openai.com/docs/guides/function-calling#custom-tools
pub(crate) fn create_apply_patch_freeform_tool() -> ToolSpec {

View File

@@ -3,9 +3,6 @@ use codex_protocol::models::ShellCommandToolCallParams;
use codex_protocol::models::ShellToolCallParams;
use std::sync::Arc;
use crate::apply_patch;
use crate::apply_patch::InternalApplyPatchInvocation;
use crate::apply_patch::convert_apply_patch_to_protocol;
use crate::codex::TurnContext;
use crate::exec::ExecParams;
use crate::exec_env::create_env;
@@ -19,11 +16,10 @@ use crate::tools::context::ToolOutput;
use crate::tools::context::ToolPayload;
use crate::tools::events::ToolEmitter;
use crate::tools::events::ToolEventCtx;
use crate::tools::handlers::apply_patch::intercept_apply_patch;
use crate::tools::orchestrator::ToolOrchestrator;
use crate::tools::registry::ToolHandler;
use crate::tools::registry::ToolKind;
use crate::tools::runtimes::apply_patch::ApplyPatchRequest;
use crate::tools::runtimes::apply_patch::ApplyPatchRuntime;
use crate::tools::runtimes::shell::ShellRequest;
use crate::tools::runtimes::shell::ShellRuntime;
use crate::tools::sandboxing::ToolCtx;
@@ -210,81 +206,19 @@ impl ShellHandler {
}
// Intercept apply_patch if present.
match codex_apply_patch::maybe_parse_apply_patch_verified(
if let Some(output) = intercept_apply_patch(
&exec_params.command,
&exec_params.cwd,
) {
codex_apply_patch::MaybeApplyPatchVerified::Body(changes) => {
match apply_patch::apply_patch(session.as_ref(), turn.as_ref(), &call_id, changes)
.await
{
InternalApplyPatchInvocation::Output(item) => {
// Programmatic apply_patch path; return its result.
let content = item?;
return Ok(ToolOutput::Function {
content,
content_items: None,
success: Some(true),
});
}
InternalApplyPatchInvocation::DelegateToExec(apply) => {
let emitter = ToolEmitter::apply_patch(
convert_apply_patch_to_protocol(&apply.action),
!apply.user_explicitly_approved_this_action,
);
let event_ctx = ToolEventCtx::new(
session.as_ref(),
turn.as_ref(),
&call_id,
Some(&tracker),
);
emitter.begin(event_ctx).await;
let req = ApplyPatchRequest {
patch: apply.action.patch.clone(),
cwd: apply.action.cwd.clone(),
timeout_ms: exec_params.expiration.timeout_ms(),
user_explicitly_approved: apply.user_explicitly_approved_this_action,
codex_exe: turn.codex_linux_sandbox_exe.clone(),
};
let mut orchestrator = ToolOrchestrator::new();
let mut runtime = ApplyPatchRuntime::new();
let tool_ctx = ToolCtx {
session: session.as_ref(),
turn: turn.as_ref(),
call_id: call_id.clone(),
tool_name: tool_name.to_string(),
};
let out = orchestrator
.run(&mut runtime, &req, &tool_ctx, &turn, turn.approval_policy)
.await;
let event_ctx = ToolEventCtx::new(
session.as_ref(),
turn.as_ref(),
&call_id,
Some(&tracker),
);
let content = emitter.finish(event_ctx, out).await?;
return Ok(ToolOutput::Function {
content,
content_items: None,
success: Some(true),
});
}
}
}
codex_apply_patch::MaybeApplyPatchVerified::CorrectnessError(parse_error) => {
return Err(FunctionCallError::RespondToModel(format!(
"apply_patch verification failed: {parse_error}"
)));
}
codex_apply_patch::MaybeApplyPatchVerified::ShellParseError(error) => {
tracing::trace!("Failed to parse shell command, {error:?}");
// Fall through to regular shell execution.
}
codex_apply_patch::MaybeApplyPatchVerified::NotApplyPatch => {
// Fall through to regular shell execution.
}
exec_params.expiration.timeout_ms(),
session.as_ref(),
turn.as_ref(),
Some(&tracker),
&call_id,
tool_name,
)
.await?
{
return Ok(output);
}
let source = ExecCommandSource::Agent;
@@ -297,6 +231,15 @@ impl ShellHandler {
let event_ctx = ToolEventCtx::new(session.as_ref(), turn.as_ref(), &call_id, None);
emitter.begin(event_ctx).await;
let approval_requirement = create_approval_requirement_for_command(
&turn.exec_policy,
&exec_params.command,
turn.approval_policy,
&turn.sandbox_policy,
SandboxPermissions::from(exec_params.with_escalated_permissions.unwrap_or(false)),
)
.await;
let req = ShellRequest {
command: exec_params.command.clone(),
cwd: exec_params.cwd.clone(),
@@ -304,13 +247,7 @@ impl ShellHandler {
env: exec_params.env.clone(),
with_escalated_permissions: exec_params.with_escalated_permissions,
justification: exec_params.justification.clone(),
approval_requirement: create_approval_requirement_for_command(
&turn.exec_policy,
&exec_params.command,
turn.approval_policy,
&turn.sandbox_policy,
SandboxPermissions::from(exec_params.with_escalated_permissions.unwrap_or(false)),
),
approval_requirement,
};
let mut orchestrator = ToolOrchestrator::new();
let mut runtime = ShellRuntime::new();

View File

@@ -6,6 +6,7 @@ use crate::protocol::EventMsg;
use crate::protocol::ExecCommandOutputDeltaEvent;
use crate::protocol::ExecCommandSource;
use crate::protocol::ExecOutputStream;
use crate::shell::default_user_shell;
use crate::shell::get_shell_by_model_provided_path;
use crate::tools::context::ToolInvocation;
use crate::tools::context::ToolOutput;
@@ -13,6 +14,7 @@ use crate::tools::context::ToolPayload;
use crate::tools::events::ToolEmitter;
use crate::tools::events::ToolEventCtx;
use crate::tools::events::ToolEventStage;
use crate::tools::handlers::apply_patch::intercept_apply_patch;
use crate::tools::registry::ToolHandler;
use crate::tools::registry::ToolKind;
use crate::unified_exec::ExecCommandRequest;
@@ -30,8 +32,8 @@ struct ExecCommandArgs {
cmd: String,
#[serde(default)]
workdir: Option<String>,
#[serde(default = "default_shell")]
shell: String,
#[serde(default)]
shell: Option<String>,
#[serde(default = "default_login")]
login: bool,
#[serde(default = "default_exec_yield_time_ms")]
@@ -46,6 +48,7 @@ struct ExecCommandArgs {
#[derive(Debug, Deserialize)]
struct WriteStdinArgs {
// The model is trained on `session_id`.
session_id: i32,
#[serde(default)]
chars: String,
@@ -63,10 +66,6 @@ fn default_write_stdin_yield_time_ms() -> u64 {
250
}
fn default_shell() -> String {
"/bin/bash".to_string()
}
fn default_login() -> bool {
true
}
@@ -102,6 +101,7 @@ impl ToolHandler for UnifiedExecHandler {
let ToolInvocation {
session,
turn,
tracker,
call_id,
tool_name,
payload,
@@ -128,6 +128,7 @@ impl ToolHandler for UnifiedExecHandler {
"failed to parse exec_command arguments: {err:?}"
))
})?;
let process_id = manager.allocate_process_id().await;
let command = get_command(&args);
let ExecCommandArgs {
@@ -151,12 +152,26 @@ impl ToolHandler for UnifiedExecHandler {
)));
}
let workdir = workdir
.as_deref()
.filter(|value| !value.is_empty())
.map(PathBuf::from);
let workdir = workdir.filter(|value| !value.is_empty());
let workdir = workdir.map(|dir| context.turn.resolve_path(Some(dir)));
let cwd = workdir.clone().unwrap_or_else(|| context.turn.cwd.clone());
if let Some(output) = intercept_apply_patch(
&command,
&cwd,
Some(yield_time_ms),
context.session.as_ref(),
context.turn.as_ref(),
Some(&tracker),
&context.call_id,
tool_name.as_str(),
)
.await?
{
return Ok(output);
}
let event_ctx = ToolEventCtx::new(
context.session.as_ref(),
context.turn.as_ref(),
@@ -168,6 +183,7 @@ impl ToolHandler for UnifiedExecHandler {
cwd.clone(),
ExecCommandSource::UnifiedExecStartup,
None,
Some(process_id.clone()),
);
emitter.emit(event_ctx, ToolEventStage::Begin).await;
@@ -175,6 +191,7 @@ impl ToolHandler for UnifiedExecHandler {
.exec_command(
ExecCommandRequest {
command,
process_id,
yield_time_ms,
max_output_tokens,
workdir,
@@ -197,7 +214,7 @@ impl ToolHandler for UnifiedExecHandler {
manager
.write_stdin(WriteStdinRequest {
call_id: &call_id,
session_id: args.session_id,
process_id: &args.session_id.to_string(),
input: &args.chars,
yield_time_ms: args.yield_time_ms,
max_output_tokens: args.max_output_tokens,
@@ -237,7 +254,12 @@ impl ToolHandler for UnifiedExecHandler {
}
fn get_command(args: &ExecCommandArgs) -> Vec<String> {
let shell = get_shell_by_model_provided_path(&PathBuf::from(args.shell.clone()));
let shell = if let Some(shell_str) = &args.shell {
get_shell_by_model_provided_path(&PathBuf::from(shell_str))
} else {
default_user_shell()
};
shell.derive_exec_args(&args.cmd, args.login)
}
@@ -255,8 +277,9 @@ fn format_response(response: &UnifiedExecResponse) -> String {
sections.push(format!("Process exited with code {exit_code}"));
}
if let Some(session_id) = response.session_id {
sections.push(format!("Process running with session ID {session_id}"));
if let Some(process_id) = &response.process_id {
// Training still uses "session ID".
sections.push(format!("Process running with session ID {process_id}"));
}
if let Some(original_token_count) = response.original_token_count {
@@ -268,3 +291,65 @@ fn format_response(response: &UnifiedExecResponse) -> String {
sections.join("\n")
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
fn test_get_command_uses_default_shell_when_unspecified() {
let json = r#"{"cmd": "echo hello"}"#;
let args: ExecCommandArgs =
serde_json::from_str(json).expect("deserialize ExecCommandArgs");
assert!(args.shell.is_none());
let command = get_command(&args);
assert_eq!(command.len(), 3);
assert_eq!(command[2], "echo hello");
}
#[test]
fn test_get_command_respects_explicit_bash_shell() {
let json = r#"{"cmd": "echo hello", "shell": "/bin/bash"}"#;
let args: ExecCommandArgs =
serde_json::from_str(json).expect("deserialize ExecCommandArgs");
assert_eq!(args.shell.as_deref(), Some("/bin/bash"));
let command = get_command(&args);
assert_eq!(command[2], "echo hello");
}
#[test]
fn test_get_command_respects_explicit_powershell_shell() {
let json = r#"{"cmd": "echo hello", "shell": "powershell"}"#;
let args: ExecCommandArgs =
serde_json::from_str(json).expect("deserialize ExecCommandArgs");
assert_eq!(args.shell.as_deref(), Some("powershell"));
let command = get_command(&args);
assert_eq!(command[2], "echo hello");
}
#[test]
fn test_get_command_respects_explicit_cmd_shell() {
let json = r#"{"cmd": "echo hello", "shell": "cmd"}"#;
let args: ExecCommandArgs =
serde_json::from_str(json).expect("deserialize ExecCommandArgs");
assert_eq!(args.shell.as_deref(), Some("cmd"));
let command = get_command(&args);
assert_eq!(command[2], "echo hello");
}
}

View File

@@ -5,8 +5,9 @@ use thiserror::Error;
pub(crate) enum UnifiedExecError {
#[error("Failed to create unified exec session: {message}")]
CreateSession { message: String },
#[error("Unknown session id {session_id}")]
UnknownSessionId { session_id: i32 },
// Called "session" in the model's training.
#[error("Unknown session id {process_id}")]
UnknownSessionId { process_id: String },
#[error("failed to write to stdin")]
WriteToStdin,
#[error("missing command line for unified exec request")]

View File

@@ -22,9 +22,9 @@
//! - `session_manager.rs`: orchestration (approvals, sandboxing, reuse) and request handling.
use std::collections::HashMap;
use std::collections::HashSet;
use std::path::PathBuf;
use std::sync::Arc;
use std::sync::atomic::AtomicI32;
use std::time::Duration;
use rand::Rng;
@@ -48,6 +48,9 @@ pub(crate) const UNIFIED_EXEC_OUTPUT_MAX_BYTES: usize = 1024 * 1024; // 1 MiB
pub(crate) const UNIFIED_EXEC_OUTPUT_MAX_TOKENS: usize = UNIFIED_EXEC_OUTPUT_MAX_BYTES / 4;
pub(crate) const MAX_UNIFIED_EXEC_SESSIONS: usize = 64;
// Send a warning message to the models when it reaches this number of sessions.
pub(crate) const WARNING_UNIFIED_EXEC_SESSIONS: usize = 60;
pub(crate) struct UnifiedExecContext {
pub session: Arc<Session>,
pub turn: Arc<TurnContext>,
@@ -67,6 +70,7 @@ impl UnifiedExecContext {
#[derive(Debug)]
pub(crate) struct ExecCommandRequest {
pub command: Vec<String>,
pub process_id: String,
pub yield_time_ms: u64,
pub max_output_tokens: Option<usize>,
pub workdir: Option<PathBuf>,
@@ -77,7 +81,7 @@ pub(crate) struct ExecCommandRequest {
#[derive(Debug)]
pub(crate) struct WriteStdinRequest<'a> {
pub call_id: &'a str,
pub session_id: i32,
pub process_id: &'a str,
pub input: &'a str,
pub yield_time_ms: u64,
pub max_output_tokens: Option<usize>,
@@ -89,7 +93,7 @@ pub(crate) struct UnifiedExecResponse {
pub chunk_id: String,
pub wall_time: Duration,
pub output: String,
pub session_id: Option<i32>,
pub process_id: Option<String>,
pub exit_code: Option<i32>,
pub original_token_count: Option<usize>,
pub session_command: Option<Vec<String>>,
@@ -97,15 +101,34 @@ pub(crate) struct UnifiedExecResponse {
#[derive(Default)]
pub(crate) struct UnifiedExecSessionManager {
next_session_id: AtomicI32,
sessions: Mutex<HashMap<i32, SessionEntry>>,
session_store: Mutex<SessionStore>,
}
// Required for mutex sharing.
#[derive(Default)]
pub(crate) struct SessionStore {
sessions: HashMap<String, SessionEntry>,
reserved_sessions_id: HashSet<String>,
}
impl SessionStore {
fn remove(&mut self, session_id: &str) -> Option<SessionEntry> {
self.reserved_sessions_id.remove(session_id);
self.sessions.remove(session_id)
}
pub(crate) fn clear(&mut self) {
self.reserved_sessions_id.clear();
self.sessions.clear();
}
}
struct SessionEntry {
session: session::UnifiedExecSession,
session: UnifiedExecSession,
session_ref: Arc<Session>,
turn_ref: Arc<TurnContext>,
call_id: String,
process_id: String,
command: Vec<String>,
cwd: PathBuf,
started_at: tokio::time::Instant,
@@ -159,6 +182,11 @@ mod tests {
) -> Result<UnifiedExecResponse, UnifiedExecError> {
let context =
UnifiedExecContext::new(Arc::clone(session), Arc::clone(turn), "call".to_string());
let process_id = session
.services
.unified_exec_manager
.allocate_process_id()
.await;
session
.services
@@ -166,6 +194,7 @@ mod tests {
.exec_command(
ExecCommandRequest {
command: vec!["bash".to_string(), "-lc".to_string(), cmd.to_string()],
process_id,
yield_time_ms,
max_output_tokens: None,
workdir: None,
@@ -179,7 +208,7 @@ mod tests {
async fn write_stdin(
session: &Arc<Session>,
session_id: i32,
process_id: &str,
input: &str,
yield_time_ms: u64,
) -> Result<UnifiedExecResponse, UnifiedExecError> {
@@ -188,7 +217,7 @@ mod tests {
.unified_exec_manager
.write_stdin(WriteStdinRequest {
call_id: "write-stdin",
session_id,
process_id,
input,
yield_time_ms,
max_output_tokens: None,
@@ -221,11 +250,15 @@ mod tests {
let (session, turn) = test_session_and_turn();
let open_shell = exec_command(&session, &turn, "bash -i", 2_500).await?;
let session_id = open_shell.session_id.expect("expected session_id");
let process_id = open_shell
.process_id
.as_ref()
.expect("expected process_id")
.as_str();
write_stdin(
&session,
session_id,
process_id,
"export CODEX_INTERACTIVE_SHELL_VAR=codex\n",
2_500,
)
@@ -233,7 +266,7 @@ mod tests {
let out_2 = write_stdin(
&session,
session_id,
process_id,
"echo $CODEX_INTERACTIVE_SHELL_VAR\n",
2_500,
)
@@ -253,11 +286,15 @@ mod tests {
let (session, turn) = test_session_and_turn();
let shell_a = exec_command(&session, &turn, "bash -i", 2_500).await?;
let session_a = shell_a.session_id.expect("expected session id");
let session_a = shell_a
.process_id
.as_ref()
.expect("expected process id")
.clone();
write_stdin(
&session,
session_a,
session_a.as_str(),
"export CODEX_INTERACTIVE_SHELL_VAR=codex\n",
2_500,
)
@@ -265,9 +302,10 @@ mod tests {
let out_2 =
exec_command(&session, &turn, "echo $CODEX_INTERACTIVE_SHELL_VAR", 2_500).await?;
tokio::time::sleep(Duration::from_secs(2)).await;
assert!(
out_2.session_id.is_none(),
"short command should not retain a session"
out_2.process_id.is_none(),
"short command should not report a process id if it exits quickly"
);
assert!(
!out_2.output.contains("codex"),
@@ -276,7 +314,11 @@ mod tests {
let out_3 = write_stdin(
&session,
session_a,
shell_a
.process_id
.as_ref()
.expect("expected process id")
.as_str(),
"echo $CODEX_INTERACTIVE_SHELL_VAR\n",
2_500,
)
@@ -296,11 +338,15 @@ mod tests {
let (session, turn) = test_session_and_turn();
let open_shell = exec_command(&session, &turn, "bash -i", 2_500).await?;
let session_id = open_shell.session_id.expect("expected session id");
let process_id = open_shell
.process_id
.as_ref()
.expect("expected process id")
.as_str();
write_stdin(
&session,
session_id,
process_id,
"export CODEX_INTERACTIVE_SHELL_VAR=codex\n",
2_500,
)
@@ -308,7 +354,7 @@ mod tests {
let out_2 = write_stdin(
&session,
session_id,
process_id,
"sleep 5 && echo $CODEX_INTERACTIVE_SHELL_VAR\n",
10,
)
@@ -320,7 +366,7 @@ mod tests {
tokio::time::sleep(Duration::from_secs(7)).await;
let out_3 = write_stdin(&session, session_id, "", 100).await?;
let out_3 = write_stdin(&session, process_id, "", 100).await?;
assert!(
out_3.output.contains("codex"),
@@ -337,7 +383,7 @@ mod tests {
let result = exec_command(&session, &turn, "echo codex", 120_000).await?;
assert!(result.session_id.is_none());
assert!(result.process_id.is_some());
assert!(result.output.contains("codex"));
Ok(())
@@ -350,8 +396,8 @@ mod tests {
let result = exec_command(&session, &turn, "echo codex", 2_500).await?;
assert!(
result.session_id.is_none(),
"completed command should not retain session"
result.process_id.is_some(),
"completed command should report a process id"
);
assert!(result.output.contains("codex"));
@@ -359,9 +405,10 @@ mod tests {
session
.services
.unified_exec_manager
.sessions
.session_store
.lock()
.await
.sessions
.is_empty()
);
@@ -375,31 +422,36 @@ mod tests {
let (session, turn) = test_session_and_turn();
let open_shell = exec_command(&session, &turn, "bash -i", 2_500).await?;
let session_id = open_shell.session_id.expect("expected session id");
let process_id = open_shell
.process_id
.as_ref()
.expect("expected process id")
.as_str();
write_stdin(&session, session_id, "exit\n", 2_500).await?;
write_stdin(&session, process_id, "exit\n", 2_500).await?;
tokio::time::sleep(Duration::from_millis(200)).await;
let err = write_stdin(&session, session_id, "", 100)
let err = write_stdin(&session, process_id, "", 100)
.await
.expect_err("expected unknown session error");
match err {
UnifiedExecError::UnknownSessionId { session_id: err_id } => {
assert_eq!(err_id, session_id);
UnifiedExecError::UnknownSessionId { process_id: err_id } => {
assert_eq!(err_id, process_id, "process id should match request");
}
other => panic!("expected UnknownSessionId, got {other:?}"),
}
assert!(
!session
session
.services
.unified_exec_manager
.sessions
.session_store
.lock()
.await
.contains_key(&session_id)
.sessions
.is_empty()
);
Ok(())

View File

@@ -1,9 +1,9 @@
use rand::Rng;
use std::cmp::Reverse;
use std::collections::HashMap;
use std::collections::HashSet;
use std::path::PathBuf;
use std::sync::Arc;
use tokio::sync::Notify;
use tokio::sync::mpsc;
use tokio::time::Duration;
@@ -36,10 +36,12 @@ use crate::truncate::formatted_truncate_text;
use super::ExecCommandRequest;
use super::MAX_UNIFIED_EXEC_SESSIONS;
use super::SessionEntry;
use super::SessionStore;
use super::UnifiedExecContext;
use super::UnifiedExecError;
use super::UnifiedExecResponse;
use super::UnifiedExecSessionManager;
use super::WARNING_UNIFIED_EXEC_SESSIONS;
use super::WriteStdinRequest;
use super::clamp_yield_time;
use super::generate_chunk_id;
@@ -75,9 +77,39 @@ struct PreparedSessionHandles {
turn_ref: Arc<TurnContext>,
command: Vec<String>,
cwd: PathBuf,
process_id: String,
}
impl UnifiedExecSessionManager {
pub(crate) async fn allocate_process_id(&self) -> String {
loop {
let mut store = self.session_store.lock().await;
let process_id = if !cfg!(test) && !cfg!(feature = "deterministic_process_ids") {
// production mode → random
rand::rng().random_range(1_000..100_000).to_string()
} else {
// test or deterministic mode
let next = store
.reserved_sessions_id
.iter()
.filter_map(|s| s.parse::<i32>().ok())
.max()
.map(|m| std::cmp::max(m, 999) + 1)
.unwrap_or(1000);
next.to_string()
};
if store.reserved_sessions_id.contains(&process_id) {
continue;
}
store.reserved_sessions_id.insert(process_id.clone());
return process_id;
}
}
pub(crate) async fn exec_command(
&self,
request: ExecCommandRequest,
@@ -122,14 +154,20 @@ impl UnifiedExecSessionManager {
let has_exited = session.has_exited();
let exit_code = session.exit_code();
let chunk_id = generate_chunk_id();
let session_id = if has_exited {
let process_id = if has_exited {
None
} else {
// Only store session if not exited.
let stored_id = self
.store_session(session, context, &request.command, cwd.clone(), start)
.await;
Some(stored_id)
self.store_session(
session,
context,
&request.command,
cwd.clone(),
start,
request.process_id.clone(),
)
.await;
Some(request.process_id.clone())
};
let original_token_count = approx_token_count(&text);
@@ -138,18 +176,18 @@ impl UnifiedExecSessionManager {
chunk_id,
wall_time,
output,
session_id,
process_id: process_id.clone(),
exit_code,
original_token_count: Some(original_token_count),
session_command: Some(request.command.clone()),
};
if response.session_id.is_some() {
if !has_exited {
Self::emit_waiting_status(&context.session, &context.turn, &request.command).await;
}
// If the command completed during this call, emit an ExecCommandEnd via the emitter.
if response.session_id.is_none() {
if has_exited {
let exit = response.exit_code.unwrap_or(-1);
Self::emit_exec_end_from_context(
context,
@@ -158,6 +196,9 @@ impl UnifiedExecSessionManager {
response.output.clone(),
exit,
response.wall_time,
// We always emit the process ID in order to keep consistency between the Begin
// event and the End event.
Some(request.process_id),
)
.await;
}
@@ -169,7 +210,7 @@ impl UnifiedExecSessionManager {
&self,
request: WriteStdinRequest<'_>,
) -> Result<UnifiedExecResponse, UnifiedExecError> {
let session_id = request.session_id;
let process_id = request.process_id.to_string();
let PreparedSessionHandles {
writer_tx,
@@ -180,13 +221,15 @@ impl UnifiedExecSessionManager {
turn_ref,
command: session_command,
cwd: session_cwd,
} = self.prepare_session_handles(session_id).await?;
process_id,
} = self.prepare_session_handles(process_id.as_str()).await?;
let interaction_emitter = ToolEmitter::unified_exec(
&session_command,
session_cwd.clone(),
ExecCommandSource::UnifiedExecInteraction,
(!request.input.is_empty()).then(|| request.input.to_string()),
Some(process_id.clone()),
);
let make_event_ctx = || {
ToolEventCtx::new(
@@ -233,17 +276,21 @@ impl UnifiedExecSessionManager {
let original_token_count = approx_token_count(&text);
let chunk_id = generate_chunk_id();
let status = self.refresh_session_state(session_id).await;
let (session_id, exit_code, completion_entry, event_call_id) = match status {
SessionStatus::Alive { exit_code, call_id } => {
(Some(session_id), exit_code, None, call_id)
}
let status = self.refresh_session_state(process_id.as_str()).await;
let (process_id, exit_code, completion_entry, event_call_id) = match status {
SessionStatus::Alive {
exit_code,
call_id,
process_id,
} => (Some(process_id), exit_code, None, call_id),
SessionStatus::Exited { exit_code, entry } => {
let call_id = entry.call_id.clone();
(None, exit_code, Some(*entry), call_id)
}
SessionStatus::Unknown => {
return Err(UnifiedExecError::UnknownSessionId { session_id });
return Err(UnifiedExecError::UnknownSessionId {
process_id: request.process_id.to_string(),
});
}
};
@@ -252,7 +299,7 @@ impl UnifiedExecSessionManager {
chunk_id,
wall_time,
output,
session_id,
process_id,
exit_code,
original_token_count: Some(original_token_count),
session_command: Some(session_command.clone()),
@@ -273,7 +320,7 @@ impl UnifiedExecSessionManager {
)
.await;
if response.session_id.is_some() {
if response.process_id.is_some() {
Self::emit_waiting_status(&session_ref, &turn_ref, &session_command).await;
}
@@ -286,16 +333,17 @@ impl UnifiedExecSessionManager {
Ok(response)
}
async fn refresh_session_state(&self, session_id: i32) -> SessionStatus {
let mut sessions = self.sessions.lock().await;
let Some(entry) = sessions.get(&session_id) else {
async fn refresh_session_state(&self, process_id: &str) -> SessionStatus {
let mut store = self.session_store.lock().await;
let Some(entry) = store.sessions.get(process_id) else {
return SessionStatus::Unknown;
};
let exit_code = entry.session.exit_code();
let process_id = entry.process_id.clone();
if entry.session.has_exited() {
let Some(entry) = sessions.remove(&session_id) else {
let Some(entry) = store.remove(&process_id) else {
return SessionStatus::Unknown;
};
SessionStatus::Exited {
@@ -306,18 +354,23 @@ impl UnifiedExecSessionManager {
SessionStatus::Alive {
exit_code,
call_id: entry.call_id.clone(),
process_id,
}
}
}
async fn prepare_session_handles(
&self,
session_id: i32,
process_id: &str,
) -> Result<PreparedSessionHandles, UnifiedExecError> {
let mut sessions = self.sessions.lock().await;
let entry = sessions
.get_mut(&session_id)
.ok_or(UnifiedExecError::UnknownSessionId { session_id })?;
let mut store = self.session_store.lock().await;
let entry =
store
.sessions
.get_mut(process_id)
.ok_or(UnifiedExecError::UnknownSessionId {
process_id: process_id.to_string(),
})?;
entry.last_used = Instant::now();
let OutputHandles {
output_buffer,
@@ -334,6 +387,7 @@ impl UnifiedExecSessionManager {
turn_ref: Arc::clone(&entry.turn_ref),
command: entry.command.clone(),
cwd: entry.cwd.clone(),
process_id: entry.process_id.clone(),
})
}
@@ -347,6 +401,7 @@ impl UnifiedExecSessionManager {
.map_err(|_| UnifiedExecError::WriteToStdin)
}
#[allow(clippy::too_many_arguments)]
async fn store_session(
&self,
session: UnifiedExecSession,
@@ -354,24 +409,35 @@ impl UnifiedExecSessionManager {
command: &[String],
cwd: PathBuf,
started_at: Instant,
) -> i32 {
let session_id = self
.next_session_id
.fetch_add(1, std::sync::atomic::Ordering::SeqCst);
process_id: String,
) {
let entry = SessionEntry {
session,
session_ref: Arc::clone(&context.session),
turn_ref: Arc::clone(&context.turn),
call_id: context.call_id.clone(),
process_id: process_id.clone(),
command: command.to_vec(),
cwd,
started_at,
last_used: started_at,
};
let mut sessions = self.sessions.lock().await;
Self::prune_sessions_if_needed(&mut sessions);
sessions.insert(session_id, entry);
session_id
let number_sessions = {
let mut store = self.session_store.lock().await;
Self::prune_sessions_if_needed(&mut store);
store.sessions.insert(process_id, entry);
store.sessions.len()
};
if number_sessions >= WARNING_UNIFIED_EXEC_SESSIONS {
context
.session
.record_model_warning(
format!("The maximum number of unified exec sessions you can keep open is {WARNING_UNIFIED_EXEC_SESSIONS} and you currently have {number_sessions} sessions open. Reuse older sessions or close them to prevent automatic pruning of old session"),
&context.turn
)
.await;
};
}
async fn emit_exec_end_from_entry(
@@ -399,6 +465,7 @@ impl UnifiedExecSessionManager {
entry.cwd,
ExecCommandSource::UnifiedExecStartup,
None,
Some(entry.process_id.clone()),
);
emitter
.emit(event_ctx, ToolEventStage::Success(output))
@@ -412,6 +479,7 @@ impl UnifiedExecSessionManager {
aggregated_output: String,
exit_code: i32,
duration: Duration,
process_id: Option<String>,
) {
let output = ExecToolCallOutput {
exit_code,
@@ -427,8 +495,13 @@ impl UnifiedExecSessionManager {
&context.call_id,
None,
);
let emitter =
ToolEmitter::unified_exec(command, cwd, ExecCommandSource::UnifiedExecStartup, None);
let emitter = ToolEmitter::unified_exec(
command,
cwd,
ExecCommandSource::UnifiedExecStartup,
None,
process_id,
);
emitter
.emit(event_ctx, ToolEventStage::Success(output))
.await;
@@ -481,19 +554,21 @@ impl UnifiedExecSessionManager {
let env = apply_unified_exec_env(create_env(&context.turn.shell_environment_policy));
let mut orchestrator = ToolOrchestrator::new();
let mut runtime = UnifiedExecRuntime::new(self);
let approval_requirement = create_approval_requirement_for_command(
&context.turn.exec_policy,
command,
context.turn.approval_policy,
&context.turn.sandbox_policy,
SandboxPermissions::from(with_escalated_permissions.unwrap_or(false)),
)
.await;
let req = UnifiedExecToolRequest::new(
command.to_vec(),
cwd,
env,
with_escalated_permissions,
justification,
create_approval_requirement_for_command(
&context.turn.exec_policy,
command,
context.turn.approval_policy,
&context.turn.sandbox_policy,
SandboxPermissions::from(with_escalated_permissions.unwrap_or(false)),
),
approval_requirement,
);
let tool_ctx = ToolCtx {
session: context.session.as_ref(),
@@ -574,52 +649,56 @@ impl UnifiedExecSessionManager {
collected
}
fn prune_sessions_if_needed(sessions: &mut HashMap<i32, SessionEntry>) {
if sessions.len() < MAX_UNIFIED_EXEC_SESSIONS {
return;
fn prune_sessions_if_needed(store: &mut SessionStore) -> bool {
if store.sessions.len() < MAX_UNIFIED_EXEC_SESSIONS {
return false;
}
let meta: Vec<(i32, Instant, bool)> = sessions
let meta: Vec<(String, Instant, bool)> = store
.sessions
.iter()
.map(|(id, entry)| (*id, entry.last_used, entry.session.has_exited()))
.map(|(id, entry)| (id.clone(), entry.last_used, entry.session.has_exited()))
.collect();
if let Some(session_id) = Self::session_id_to_prune_from_meta(&meta) {
sessions.remove(&session_id);
store.remove(&session_id);
return true;
}
false
}
// Centralized pruning policy so we can easily swap strategies later.
fn session_id_to_prune_from_meta(meta: &[(i32, Instant, bool)]) -> Option<i32> {
fn session_id_to_prune_from_meta(meta: &[(String, Instant, bool)]) -> Option<String> {
if meta.is_empty() {
return None;
}
let mut by_recency = meta.to_vec();
by_recency.sort_by_key(|(_, last_used, _)| Reverse(*last_used));
let protected: HashSet<i32> = by_recency
let protected: HashSet<String> = by_recency
.iter()
.take(8)
.map(|(session_id, _, _)| *session_id)
.map(|(process_id, _, _)| process_id.clone())
.collect();
let mut lru = meta.to_vec();
lru.sort_by_key(|(_, last_used, _)| *last_used);
if let Some((session_id, _, _)) = lru
if let Some((process_id, _, _)) = lru
.iter()
.find(|(session_id, _, exited)| !protected.contains(session_id) && *exited)
.find(|(process_id, _, exited)| !protected.contains(process_id) && *exited)
{
return Some(*session_id);
return Some(process_id.clone());
}
lru.into_iter()
.find(|(session_id, _, _)| !protected.contains(session_id))
.map(|(session_id, _, _)| session_id)
.find(|(process_id, _, _)| !protected.contains(process_id))
.map(|(process_id, _, _)| process_id)
}
pub(crate) async fn terminate_all_sessions(&self) {
let mut sessions = self.sessions.lock().await;
let mut sessions = self.session_store.lock().await;
sessions.clear();
}
}
@@ -628,6 +707,7 @@ enum SessionStatus {
Alive {
exit_code: Option<i32>,
call_id: String,
process_id: String,
},
Exited {
exit_code: Option<i32>,
@@ -675,64 +755,67 @@ mod tests {
#[test]
fn pruning_prefers_exited_sessions_outside_recently_used() {
let now = Instant::now();
let id = |n: i32| n.to_string();
let meta = vec![
(1, now - Duration::from_secs(40), false),
(2, now - Duration::from_secs(30), true),
(3, now - Duration::from_secs(20), false),
(4, now - Duration::from_secs(19), false),
(5, now - Duration::from_secs(18), false),
(6, now - Duration::from_secs(17), false),
(7, now - Duration::from_secs(16), false),
(8, now - Duration::from_secs(15), false),
(9, now - Duration::from_secs(14), false),
(10, now - Duration::from_secs(13), false),
(id(1), now - Duration::from_secs(40), false),
(id(2), now - Duration::from_secs(30), true),
(id(3), now - Duration::from_secs(20), false),
(id(4), now - Duration::from_secs(19), false),
(id(5), now - Duration::from_secs(18), false),
(id(6), now - Duration::from_secs(17), false),
(id(7), now - Duration::from_secs(16), false),
(id(8), now - Duration::from_secs(15), false),
(id(9), now - Duration::from_secs(14), false),
(id(10), now - Duration::from_secs(13), false),
];
let candidate = UnifiedExecSessionManager::session_id_to_prune_from_meta(&meta);
assert_eq!(candidate, Some(2));
assert_eq!(candidate, Some(id(2)));
}
#[test]
fn pruning_falls_back_to_lru_when_no_exited() {
let now = Instant::now();
let id = |n: i32| n.to_string();
let meta = vec![
(1, now - Duration::from_secs(40), false),
(2, now - Duration::from_secs(30), false),
(3, now - Duration::from_secs(20), false),
(4, now - Duration::from_secs(19), false),
(5, now - Duration::from_secs(18), false),
(6, now - Duration::from_secs(17), false),
(7, now - Duration::from_secs(16), false),
(8, now - Duration::from_secs(15), false),
(9, now - Duration::from_secs(14), false),
(10, now - Duration::from_secs(13), false),
(id(1), now - Duration::from_secs(40), false),
(id(2), now - Duration::from_secs(30), false),
(id(3), now - Duration::from_secs(20), false),
(id(4), now - Duration::from_secs(19), false),
(id(5), now - Duration::from_secs(18), false),
(id(6), now - Duration::from_secs(17), false),
(id(7), now - Duration::from_secs(16), false),
(id(8), now - Duration::from_secs(15), false),
(id(9), now - Duration::from_secs(14), false),
(id(10), now - Duration::from_secs(13), false),
];
let candidate = UnifiedExecSessionManager::session_id_to_prune_from_meta(&meta);
assert_eq!(candidate, Some(1));
assert_eq!(candidate, Some(id(1)));
}
#[test]
fn pruning_protects_recent_sessions_even_if_exited() {
let now = Instant::now();
let id = |n: i32| n.to_string();
let meta = vec![
(1, now - Duration::from_secs(40), false),
(2, now - Duration::from_secs(30), false),
(3, now - Duration::from_secs(20), true),
(4, now - Duration::from_secs(19), false),
(5, now - Duration::from_secs(18), false),
(6, now - Duration::from_secs(17), false),
(7, now - Duration::from_secs(16), false),
(8, now - Duration::from_secs(15), false),
(9, now - Duration::from_secs(14), false),
(10, now - Duration::from_secs(13), true),
(id(1), now - Duration::from_secs(40), false),
(id(2), now - Duration::from_secs(30), false),
(id(3), now - Duration::from_secs(20), true),
(id(4), now - Duration::from_secs(19), false),
(id(5), now - Duration::from_secs(18), false),
(id(6), now - Duration::from_secs(17), false),
(id(7), now - Duration::from_secs(16), false),
(id(8), now - Duration::from_secs(15), false),
(id(9), now - Duration::from_secs(14), false),
(id(10), now - Duration::from_secs(13), true),
];
let candidate = UnifiedExecSessionManager::session_id_to_prune_from_meta(&meta);
// (10) is exited but among the last 8; we should drop the LRU outside that set.
assert_eq!(candidate, Some(1));
assert_eq!(candidate, Some(id(1)));
}
}

View File

@@ -1,3 +1,5 @@
use std::path::Path;
use std::path::PathBuf;
use std::time::Duration;
use rand::Rng;
@@ -14,11 +16,11 @@ pub(crate) fn backoff(attempt: u64) -> Duration {
Duration::from_millis((base as f64 * jitter) as u64)
}
pub(crate) fn error_or_panic(message: String) {
pub(crate) fn error_or_panic(message: impl std::string::ToString) {
if cfg!(debug_assertions) || env!("CARGO_PKG_VERSION").contains("alpha") {
panic!("{message}");
panic!("{}", message.to_string());
} else {
error!("{message}");
error!("{}", message.to_string());
}
}
@@ -37,6 +39,14 @@ pub(crate) fn try_parse_error_message(text: &str) -> String {
text.to_string()
}
pub fn resolve_path(base: &Path, path: &PathBuf) -> PathBuf {
if path.is_absolute() {
path.clone()
} else {
base.join(path)
}
}
#[cfg(test)]
mod tests {
use super::*;

View File

@@ -431,6 +431,9 @@ pub fn ev_apply_patch_call(
ApplyPatchModelOutput::ShellViaHeredoc => {
ev_apply_patch_shell_call_via_heredoc(call_id, patch)
}
ApplyPatchModelOutput::ShellCommandViaHeredoc => {
ev_apply_patch_shell_command_call_via_heredoc(call_id, patch)
}
}
}
@@ -492,6 +495,13 @@ pub fn ev_apply_patch_shell_call_via_heredoc(call_id: &str, patch: &str) -> Valu
ev_function_call(call_id, "shell", &arguments)
}
pub fn ev_apply_patch_shell_command_call_via_heredoc(call_id: &str, patch: &str) -> Value {
let args = serde_json::json!({ "command": format!("apply_patch <<'EOF'\n{patch}\nEOF\n") });
let arguments = serde_json::to_string(&args).expect("serialize apply_patch arguments");
ev_function_call(call_id, "shell_command", &arguments)
}
pub fn sse_failed(id: &str, code: &str, message: &str) -> String {
sse(vec![serde_json::json!({
"type": "response.failed",
@@ -508,6 +518,32 @@ pub fn sse_response(body: String) -> ResponseTemplate {
.set_body_raw(body, "text/event-stream")
}
pub async fn mount_response_once(server: &MockServer, response: ResponseTemplate) -> ResponseMock {
let (mock, response_mock) = base_mock();
mock.respond_with(response)
.up_to_n_times(1)
.mount(server)
.await;
response_mock
}
pub async fn mount_response_once_match<M>(
server: &MockServer,
matcher: M,
response: ResponseTemplate,
) -> ResponseMock
where
M: wiremock::Match + Send + Sync + 'static,
{
let (mock, response_mock) = base_mock();
mock.and(matcher)
.respond_with(response)
.up_to_n_times(1)
.mount(server)
.await;
response_mock
}
fn base_mock() -> (MockBuilder, ResponseMock) {
let response_mock = ResponseMock::new();
let mock = Mock::given(method("POST"))

View File

@@ -36,6 +36,7 @@ pub enum ApplyPatchModelOutput {
Function,
Shell,
ShellViaHeredoc,
ShellCommandViaHeredoc,
}
/// A collection of different ways the model can output an apply_patch call
@@ -312,7 +313,10 @@ impl TestCodexHarness {
ApplyPatchModelOutput::Freeform => self.custom_tool_call_output(call_id).await,
ApplyPatchModelOutput::Function
| ApplyPatchModelOutput::Shell
| ApplyPatchModelOutput::ShellViaHeredoc => self.function_call_stdout(call_id).await,
| ApplyPatchModelOutput::ShellViaHeredoc
| ApplyPatchModelOutput::ShellCommandViaHeredoc => {
self.function_call_stdout(call_id).await
}
}
}
}

View File

@@ -2,6 +2,7 @@
use anyhow::Result;
use core_test_support::responses::ev_apply_patch_call;
use core_test_support::responses::ev_shell_command_call;
use core_test_support::test_codex::ApplyPatchModelOutput;
use pretty_assertions::assert_eq;
use std::fs;
@@ -127,6 +128,7 @@ D delete.txt
#[test_case(ApplyPatchModelOutput::Function)]
#[test_case(ApplyPatchModelOutput::Shell)]
#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)]
#[test_case(ApplyPatchModelOutput::ShellCommandViaHeredoc)]
async fn apply_patch_cli_multiple_chunks(model_output: ApplyPatchModelOutput) -> Result<()> {
skip_if_no_network!(Ok(()));
@@ -153,6 +155,7 @@ async fn apply_patch_cli_multiple_chunks(model_output: ApplyPatchModelOutput) ->
#[test_case(ApplyPatchModelOutput::Function)]
#[test_case(ApplyPatchModelOutput::Shell)]
#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)]
#[test_case(ApplyPatchModelOutput::ShellCommandViaHeredoc)]
async fn apply_patch_cli_moves_file_to_new_directory(
model_output: ApplyPatchModelOutput,
) -> Result<()> {
@@ -181,6 +184,7 @@ async fn apply_patch_cli_moves_file_to_new_directory(
#[test_case(ApplyPatchModelOutput::Function)]
#[test_case(ApplyPatchModelOutput::Shell)]
#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)]
#[test_case(ApplyPatchModelOutput::ShellCommandViaHeredoc)]
async fn apply_patch_cli_updates_file_appends_trailing_newline(
model_output: ApplyPatchModelOutput,
) -> Result<()> {
@@ -208,6 +212,7 @@ async fn apply_patch_cli_updates_file_appends_trailing_newline(
#[test_case(ApplyPatchModelOutput::Function)]
#[test_case(ApplyPatchModelOutput::Shell)]
#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)]
#[test_case(ApplyPatchModelOutput::ShellCommandViaHeredoc)]
async fn apply_patch_cli_insert_only_hunk_modifies_file(
model_output: ApplyPatchModelOutput,
) -> Result<()> {
@@ -233,6 +238,7 @@ async fn apply_patch_cli_insert_only_hunk_modifies_file(
#[test_case(ApplyPatchModelOutput::Function)]
#[test_case(ApplyPatchModelOutput::Shell)]
#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)]
#[test_case(ApplyPatchModelOutput::ShellCommandViaHeredoc)]
async fn apply_patch_cli_move_overwrites_existing_destination(
model_output: ApplyPatchModelOutput,
) -> Result<()> {
@@ -263,6 +269,7 @@ async fn apply_patch_cli_move_overwrites_existing_destination(
#[test_case(ApplyPatchModelOutput::Function)]
#[test_case(ApplyPatchModelOutput::Shell)]
#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)]
#[test_case(ApplyPatchModelOutput::ShellCommandViaHeredoc)]
async fn apply_patch_cli_move_without_content_change_has_no_turn_diff(
model_output: ApplyPatchModelOutput,
) -> Result<()> {
@@ -320,6 +327,7 @@ async fn apply_patch_cli_move_without_content_change_has_no_turn_diff(
#[test_case(ApplyPatchModelOutput::Function)]
#[test_case(ApplyPatchModelOutput::Shell)]
#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)]
#[test_case(ApplyPatchModelOutput::ShellCommandViaHeredoc)]
async fn apply_patch_cli_add_overwrites_existing_file(
model_output: ApplyPatchModelOutput,
) -> Result<()> {
@@ -345,6 +353,7 @@ async fn apply_patch_cli_add_overwrites_existing_file(
#[test_case(ApplyPatchModelOutput::Function)]
#[test_case(ApplyPatchModelOutput::Shell)]
#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)]
#[test_case(ApplyPatchModelOutput::ShellCommandViaHeredoc)]
async fn apply_patch_cli_rejects_invalid_hunk_header(
model_output: ApplyPatchModelOutput,
) -> Result<()> {
@@ -376,6 +385,7 @@ async fn apply_patch_cli_rejects_invalid_hunk_header(
#[test_case(ApplyPatchModelOutput::Function)]
#[test_case(ApplyPatchModelOutput::Shell)]
#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)]
#[test_case(ApplyPatchModelOutput::ShellCommandViaHeredoc)]
async fn apply_patch_cli_reports_missing_context(
model_output: ApplyPatchModelOutput,
) -> Result<()> {
@@ -409,6 +419,7 @@ async fn apply_patch_cli_reports_missing_context(
#[test_case(ApplyPatchModelOutput::Function)]
#[test_case(ApplyPatchModelOutput::Shell)]
#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)]
#[test_case(ApplyPatchModelOutput::ShellCommandViaHeredoc)]
async fn apply_patch_cli_reports_missing_target_file(
model_output: ApplyPatchModelOutput,
) -> Result<()> {
@@ -444,6 +455,7 @@ async fn apply_patch_cli_reports_missing_target_file(
#[test_case(ApplyPatchModelOutput::Function)]
#[test_case(ApplyPatchModelOutput::Shell)]
#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)]
#[test_case(ApplyPatchModelOutput::ShellCommandViaHeredoc)]
async fn apply_patch_cli_delete_missing_file_reports_error(
model_output: ApplyPatchModelOutput,
) -> Result<()> {
@@ -480,6 +492,7 @@ async fn apply_patch_cli_delete_missing_file_reports_error(
#[test_case(ApplyPatchModelOutput::Function)]
#[test_case(ApplyPatchModelOutput::Shell)]
#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)]
#[test_case(ApplyPatchModelOutput::ShellCommandViaHeredoc)]
async fn apply_patch_cli_rejects_empty_patch(model_output: ApplyPatchModelOutput) -> Result<()> {
skip_if_no_network!(Ok(()));
@@ -504,6 +517,7 @@ async fn apply_patch_cli_rejects_empty_patch(model_output: ApplyPatchModelOutput
#[test_case(ApplyPatchModelOutput::Function)]
#[test_case(ApplyPatchModelOutput::Shell)]
#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)]
#[test_case(ApplyPatchModelOutput::ShellCommandViaHeredoc)]
async fn apply_patch_cli_delete_directory_reports_verification_error(
model_output: ApplyPatchModelOutput,
) -> Result<()> {
@@ -530,6 +544,7 @@ async fn apply_patch_cli_delete_directory_reports_verification_error(
#[test_case(ApplyPatchModelOutput::Function)]
#[test_case(ApplyPatchModelOutput::Shell)]
#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)]
#[test_case(ApplyPatchModelOutput::ShellCommandViaHeredoc)]
async fn apply_patch_cli_rejects_path_traversal_outside_workspace(
model_output: ApplyPatchModelOutput,
) -> Result<()> {
@@ -582,6 +597,7 @@ async fn apply_patch_cli_rejects_path_traversal_outside_workspace(
#[test_case(ApplyPatchModelOutput::Function)]
#[test_case(ApplyPatchModelOutput::Shell)]
#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)]
#[test_case(ApplyPatchModelOutput::ShellCommandViaHeredoc)]
async fn apply_patch_cli_rejects_move_path_traversal_outside_workspace(
model_output: ApplyPatchModelOutput,
) -> Result<()> {
@@ -635,6 +651,7 @@ async fn apply_patch_cli_rejects_move_path_traversal_outside_workspace(
#[test_case(ApplyPatchModelOutput::Function)]
#[test_case(ApplyPatchModelOutput::Shell)]
#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)]
#[test_case(ApplyPatchModelOutput::ShellCommandViaHeredoc)]
async fn apply_patch_cli_verification_failure_has_no_side_effects(
model_output: ApplyPatchModelOutput,
) -> Result<()> {
@@ -677,11 +694,10 @@ async fn apply_patch_shell_command_heredoc_with_cd_updates_relative_workdir() ->
let script = "cd sub && apply_patch <<'EOF'\n*** Begin Patch\n*** Update File: in_sub.txt\n@@\n-before\n+after\n*** End Patch\nEOF\n";
let call_id = "shell-heredoc-cd";
let args = json!({ "command": script, "timeout_ms": 5_000 });
let bodies = vec![
sse(vec![
ev_response_created("resp-1"),
ev_function_call(call_id, "shell_command", &serde_json::to_string(&args)?),
ev_shell_command_call(call_id, script),
ev_completed("resp-1"),
]),
sse(vec![
@@ -702,6 +718,86 @@ async fn apply_patch_shell_command_heredoc_with_cd_updates_relative_workdir() ->
Ok(())
}
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
async fn apply_patch_shell_command_heredoc_with_cd_emits_turn_diff() -> Result<()> {
skip_if_no_network!(Ok(()));
let harness = apply_patch_harness_with(|builder| builder.with_model("gpt-5.1")).await?;
let test = harness.test();
let codex = test.codex.clone();
let cwd = test.cwd.clone();
// Prepare a file inside a subdir; update it via cd && apply_patch heredoc form.
let sub = test.workspace_path("sub");
fs::create_dir_all(&sub)?;
let target = sub.join("in_sub.txt");
fs::write(&target, "before\n")?;
let script = "cd sub && apply_patch <<'EOF'\n*** Begin Patch\n*** Update File: in_sub.txt\n@@\n-before\n+after\n*** End Patch\nEOF\n";
let call_id = "shell-heredoc-cd";
let args = json!({ "command": script, "timeout_ms": 5_000 });
let bodies = vec![
sse(vec![
ev_response_created("resp-1"),
ev_function_call(call_id, "shell_command", &serde_json::to_string(&args)?),
ev_completed("resp-1"),
]),
sse(vec![
ev_assistant_message("msg-1", "ok"),
ev_completed("resp-2"),
]),
];
mount_sse_sequence(harness.server(), bodies).await;
let model = test.session_configured.model.clone();
codex
.submit(Op::UserTurn {
items: vec![UserInput::Text {
text: "apply via shell heredoc with cd".into(),
}],
final_output_json_schema: None,
cwd: cwd.path().to_path_buf(),
approval_policy: AskForApproval::Never,
sandbox_policy: SandboxPolicy::DangerFullAccess,
model,
effort: None,
summary: ReasoningSummary::Auto,
})
.await?;
let mut saw_turn_diff = None;
let mut saw_patch_begin = false;
let mut patch_end_success = None;
wait_for_event(&codex, |event| match event {
EventMsg::PatchApplyBegin(begin) => {
saw_patch_begin = true;
assert_eq!(begin.call_id, call_id);
false
}
EventMsg::PatchApplyEnd(end) => {
assert_eq!(end.call_id, call_id);
patch_end_success = Some(end.success);
false
}
EventMsg::TurnDiff(ev) => {
saw_turn_diff = Some(ev.unified_diff.clone());
false
}
EventMsg::TaskComplete(_) => true,
_ => false,
})
.await;
assert!(saw_patch_begin, "expected PatchApplyBegin event");
let patch_end_success =
patch_end_success.expect("expected PatchApplyEnd event to capture success flag");
assert!(patch_end_success);
let diff = saw_turn_diff.expect("expected TurnDiff event");
assert!(diff.contains("diff --git"), "diff header missing: {diff:?}");
Ok(())
}
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
async fn apply_patch_shell_command_failure_propagates_error_and_skips_diff() -> Result<()> {
skip_if_no_network!(Ok(()));
@@ -776,7 +872,11 @@ async fn apply_patch_shell_command_failure_propagates_error_and_skips_diff() ->
}
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
async fn apply_patch_function_accepts_lenient_heredoc_wrapped_patch() -> Result<()> {
#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)]
#[test_case(ApplyPatchModelOutput::ShellCommandViaHeredoc)]
async fn apply_patch_function_accepts_lenient_heredoc_wrapped_patch(
model_output: ApplyPatchModelOutput,
) -> Result<()> {
skip_if_no_network!(Ok(()));
let harness = apply_patch_harness().await?;
@@ -784,16 +884,8 @@ async fn apply_patch_function_accepts_lenient_heredoc_wrapped_patch() -> Result<
let file_name = "lenient.txt";
let patch_inner =
format!("*** Begin Patch\n*** Add File: {file_name}\n+lenient\n*** End Patch\n");
let wrapped = format!("<<'EOF'\n{patch_inner}EOF\n");
let call_id = "apply-lenient";
mount_apply_patch(
&harness,
call_id,
wrapped.as_str(),
"ok",
ApplyPatchModelOutput::Function,
)
.await;
mount_apply_patch(&harness, call_id, patch_inner.as_str(), "ok", model_output).await;
harness.submit("apply lenient heredoc patch").await?;
@@ -807,6 +899,7 @@ async fn apply_patch_function_accepts_lenient_heredoc_wrapped_patch() -> Result<
#[test_case(ApplyPatchModelOutput::Function)]
#[test_case(ApplyPatchModelOutput::Shell)]
#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)]
#[test_case(ApplyPatchModelOutput::ShellCommandViaHeredoc)]
async fn apply_patch_cli_end_of_file_anchor(model_output: ApplyPatchModelOutput) -> Result<()> {
skip_if_no_network!(Ok(()));
@@ -829,6 +922,7 @@ async fn apply_patch_cli_end_of_file_anchor(model_output: ApplyPatchModelOutput)
#[test_case(ApplyPatchModelOutput::Function)]
#[test_case(ApplyPatchModelOutput::Shell)]
#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)]
#[test_case(ApplyPatchModelOutput::ShellCommandViaHeredoc)]
async fn apply_patch_cli_missing_second_chunk_context_rejected(
model_output: ApplyPatchModelOutput,
) -> Result<()> {
@@ -863,6 +957,7 @@ async fn apply_patch_cli_missing_second_chunk_context_rejected(
#[test_case(ApplyPatchModelOutput::Function)]
#[test_case(ApplyPatchModelOutput::Shell)]
#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)]
#[test_case(ApplyPatchModelOutput::ShellCommandViaHeredoc)]
async fn apply_patch_emits_turn_diff_event_with_unified_diff(
model_output: ApplyPatchModelOutput,
) -> Result<()> {
@@ -918,6 +1013,7 @@ async fn apply_patch_emits_turn_diff_event_with_unified_diff(
#[test_case(ApplyPatchModelOutput::Function)]
#[test_case(ApplyPatchModelOutput::Shell)]
#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)]
#[test_case(ApplyPatchModelOutput::ShellCommandViaHeredoc)]
async fn apply_patch_turn_diff_for_rename_with_content_change(
model_output: ApplyPatchModelOutput,
) -> Result<()> {
@@ -1132,6 +1228,7 @@ async fn apply_patch_aggregates_diff_preserves_success_after_failure() -> Result
#[test_case(ApplyPatchModelOutput::Function)]
#[test_case(ApplyPatchModelOutput::Shell)]
#[test_case(ApplyPatchModelOutput::ShellViaHeredoc)]
#[test_case(ApplyPatchModelOutput::ShellCommandViaHeredoc)]
async fn apply_patch_change_context_disambiguates_target(
model_output: ApplyPatchModelOutput,
) -> Result<()> {

View File

@@ -15,6 +15,7 @@ use codex_core::WireApi;
use codex_core::auth::AuthCredentialsStoreMode;
use codex_core::built_in_model_providers;
use codex_core::error::CodexErr;
use codex_core::features::Feature;
use codex_core::model_family::find_family_for_model;
use codex_core::protocol::EventMsg;
use codex_core::protocol::Op;
@@ -34,6 +35,7 @@ use core_test_support::skip_if_no_network;
use core_test_support::test_codex::TestCodex;
use core_test_support::test_codex::test_codex;
use core_test_support::wait_for_event;
use dunce::canonicalize as normalize_path;
use futures::StreamExt;
use serde_json::json;
use std::io::Write;
@@ -620,6 +622,74 @@ async fn includes_user_instructions_message_in_request() {
assert_message_ends_with(&request_body["input"][1], "</environment_context>");
}
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
async fn skills_append_to_instructions_when_feature_enabled() {
skip_if_no_network!();
let server = MockServer::start().await;
let resp_mock = responses::mount_sse_once(&server, sse_completed("resp1")).await;
let model_provider = ModelProviderInfo {
base_url: Some(format!("{}/v1", server.uri())),
..built_in_model_providers()["openai"].clone()
};
let codex_home = TempDir::new().unwrap();
let skill_dir = codex_home.path().join("skills/demo");
std::fs::create_dir_all(&skill_dir).expect("create skill dir");
std::fs::write(
skill_dir.join("SKILL.md"),
"---\nname: demo\ndescription: build charts\n---\n\n# body\n",
)
.expect("write skill");
let mut config = load_default_config_for_test(&codex_home);
config.model_provider = model_provider;
config.features.enable(Feature::Skills);
config.cwd = codex_home.path().to_path_buf();
let conversation_manager =
ConversationManager::with_auth(CodexAuth::from_api_key("Test API Key"));
let codex = conversation_manager
.new_conversation(config)
.await
.expect("create new conversation")
.conversation;
codex
.submit(Op::UserInput {
items: vec![UserInput::Text {
text: "hello".into(),
}],
})
.await
.unwrap();
wait_for_event(&codex, |ev| matches!(ev, EventMsg::TaskComplete(_))).await;
let request = resp_mock.single_request();
let request_body = request.body_json();
assert_message_role(&request_body["input"][0], "user");
let instructions_text = request_body["input"][0]["content"][0]["text"]
.as_str()
.expect("instructions text");
assert!(
instructions_text.contains("## Skills"),
"expected skills section present"
);
assert!(
instructions_text.contains("demo: build charts"),
"expected skill summary"
);
let expected_path = normalize_path(skill_dir.join("SKILL.md")).unwrap();
let expected_path_str = expected_path.to_string_lossy().replace('\\', "/");
assert!(
instructions_text.contains(&expected_path_str),
"expected path {expected_path_str} in instructions"
);
}
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
async fn includes_configured_effort_in_request() -> anyhow::Result<()> {
skip_if_no_network!(Ok(()));

View File

@@ -3,6 +3,7 @@ use codex_core::protocol::EventMsg;
use codex_core::protocol::Op;
use codex_core::protocol::ReviewDecision;
use codex_core::protocol::ReviewRequest;
use codex_core::protocol::ReviewTarget;
use codex_core::protocol::SandboxPolicy;
use core_test_support::responses::ev_apply_patch_function_call;
use core_test_support::responses::ev_assistant_message;
@@ -68,9 +69,10 @@ async fn codex_delegate_forwards_exec_approval_and_proceeds_on_approval() {
test.codex
.submit(Op::Review {
review_request: ReviewRequest {
prompt: "Please review".to_string(),
user_facing_hint: "review".to_string(),
append_to_original_thread: true,
target: ReviewTarget::Custom {
instructions: "Please review".to_string(),
},
user_facing_hint: None,
},
})
.await
@@ -144,9 +146,10 @@ async fn codex_delegate_forwards_patch_approval_and_proceeds_on_decision() {
test.codex
.submit(Op::Review {
review_request: ReviewRequest {
prompt: "Please review".to_string(),
user_facing_hint: "review".to_string(),
append_to_original_thread: true,
target: ReviewTarget::Custom {
instructions: "Please review".to_string(),
},
user_facing_hint: None,
},
})
.await
@@ -199,9 +202,10 @@ async fn codex_delegate_ignores_legacy_deltas() {
test.codex
.submit(Op::Review {
review_request: ReviewRequest {
prompt: "Please review".to_string(),
user_facing_hint: "review".to_string(),
append_to_original_thread: true,
target: ReviewTarget::Custom {
instructions: "Please review".to_string(),
},
user_facing_hint: None,
},
})
.await

View File

@@ -45,6 +45,7 @@ mod resume;
mod review;
mod rmcp_client;
mod rollout_list_find;
mod saved_sessions;
mod seatbelt;
mod shell_serialization;
mod stream_error_allows_next_turn;

View File

@@ -16,8 +16,10 @@ use codex_core::protocol::ReviewFinding;
use codex_core::protocol::ReviewLineRange;
use codex_core::protocol::ReviewOutputEvent;
use codex_core::protocol::ReviewRequest;
use codex_core::protocol::ReviewTarget;
use codex_core::protocol::RolloutItem;
use codex_core::protocol::RolloutLine;
use codex_core::review_format::render_review_output_text;
use codex_protocol::user_input::UserInput;
use core_test_support::load_default_config_for_test;
use core_test_support::load_sse_fixture_with_id_from_str;
@@ -80,9 +82,10 @@ async fn review_op_emits_lifecycle_and_review_output() {
codex
.submit(Op::Review {
review_request: ReviewRequest {
prompt: "Please review my changes".to_string(),
user_facing_hint: "my changes".to_string(),
append_to_original_thread: true,
target: ReviewTarget::Custom {
instructions: "Please review my changes".to_string(),
},
user_facing_hint: None,
},
})
.await
@@ -124,22 +127,36 @@ async fn review_op_emits_lifecycle_and_review_output() {
let mut saw_header = false;
let mut saw_finding_line = false;
let expected_assistant_text = render_review_output_text(&expected);
let mut saw_assistant_plain = false;
let mut saw_assistant_xml = false;
for line in text.lines() {
if line.trim().is_empty() {
continue;
}
let v: serde_json::Value = serde_json::from_str(line).expect("jsonl line");
let rl: RolloutLine = serde_json::from_value(v).expect("rollout line");
if let RolloutItem::ResponseItem(ResponseItem::Message { role, content, .. }) = rl.item
&& role == "user"
{
for c in content {
if let ContentItem::InputText { text } = c {
if text.contains("full review output from reviewer model") {
saw_header = true;
if let RolloutItem::ResponseItem(ResponseItem::Message { role, content, .. }) = rl.item {
if role == "user" {
for c in content {
if let ContentItem::InputText { text } = c {
if text.contains("full review output from reviewer model") {
saw_header = true;
}
if text.contains("- Prefer Stylize helpers — /tmp/file.rs:10-20") {
saw_finding_line = true;
}
}
if text.contains("- Prefer Stylize helpers — /tmp/file.rs:10-20") {
saw_finding_line = true;
}
} else if role == "assistant" {
for c in content {
if let ContentItem::OutputText { text } = c {
if text.contains("<user_action>") {
saw_assistant_xml = true;
}
if text == expected_assistant_text {
saw_assistant_plain = true;
}
}
}
}
@@ -150,6 +167,14 @@ async fn review_op_emits_lifecycle_and_review_output() {
saw_finding_line,
"formatted finding line missing from rollout"
);
assert!(
saw_assistant_plain,
"assistant review output missing from rollout"
);
assert!(
!saw_assistant_xml,
"assistant review output contains user_action markup"
);
server.verify().await;
}
@@ -177,9 +202,10 @@ async fn review_op_with_plain_text_emits_review_fallback() {
codex
.submit(Op::Review {
review_request: ReviewRequest {
prompt: "Plain text review".to_string(),
user_facing_hint: "plain text review".to_string(),
append_to_original_thread: true,
target: ReviewTarget::Custom {
instructions: "Plain text review".to_string(),
},
user_facing_hint: None,
},
})
.await
@@ -236,9 +262,10 @@ async fn review_filters_agent_message_related_events() {
codex
.submit(Op::Review {
review_request: ReviewRequest {
prompt: "Filter streaming events".to_string(),
user_facing_hint: "Filter streaming events".to_string(),
append_to_original_thread: true,
target: ReviewTarget::Custom {
instructions: "Filter streaming events".to_string(),
},
user_facing_hint: None,
},
})
.await
@@ -247,7 +274,7 @@ async fn review_filters_agent_message_related_events() {
let mut saw_entered = false;
let mut saw_exited = false;
// Drain until TaskComplete; assert filtered events never surface.
// Drain until TaskComplete; assert streaming-related events never surface.
wait_for_event(&codex, |event| match event {
EventMsg::TaskComplete(_) => true,
EventMsg::EnteredReviewMode(_) => {
@@ -265,12 +292,6 @@ async fn review_filters_agent_message_related_events() {
EventMsg::AgentMessageDelta(_) => {
panic!("unexpected AgentMessageDelta surfaced during review")
}
EventMsg::ItemCompleted(ev) => match &ev.item {
codex_protocol::items::TurnItem::AgentMessage(_) => {
panic!("unexpected ItemCompleted for TurnItem::AgentMessage surfaced during review")
}
_ => false,
},
_ => false,
})
.await;
@@ -279,8 +300,9 @@ async fn review_filters_agent_message_related_events() {
server.verify().await;
}
/// When the model returns structured JSON in a review, ensure no AgentMessage
/// is emitted; the UI consumes the structured result via ExitedReviewMode.
/// When the model returns structured JSON in a review, ensure only a single
/// non-streaming AgentMessage is emitted; the UI consumes the structured
/// result via ExitedReviewMode plus a final assistant message.
// Windows CI only: bump to 4 workers to prevent SSE/event starvation and test timeouts.
#[cfg_attr(windows, tokio::test(flavor = "multi_thread", worker_threads = 4))]
#[cfg_attr(not(windows), tokio::test(flavor = "multi_thread", worker_threads = 2))]
@@ -321,21 +343,25 @@ async fn review_does_not_emit_agent_message_on_structured_output() {
codex
.submit(Op::Review {
review_request: ReviewRequest {
prompt: "check structured".to_string(),
user_facing_hint: "check structured".to_string(),
append_to_original_thread: true,
target: ReviewTarget::Custom {
instructions: "check structured".to_string(),
},
user_facing_hint: None,
},
})
.await
.unwrap();
// Drain events until TaskComplete; ensure none are AgentMessage.
// Drain events until TaskComplete; ensure we only see a final
// AgentMessage (no streaming assistant messages).
let mut saw_entered = false;
let mut saw_exited = false;
let mut agent_messages = 0;
wait_for_event(&codex, |event| match event {
EventMsg::TaskComplete(_) => true,
EventMsg::AgentMessage(_) => {
panic!("unexpected AgentMessage during review with structured output")
agent_messages += 1;
false
}
EventMsg::EnteredReviewMode(_) => {
saw_entered = true;
@@ -348,6 +374,7 @@ async fn review_does_not_emit_agent_message_on_structured_output() {
_ => false,
})
.await;
assert_eq!(1, agent_messages, "expected exactly one AgentMessage event");
assert!(saw_entered && saw_exited, "missing review lifecycle events");
server.verify().await;
@@ -375,9 +402,10 @@ async fn review_uses_custom_review_model_from_config() {
codex
.submit(Op::Review {
review_request: ReviewRequest {
prompt: "use custom model".to_string(),
user_facing_hint: "use custom model".to_string(),
append_to_original_thread: true,
target: ReviewTarget::Custom {
instructions: "use custom model".to_string(),
},
user_facing_hint: None,
},
})
.await
@@ -493,9 +521,10 @@ async fn review_input_isolated_from_parent_history() {
codex
.submit(Op::Review {
review_request: ReviewRequest {
prompt: review_prompt.clone(),
user_facing_hint: review_prompt.clone(),
append_to_original_thread: true,
target: ReviewTarget::Custom {
instructions: review_prompt.clone(),
},
user_facing_hint: None,
},
})
.await
@@ -583,11 +612,10 @@ async fn review_input_isolated_from_parent_history() {
server.verify().await;
}
/// After a review thread finishes, its conversation should not leak into the
/// parent session. A subsequent parent turn must not include any review
/// messages in its request `input`.
/// After a review thread finishes, its conversation should be visible in the
/// parent session so later turns can reference the results.
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
async fn review_history_does_not_leak_into_parent_session() {
async fn review_history_surfaces_in_parent_session() {
skip_if_no_network!();
// Respond to both the review request and the subsequent parent request.
@@ -606,9 +634,10 @@ async fn review_history_does_not_leak_into_parent_session() {
codex
.submit(Op::Review {
review_request: ReviewRequest {
prompt: "Start a review".to_string(),
user_facing_hint: "Start a review".to_string(),
append_to_original_thread: true,
target: ReviewTarget::Custom {
instructions: "Start a review".to_string(),
},
user_facing_hint: None,
},
})
.await
@@ -651,20 +680,26 @@ async fn review_history_does_not_leak_into_parent_session() {
let last_text = last["content"][0]["text"].as_str().unwrap();
assert_eq!(last_text, followup);
// Ensure no review-thread content leaked into the parent request
let contains_review_prompt = input
.iter()
.any(|msg| msg["content"][0]["text"].as_str().unwrap_or_default() == "Start a review");
// Ensure review-thread content is present for downstream turns.
let contains_review_rollout_user = input.iter().any(|msg| {
msg["content"][0]["text"]
.as_str()
.unwrap_or_default()
.contains("User initiated a review task.")
});
let contains_review_assistant = input.iter().any(|msg| {
msg["content"][0]["text"].as_str().unwrap_or_default() == "review assistant output"
msg["content"][0]["text"]
.as_str()
.unwrap_or_default()
.contains("review assistant output")
});
assert!(
!contains_review_prompt,
"review prompt leaked into parent turn input"
contains_review_rollout_user,
"review rollout user message missing from parent turn input"
);
assert!(
!contains_review_assistant,
"review assistant output leaked into parent turn input"
contains_review_assistant,
"review assistant output missing from parent turn input"
);
server.verify().await;

View File

@@ -0,0 +1,457 @@
#![allow(clippy::expect_used)]
use anyhow::Result;
use codex_core::AuthManager;
use codex_core::CodexAuth;
use codex_core::CodexConversation;
use codex_core::ConversationManager;
use codex_core::SavedSessionEntry;
use codex_core::build_saved_session_entry;
use codex_core::config::Config;
use codex_core::protocol::EventMsg;
use codex_core::protocol::Op;
use codex_core::protocol::RolloutItem;
use codex_core::protocol::RolloutLine;
use codex_core::protocol::SaveSessionResponseEvent;
use codex_core::protocol::SessionSource;
use codex_core::resolve_saved_session;
use codex_core::upsert_saved_session;
use codex_protocol::user_input::UserInput;
use core_test_support::responses::ev_assistant_message;
use core_test_support::responses::ev_completed;
use core_test_support::responses::ev_response_created;
use core_test_support::responses::mount_sse_sequence;
use core_test_support::responses::sse;
use core_test_support::responses::start_mock_server;
use core_test_support::skip_if_no_network;
use core_test_support::test_codex::test_codex;
use core_test_support::wait_for_event;
use core_test_support::wait_for_event_match;
use pretty_assertions::assert_eq;
use serde_json::json;
use std::path::Path;
use std::sync::Arc;
fn completion_body(idx: usize, message: &str) -> String {
let resp_id = format!("resp-{idx}");
let msg_id = format!("msg-{idx}");
sse(vec![
ev_response_created(&resp_id),
ev_assistant_message(&msg_id, message),
ev_completed(&resp_id),
])
}
fn rollout_lines(path: &Path) -> Vec<RolloutLine> {
let text = std::fs::read_to_string(path).expect("read rollout");
text.lines()
.filter_map(|line| {
if line.trim().is_empty() {
return None;
}
let value: serde_json::Value = serde_json::from_str(line).expect("rollout line json");
Some(serde_json::from_value::<RolloutLine>(value).expect("rollout line"))
})
.collect()
}
fn rollout_items_without_meta(path: &Path) -> Vec<RolloutItem> {
rollout_lines(path)
.into_iter()
.filter_map(|line| match line.item {
RolloutItem::SessionMeta(_) => None,
other => Some(other),
})
.collect()
}
fn session_meta_count(path: &Path) -> usize {
rollout_lines(path)
.iter()
.filter(|line| matches!(line.item, RolloutItem::SessionMeta(_)))
.count()
}
async fn submit_text(codex: &Arc<CodexConversation>, text: &str) -> Result<()> {
codex
.submit(Op::UserInput {
items: vec![UserInput::Text {
text: text.to_string(),
}],
})
.await?;
let _ = wait_for_event(codex, |ev| matches!(ev, EventMsg::TaskComplete(_))).await;
Ok(())
}
async fn save_session(
name: &str,
codex: &Arc<CodexConversation>,
config: &Config,
) -> Result<SavedSessionEntry> {
codex.flush_rollout().await?;
codex.set_session_name(Some(name.to_string())).await?;
let entry =
build_saved_session_entry(name.to_string(), codex.rollout_path(), codex.model().await)
.await?;
upsert_saved_session(&config.codex_home, entry.clone()).await?;
Ok(entry)
}
async fn save_session_via_op(
codex: &Arc<CodexConversation>,
name: &str,
) -> Result<SaveSessionResponseEvent> {
codex
.submit(Op::SaveSession {
name: name.to_string(),
})
.await?;
let response: SaveSessionResponseEvent = wait_for_event_match(codex, |ev| match ev {
EventMsg::SaveSessionResponse(resp) => Some(resp.clone()),
_ => None,
})
.await;
Ok(response)
}
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
async fn save_and_resume_by_name() -> Result<()> {
skip_if_no_network!(Ok(()));
let server = start_mock_server().await;
mount_sse_sequence(&server, vec![completion_body(1, "initial")]).await;
let mut builder = test_codex();
let initial = builder.build(&server).await?;
submit_text(&initial.codex, "first turn").await?;
let name = "alpha";
let entry = save_session(name, &initial.codex, &initial.config).await?;
let resolved = resolve_saved_session(&initial.config.codex_home, name)
.await?
.expect("saved session");
assert_eq!(entry, resolved);
assert_eq!(session_meta_count(&entry.rollout_path), 1);
let saved_items = rollout_items_without_meta(&entry.rollout_path);
let resumed = builder
.resume(&server, initial.home.clone(), entry.rollout_path.clone())
.await?;
assert_eq!(resumed.session_configured.session_id, entry.conversation_id);
let resumed_items = rollout_items_without_meta(&resumed.session_configured.rollout_path);
assert_eq!(
serde_json::to_value(saved_items)?,
serde_json::to_value(resumed_items)?
);
Ok(())
}
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
async fn save_session_op_persists_and_emits_response() -> Result<()> {
skip_if_no_network!(Ok(()));
let server = start_mock_server().await;
mount_sse_sequence(&server, vec![completion_body(1, "initial")]).await;
let mut builder = test_codex();
let initial = builder.build(&server).await?;
submit_text(&initial.codex, "first turn").await?;
let name = "via-op";
let response = save_session_via_op(&initial.codex, name).await?;
assert_eq!(response.name, name);
assert_eq!(
response.conversation_id,
initial.session_configured.session_id
);
assert!(response.rollout_path.exists());
let resolved = resolve_saved_session(&initial.config.codex_home, name)
.await?
.expect("saved session");
assert_eq!(resolved.rollout_path, response.rollout_path);
assert_eq!(resolved.conversation_id, response.conversation_id);
assert_eq!(session_meta_count(&resolved.rollout_path), 1);
Ok(())
}
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
async fn fork_from_identifier_after_save_op() -> Result<()> {
skip_if_no_network!(Ok(()));
let server = start_mock_server().await;
mount_sse_sequence(
&server,
vec![
completion_body(1, "seed"),
completion_body(2, "fork-extra-1"),
completion_body(3, "fork-extra-2"),
],
)
.await;
let mut builder = test_codex();
let initial = builder.build(&server).await?;
submit_text(&initial.codex, "seeded").await?;
let name = "forkable-op";
let response = save_session_via_op(&initial.codex, name).await?;
let auth_manager = AuthManager::from_auth_for_testing(CodexAuth::from_api_key("dummy"));
let conversation_manager = ConversationManager::new(auth_manager.clone(), SessionSource::Exec);
let forked = conversation_manager
.fork_from_identifier(initial.config.clone(), name, auth_manager)
.await?;
assert_ne!(
forked.session_configured.session_id,
response.conversation_id
);
// Record the baseline rollout for the saved session.
let base_items = rollout_items_without_meta(&response.rollout_path);
// Send additional turns to the forked conversation and flush.
submit_text(&forked.conversation, "fork one").await?;
submit_text(&forked.conversation, "fork two").await?;
forked.conversation.flush_rollout().await?;
// Re-read both rollouts: source should remain unchanged.
let base_after = rollout_items_without_meta(&response.rollout_path);
assert_eq!(
serde_json::to_value(&base_items)?,
serde_json::to_value(&base_after)?
);
// Forked rollout should extend the baseline.
let fork_items = rollout_items_without_meta(&forked.conversation.rollout_path());
assert!(
fork_items.len() > base_items.len(),
"expected forked rollout to contain additional items"
);
let fork_prefix: Vec<_> = fork_items.iter().take(base_items.len()).cloned().collect();
assert_eq!(
serde_json::to_value(&base_items)?,
serde_json::to_value(&fork_prefix)?,
"forked rollout should extend the baseline history"
);
Ok(())
}
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
async fn save_and_fork_by_name() -> Result<()> {
skip_if_no_network!(Ok(()));
let server = start_mock_server().await;
mount_sse_sequence(&server, vec![completion_body(1, "base")]).await;
let mut builder = test_codex();
let initial = builder.build(&server).await?;
submit_text(&initial.codex, "original").await?;
let entry = save_session("forkable", &initial.codex, &initial.config).await?;
let auth_manager = AuthManager::from_auth_for_testing(CodexAuth::from_api_key("dummy"));
let conversation_manager = ConversationManager::new(auth_manager.clone(), SessionSource::Exec);
let forked = conversation_manager
.fork_from_rollout(
initial.config.clone(),
entry.rollout_path.clone(),
auth_manager,
)
.await?;
assert_ne!(forked.session_configured.session_id, entry.conversation_id);
assert_ne!(forked.conversation.rollout_path(), entry.rollout_path);
assert_eq!(session_meta_count(&forked.conversation.rollout_path()), 1);
let base_items = rollout_items_without_meta(&entry.rollout_path);
let fork_items = rollout_items_without_meta(&forked.conversation.rollout_path());
assert_eq!(
serde_json::to_value(base_items)?,
serde_json::to_value(fork_items)?
);
Ok(())
}
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
async fn forked_messages_do_not_touch_original() -> Result<()> {
skip_if_no_network!(Ok(()));
let server = start_mock_server().await;
mount_sse_sequence(
&server,
vec![
completion_body(1, "base"),
completion_body(2, "fork-1"),
completion_body(3, "fork-2"),
],
)
.await;
let mut builder = test_codex();
let initial = builder.build(&server).await?;
submit_text(&initial.codex, "first").await?;
let entry = save_session("branch", &initial.codex, &initial.config).await?;
let baseline_items = rollout_items_without_meta(&entry.rollout_path);
let auth_manager = AuthManager::from_auth_for_testing(CodexAuth::from_api_key("dummy"));
let conversation_manager = ConversationManager::new(auth_manager.clone(), SessionSource::Exec);
let forked = conversation_manager
.fork_from_rollout(
initial.config.clone(),
entry.rollout_path.clone(),
auth_manager.clone(),
)
.await?;
submit_text(&forked.conversation, "fork message one").await?;
submit_text(&forked.conversation, "fork message two").await?;
let resumed = builder
.resume(&server, initial.home.clone(), entry.rollout_path.clone())
.await?;
let resumed_items = rollout_items_without_meta(&resumed.session_configured.rollout_path);
assert_eq!(
serde_json::to_value(baseline_items.clone())?,
serde_json::to_value(resumed_items)?
);
assert_eq!(
serde_json::to_value(baseline_items)?,
serde_json::to_value(rollout_items_without_meta(&entry.rollout_path))?
);
Ok(())
}
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
async fn resumed_messages_are_present_in_new_fork() -> Result<()> {
skip_if_no_network!(Ok(()));
let server = start_mock_server().await;
mount_sse_sequence(
&server,
vec![
completion_body(1, "original"),
completion_body(2, "fork-extra"),
completion_body(3, "resumed-extra"),
],
)
.await;
let mut builder = test_codex();
let initial = builder.build(&server).await?;
submit_text(&initial.codex, "start").await?;
let entry = save_session("seed", &initial.codex, &initial.config).await?;
let auth_manager = AuthManager::from_auth_for_testing(CodexAuth::from_api_key("dummy"));
let conversation_manager = ConversationManager::new(auth_manager.clone(), SessionSource::Exec);
let forked = conversation_manager
.fork_from_rollout(
initial.config.clone(),
entry.rollout_path.clone(),
auth_manager.clone(),
)
.await?;
submit_text(&forked.conversation, "fork only").await?;
let resumed = builder
.resume(&server, initial.home.clone(), entry.rollout_path.clone())
.await?;
submit_text(&resumed.codex, "resumed addition").await?;
resumed.codex.flush_rollout().await?;
let updated_base_items = rollout_items_without_meta(&entry.rollout_path);
let fork_again = conversation_manager
.fork_from_rollout(
initial.config.clone(),
entry.rollout_path.clone(),
auth_manager,
)
.await?;
let fork_again_items = rollout_items_without_meta(&fork_again.conversation.rollout_path());
assert_eq!(
serde_json::to_value(updated_base_items)?,
serde_json::to_value(fork_again_items)?
);
assert_eq!(
session_meta_count(&fork_again.conversation.rollout_path()),
1
);
Ok(())
}
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
async fn duplicate_name_overwrites_entry() -> Result<()> {
skip_if_no_network!(Ok(()));
let server = start_mock_server().await;
mount_sse_sequence(
&server,
vec![completion_body(1, "one"), completion_body(2, "two")],
)
.await;
let mut builder = test_codex();
let first = builder.build(&server).await?;
submit_text(&first.codex, "first session").await?;
let name = "shared";
let entry_one = save_session(name, &first.codex, &first.config).await?;
let second = builder.build(&server).await?;
submit_text(&second.codex, "second session").await?;
let entry_two = save_session(name, &second.codex, &second.config).await?;
let resolved = resolve_saved_session(&second.config.codex_home, name)
.await?
.expect("latest entry present");
assert_eq!(resolved, entry_two);
assert_ne!(resolved.conversation_id, entry_one.conversation_id);
Ok(())
}
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
async fn same_session_multiple_names() -> Result<()> {
skip_if_no_network!(Ok(()));
let server = start_mock_server().await;
mount_sse_sequence(&server, vec![completion_body(1, "hello")]).await;
let mut builder = test_codex();
let session = builder.build(&server).await?;
submit_text(&session.codex, "save twice").await?;
let entry_first = save_session("first", &session.codex, &session.config).await?;
let entry_second = save_session("second", &session.codex, &session.config).await?;
let resolved_first = resolve_saved_session(&session.config.codex_home, "first")
.await?
.expect("first entry");
let resolved_second = resolve_saved_session(&session.config.codex_home, "second")
.await?
.expect("second entry");
assert_eq!(entry_first.conversation_id, entry_second.conversation_id);
assert_eq!(
resolved_first.conversation_id,
resolved_second.conversation_id
);
assert_eq!(resolved_first.rollout_path, resolved_second.rollout_path);
let names: serde_json::Value = json!([entry_first.name, entry_second.name]);
assert_eq!(names, json!(["first", "second"]));
Ok(())
}

View File

@@ -1,5 +1,7 @@
#![cfg(not(target_os = "windows"))]
use std::collections::HashMap;
use std::ffi::OsStr;
use std::fs;
use std::sync::OnceLock;
use anyhow::Context;
@@ -23,6 +25,7 @@ use core_test_support::responses::start_mock_server;
use core_test_support::skip_if_no_network;
use core_test_support::skip_if_sandbox;
use core_test_support::test_codex::TestCodex;
use core_test_support::test_codex::TestCodexHarness;
use core_test_support::test_codex::test_codex;
use core_test_support::wait_for_event;
use core_test_support::wait_for_event_match;
@@ -44,7 +47,7 @@ fn extract_output_text(item: &Value) -> Option<&str> {
struct ParsedUnifiedExecOutput {
chunk_id: Option<String>,
wall_time_seconds: f64,
session_id: Option<i32>,
process_id: Option<String>,
exit_code: Option<i32>,
original_token_count: Option<usize>,
output: String,
@@ -59,7 +62,7 @@ fn parse_unified_exec_output(raw: &str) -> Result<ParsedUnifiedExecOutput> {
r#"(?:Chunk ID: (?P<chunk_id>[^\n]+)\n)?"#,
r#"Wall time: (?P<wall_time>-?\d+(?:\.\d+)?) seconds\n"#,
r#"(?:Process exited with code (?P<exit_code>-?\d+)\n)?"#,
r#"(?:Process running with session ID (?P<session_id>-?\d+)\n)?"#,
r#"(?:Process running with session ID (?P<process_id>-?\d+)\n)?"#,
r#"(?:Original token count: (?P<original_token_count>\d+)\n)?"#,
r#"Output:\n?(?P<output>.*)$"#,
))
@@ -92,15 +95,9 @@ fn parse_unified_exec_output(raw: &str) -> Result<ParsedUnifiedExecOutput> {
})
.transpose()?;
let session_id = captures
.name("session_id")
.map(|value| {
value
.as_str()
.parse::<i32>()
.context("failed to parse session id from unified exec output")
})
.transpose()?;
let process_id = captures
.name("process_id")
.map(|value| value.as_str().to_string());
let original_token_count = captures
.name("original_token_count")
@@ -121,7 +118,7 @@ fn parse_unified_exec_output(raw: &str) -> Result<ParsedUnifiedExecOutput> {
Ok(ParsedUnifiedExecOutput {
chunk_id,
wall_time_seconds,
session_id,
process_id,
exit_code,
original_token_count,
output,
@@ -154,6 +151,130 @@ fn collect_tool_outputs(bodies: &[Value]) -> Result<HashMap<String, ParsedUnifie
Ok(outputs)
}
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
async fn unified_exec_intercepts_apply_patch_exec_command() -> Result<()> {
skip_if_no_network!(Ok(()));
skip_if_sandbox!(Ok(()));
let builder = test_codex().with_config(|config| {
config.include_apply_patch_tool = true;
config.use_experimental_unified_exec_tool = true;
config.features.enable(Feature::UnifiedExec);
});
let harness = TestCodexHarness::with_builder(builder).await?;
let patch =
"*** Begin Patch\n*** Add File: uexec_apply.txt\n+hello from unified exec\n*** End Patch";
let command = format!("apply_patch <<'EOF'\n{patch}\nEOF\n");
let call_id = "uexec-apply-patch";
let args = json!({
"cmd": command,
"yield_time_ms": 250,
});
let responses = vec![
sse(vec![
ev_response_created("resp-1"),
ev_function_call(call_id, "exec_command", &serde_json::to_string(&args)?),
ev_completed("resp-1"),
]),
sse(vec![
ev_response_created("resp-2"),
ev_assistant_message("msg-1", "done"),
ev_completed("resp-2"),
]),
];
mount_sse_sequence(harness.server(), responses).await;
let test = harness.test();
let codex = test.codex.clone();
let cwd = test.cwd_path().to_path_buf();
let session_model = test.session_configured.model.clone();
codex
.submit(Op::UserTurn {
items: vec![UserInput::Text {
text: "apply patch via unified exec".into(),
}],
final_output_json_schema: None,
cwd,
approval_policy: AskForApproval::Never,
sandbox_policy: SandboxPolicy::DangerFullAccess,
model: session_model,
effort: None,
summary: ReasoningSummary::Auto,
})
.await?;
let mut saw_patch_begin = false;
let mut patch_end = None;
let mut saw_exec_begin = false;
let mut saw_exec_end = false;
wait_for_event(&codex, |event| match event {
EventMsg::PatchApplyBegin(begin) if begin.call_id == call_id => {
saw_patch_begin = true;
assert!(
begin
.changes
.keys()
.any(|path| path.file_name() == Some(OsStr::new("uexec_apply.txt"))),
"expected apply_patch changes to target uexec_apply.txt",
);
false
}
EventMsg::PatchApplyEnd(end) if end.call_id == call_id => {
patch_end = Some(end.clone());
false
}
EventMsg::ExecCommandBegin(event) if event.call_id == call_id => {
saw_exec_begin = true;
false
}
EventMsg::ExecCommandEnd(event) if event.call_id == call_id => {
saw_exec_end = true;
false
}
EventMsg::TaskComplete(_) => true,
_ => false,
})
.await;
assert!(
saw_patch_begin,
"expected apply_patch to emit PatchApplyBegin"
);
let patch_end = patch_end.expect("expected apply_patch to emit PatchApplyEnd");
assert!(
patch_end.success,
"expected apply_patch to finish successfully: stdout={:?} stderr={:?}",
patch_end.stdout, patch_end.stderr,
);
assert!(
!saw_exec_begin,
"apply_patch should be intercepted before exec_command begin"
);
assert!(
!saw_exec_end,
"apply_patch should not emit exec_command end events"
);
let output = harness.function_call_stdout(call_id).await;
assert!(
output.contains("Success. Updated the following files:"),
"expected apply_patch output, got: {output:?}"
);
assert!(
output.contains("A uexec_apply.txt"),
"expected apply_patch file summary, got: {output:?}"
);
assert_eq!(
fs::read_to_string(harness.path("uexec_apply.txt"))?,
"hello from unified exec\n"
);
Ok(())
}
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
async fn unified_exec_emits_exec_command_begin_event() -> Result<()> {
skip_if_no_network!(Ok(()));
@@ -174,6 +295,7 @@ async fn unified_exec_emits_exec_command_begin_event() -> Result<()> {
let call_id = "uexec-begin-event";
let args = json!({
"shell": "bash".to_string(),
"cmd": "/bin/echo hello unified exec".to_string(),
"yield_time_ms": 250,
});
@@ -215,14 +337,8 @@ async fn unified_exec_emits_exec_command_begin_event() -> Result<()> {
})
.await;
assert_eq!(
begin_event.command,
vec![
"/bin/bash".to_string(),
"-lc".to_string(),
"/bin/echo hello unified exec".to_string()
]
);
assert_command(&begin_event.command, "-lc", "/bin/echo hello unified exec");
assert_eq!(begin_event.cwd, cwd.path());
wait_for_event(&codex, |event| matches!(event, EventMsg::TaskComplete(_))).await;
@@ -230,6 +346,82 @@ async fn unified_exec_emits_exec_command_begin_event() -> Result<()> {
Ok(())
}
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
async fn unified_exec_resolves_relative_workdir() -> Result<()> {
skip_if_no_network!(Ok(()));
skip_if_sandbox!(Ok(()));
let server = start_mock_server().await;
let mut builder = test_codex().with_model("gpt-5").with_config(|config| {
config.use_experimental_unified_exec_tool = true;
config.features.enable(Feature::UnifiedExec);
});
let TestCodex {
codex,
cwd,
session_configured,
..
} = builder.build(&server).await?;
let workdir_rel = std::path::PathBuf::from("uexec_relative_workdir");
std::fs::create_dir_all(cwd.path().join(&workdir_rel))?;
let call_id = "uexec-workdir-relative";
let args = json!({
"cmd": "pwd",
"yield_time_ms": 250,
"workdir": workdir_rel.to_string_lossy().to_string(),
});
let responses = vec![
sse(vec![
ev_response_created("resp-1"),
ev_function_call(call_id, "exec_command", &serde_json::to_string(&args)?),
ev_completed("resp-1"),
]),
sse(vec![
ev_response_created("resp-2"),
ev_assistant_message("msg-1", "finished"),
ev_completed("resp-2"),
]),
];
mount_sse_sequence(&server, responses).await;
let session_model = session_configured.model.clone();
codex
.submit(Op::UserTurn {
items: vec![UserInput::Text {
text: "run relative workdir test".into(),
}],
final_output_json_schema: None,
cwd: cwd.path().to_path_buf(),
approval_policy: AskForApproval::Never,
sandbox_policy: SandboxPolicy::DangerFullAccess,
model: session_model,
effort: None,
summary: ReasoningSummary::Auto,
})
.await?;
let begin_event = wait_for_event_match(&codex, |msg| match msg {
EventMsg::ExecCommandBegin(event) if event.call_id == call_id => Some(event.clone()),
_ => None,
})
.await;
assert_eq!(
begin_event.cwd,
cwd.path().join(workdir_rel),
"exec_command cwd should resolve relative workdir against turn cwd",
);
wait_for_event(&codex, |event| matches!(event, EventMsg::TaskComplete(_))).await;
Ok(())
}
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
#[ignore = "flaky"]
async fn unified_exec_respects_workdir_override() -> Result<()> {
@@ -335,7 +527,7 @@ async fn unified_exec_emits_exec_command_end_event() -> Result<()> {
let poll_call_id = "uexec-end-event-poll";
let poll_args = json!({
"chars": "",
"session_id": 0,
"session_id": 1000,
"yield_time_ms": 250,
});
@@ -493,7 +685,7 @@ async fn unified_exec_emits_output_delta_for_write_stdin() -> Result<()> {
let stdin_call_id = "uexec-stdin-delta";
let stdin_args = json!({
"chars": "echo WSTDIN-MARK\\n",
"session_id": 0,
"session_id": 1000,
"yield_time_ms": 800,
});
@@ -585,6 +777,7 @@ async fn unified_exec_emits_begin_for_write_stdin() -> Result<()> {
let open_call_id = "uexec-open-for-begin";
let open_args = json!({
"shell": "bash".to_string(),
"cmd": "bash -i".to_string(),
"yield_time_ms": 200,
});
@@ -592,7 +785,7 @@ async fn unified_exec_emits_begin_for_write_stdin() -> Result<()> {
let stdin_call_id = "uexec-stdin-begin";
let stdin_args = json!({
"chars": "echo hello",
"session_id": 0,
"session_id": 1000,
"yield_time_ms": 400,
});
@@ -646,14 +839,7 @@ async fn unified_exec_emits_begin_for_write_stdin() -> Result<()> {
})
.await;
assert_eq!(
begin_event.command,
vec![
"/bin/bash".to_string(),
"-lc".to_string(),
"bash -i".to_string()
]
);
assert_command(&begin_event.command, "-lc", "bash -i");
assert_eq!(
begin_event.interaction_input,
Some("echo hello".to_string())
@@ -687,6 +873,7 @@ async fn unified_exec_emits_begin_event_for_write_stdin_requests() -> Result<()>
let open_call_id = "uexec-open-session";
let open_args = json!({
"shell": "bash".to_string(),
"cmd": "bash -i".to_string(),
"yield_time_ms": 250,
});
@@ -694,7 +881,7 @@ async fn unified_exec_emits_begin_event_for_write_stdin_requests() -> Result<()>
let poll_call_id = "uexec-poll-empty";
let poll_args = json!({
"chars": "",
"session_id": 0,
"session_id": 1000,
"yield_time_ms": 150,
});
@@ -762,14 +949,9 @@ async fn unified_exec_emits_begin_event_for_write_stdin_requests() -> Result<()>
.iter()
.find(|ev| ev.call_id == open_call_id)
.expect("missing exec_command begin");
assert_eq!(
open_event.command,
vec![
"/bin/bash".to_string(),
"-lc".to_string(),
"bash -i".to_string()
]
);
assert_command(&open_event.command, "-lc", "bash -i");
assert!(
open_event.interaction_input.is_none(),
"startup begin events should not include interaction input"
@@ -780,14 +962,9 @@ async fn unified_exec_emits_begin_event_for_write_stdin_requests() -> Result<()>
.iter()
.find(|ev| ev.call_id == poll_call_id)
.expect("missing write_stdin begin");
assert_eq!(
poll_event.command,
vec![
"/bin/bash".to_string(),
"-lc".to_string(),
"bash -i".to_string()
]
);
assert_command(&poll_event.command, "-lc", "bash -i");
assert!(
poll_event.interaction_input.is_none(),
"poll begin events should omit interaction input"
@@ -880,8 +1057,8 @@ async fn exec_command_reports_chunk_and_exit_metadata() -> Result<()> {
);
assert!(
metadata.session_id.is_none(),
"exec_command for a completed process should not include session_id"
metadata.process_id.is_none(),
"exec_command for a completed process should not include process_id"
);
let exit_code = metadata.exit_code.expect("expected exit_code");
@@ -973,7 +1150,7 @@ async fn unified_exec_respects_early_exit_notifications() -> Result<()> {
.expect("missing early exit unified_exec output");
assert!(
output.session_id.is_none(),
output.process_id.is_none(),
"short-lived process should not keep a session alive"
);
assert_eq!(
@@ -1023,12 +1200,12 @@ async fn write_stdin_returns_exit_metadata_and_clears_session() -> Result<()> {
});
let send_args = serde_json::json!({
"chars": "hello unified exec\n",
"session_id": 0,
"session_id": 1000,
"yield_time_ms": 500,
});
let exit_args = serde_json::json!({
"chars": "\u{0004}",
"session_id": 0,
"session_id": 1000,
"yield_time_ms": 500,
});
@@ -1099,12 +1276,13 @@ async fn write_stdin_returns_exit_metadata_and_clears_session() -> Result<()> {
let start_output = outputs
.get(start_call_id)
.expect("missing start output for exec_command");
let session_id = start_output
.session_id
.expect("expected session id from exec_command");
let process_id = start_output
.process_id
.clone()
.expect("expected process id from exec_command");
assert!(
session_id >= 0,
"session_id should be non-negative, got {session_id}"
process_id.len() > 3,
"process_id should be at least 4 digits, got {process_id}"
);
assert!(
start_output.exit_code.is_none(),
@@ -1120,11 +1298,12 @@ async fn write_stdin_returns_exit_metadata_and_clears_session() -> Result<()> {
"expected echoed output from cat, got {echoed:?}"
);
let echoed_session = send_output
.session_id
.expect("write_stdin should return session id while process is running");
.process_id
.clone()
.expect("write_stdin should return process id while process is running");
assert_eq!(
echoed_session, session_id,
"write_stdin should reuse existing session id"
echoed_session, process_id,
"write_stdin should reuse existing process id"
);
assert!(
send_output.exit_code.is_none(),
@@ -1135,8 +1314,8 @@ async fn write_stdin_returns_exit_metadata_and_clears_session() -> Result<()> {
.get(exit_call_id)
.expect("missing exit metadata output");
assert!(
exit_output.session_id.is_none(),
"session_id should be omitted once the process exits"
exit_output.process_id.is_none(),
"process_id should be omitted once the process exits"
);
let exit_code = exit_output
.exit_code
@@ -1182,14 +1361,14 @@ async fn unified_exec_emits_end_event_when_session_dies_via_stdin() -> Result<()
let echo_call_id = "uexec-end-on-exit-echo";
let echo_args = serde_json::json!({
"chars": "bye-END\n",
"session_id": 0,
"session_id": 1000,
"yield_time_ms": 300,
});
let exit_call_id = "uexec-end-on-exit";
let exit_args = serde_json::json!({
"chars": "\u{0004}",
"session_id": 0,
"session_id": 1000,
"yield_time_ms": 500,
});
@@ -1285,7 +1464,7 @@ async fn unified_exec_reuses_session_via_stdin() -> Result<()> {
let second_call_id = "uexec-stdin";
let second_args = serde_json::json!({
"chars": "hello unified exec\n",
"session_id": 0,
"session_id": 1000,
"yield_time_ms": 500,
});
@@ -1347,17 +1526,20 @@ async fn unified_exec_reuses_session_via_stdin() -> Result<()> {
let start_output = outputs
.get(first_call_id)
.expect("missing first unified_exec output");
let session_id = start_output.session_id.unwrap_or_default();
let process_id = start_output.process_id.clone().unwrap_or_default();
assert!(
session_id >= 0,
"expected session id in first unified_exec response"
!process_id.is_empty(),
"expected process id in first unified_exec response"
);
assert!(start_output.output.is_empty());
let reuse_output = outputs
.get(second_call_id)
.expect("missing reused unified_exec output");
assert_eq!(reuse_output.session_id.unwrap_or_default(), session_id);
assert_eq!(
reuse_output.process_id.clone().unwrap_or_default(),
process_id
);
let echoed = reuse_output.output.as_str();
assert!(
echoed.contains("hello unified exec"),
@@ -1413,7 +1595,7 @@ PY
let second_call_id = "uexec-lag-poll";
let second_args = serde_json::json!({
"chars": "",
"session_id": 0,
"session_id": 1000,
"yield_time_ms": 2_000,
});
@@ -1480,9 +1662,9 @@ PY
let start_output = outputs
.get(first_call_id)
.expect("missing initial unified_exec output");
let session_id = start_output.session_id.unwrap_or_default();
let process_id = start_output.process_id.clone().unwrap_or_default();
assert!(
session_id >= 0,
!process_id.is_empty(),
"expected session id from initial unified_exec response"
);
@@ -1524,7 +1706,7 @@ async fn unified_exec_timeout_and_followup_poll() -> Result<()> {
let second_call_id = "uexec-poll";
let second_args = serde_json::json!({
"chars": "",
"session_id": 0,
"session_id": 1000,
"yield_time_ms": 800,
});
@@ -1589,7 +1771,7 @@ async fn unified_exec_timeout_and_followup_poll() -> Result<()> {
let outputs = collect_tool_outputs(&bodies)?;
let first_output = outputs.get(first_call_id).expect("missing timeout output");
assert_eq!(first_output.session_id, Some(0));
assert!(first_output.process_id.is_some());
assert!(first_output.output.is_empty());
let poll_output = outputs.get(second_call_id).expect("missing poll output");
@@ -1824,7 +2006,7 @@ async fn unified_exec_prunes_exited_sessions_first() -> Result<()> {
let keep_write_call_id = "uexec-prune-keep-write";
let keep_write_args = serde_json::json!({
"chars": "still alive\n",
"session_id": 0,
"session_id": 1000,
"yield_time_ms": 500,
});
events.push(ev_function_call(
@@ -1836,7 +2018,7 @@ async fn unified_exec_prunes_exited_sessions_first() -> Result<()> {
let probe_call_id = "uexec-prune-probe";
let probe_args = serde_json::json!({
"chars": "should fail\n",
"session_id": 1,
"session_id": 1001,
"yield_time_ms": 500,
});
events.push(ev_function_call(
@@ -1885,7 +2067,7 @@ async fn unified_exec_prunes_exited_sessions_first() -> Result<()> {
.find_map(|req| req.function_call_output_text(keep_call_id))
.expect("missing initial keep session output");
let keep_start_output = parse_unified_exec_output(&keep_start)?;
pretty_assertions::assert_eq!(keep_start_output.session_id, Some(0));
assert!(keep_start_output.process_id.is_some());
assert!(keep_start_output.exit_code.is_none());
let prune_start = requests
@@ -1893,7 +2075,7 @@ async fn unified_exec_prunes_exited_sessions_first() -> Result<()> {
.find_map(|req| req.function_call_output_text(prune_call_id))
.expect("missing initial prune session output");
let prune_start_output = parse_unified_exec_output(&prune_start)?;
pretty_assertions::assert_eq!(prune_start_output.session_id, Some(1));
assert!(prune_start_output.process_id.is_some());
assert!(prune_start_output.exit_code.is_none());
let keep_write = requests
@@ -1901,7 +2083,7 @@ async fn unified_exec_prunes_exited_sessions_first() -> Result<()> {
.find_map(|req| req.function_call_output_text(keep_write_call_id))
.expect("missing keep write output");
let keep_write_output = parse_unified_exec_output(&keep_write)?;
pretty_assertions::assert_eq!(keep_write_output.session_id, Some(0));
assert!(keep_write_output.process_id.is_some());
assert!(
keep_write_output.output.contains("still alive"),
"expected cat session to echo input, got {:?}",
@@ -1913,9 +2095,23 @@ async fn unified_exec_prunes_exited_sessions_first() -> Result<()> {
.find_map(|req| req.function_call_output_text(probe_call_id))
.expect("missing probe output");
assert!(
pruned_probe.contains("UnknownSessionId") || pruned_probe.contains("Unknown session id"),
pruned_probe.contains("UnknownSessionId") || pruned_probe.contains("Unknown process id"),
"expected probe to fail after pruning, got {pruned_probe:?}"
);
Ok(())
}
fn assert_command(command: &[String], expected_args: &str, expected_cmd: &str) {
assert_eq!(command.len(), 3);
let shell_path = &command[0];
assert!(
shell_path == "/bin/bash"
|| shell_path == "/usr/bin/bash"
|| shell_path == "/usr/local/bin/bash"
|| shell_path.ends_with("/bash"),
"unexpected bash path: {shell_path}"
);
assert_eq!(command[1], expected_args);
assert_eq!(command[2], expected_cmd);
}

View File

@@ -474,3 +474,82 @@ async fn view_image_tool_errors_when_file_missing() -> anyhow::Result<()> {
Ok(())
}
#[cfg(not(debug_assertions))]
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
async fn replaces_invalid_local_image_after_bad_request() -> anyhow::Result<()> {
skip_if_no_network!(Ok(()));
let server = start_mock_server().await;
const INVALID_IMAGE_ERROR: &str =
"The image data you provided does not represent a valid image";
let invalid_image_mock = responses::mount_response_once_match(
&server,
body_string_contains("\"input_image\""),
ResponseTemplate::new(400)
.insert_header("content-type", "text/plain")
.set_body_string(INVALID_IMAGE_ERROR),
)
.await;
let success_response = sse(vec![
ev_response_created("resp-2"),
ev_assistant_message("msg-1", "done"),
ev_completed("resp-2"),
]);
let completion_mock = responses::mount_sse_once(&server, success_response).await;
let TestCodex {
codex,
cwd,
session_configured,
..
} = test_codex().build(&server).await?;
let rel_path = "assets/poisoned.png";
let abs_path = cwd.path().join(rel_path);
if let Some(parent) = abs_path.parent() {
std::fs::create_dir_all(parent)?;
}
let image = ImageBuffer::from_pixel(1024, 512, Rgba([10u8, 20, 30, 255]));
image.save(&abs_path)?;
let session_model = session_configured.model.clone();
codex
.submit(Op::UserTurn {
items: vec![UserInput::LocalImage {
path: abs_path.clone(),
}],
final_output_json_schema: None,
cwd: cwd.path().to_path_buf(),
approval_policy: AskForApproval::Never,
sandbox_policy: SandboxPolicy::DangerFullAccess,
model: session_model,
effort: None,
summary: ReasoningSummary::Auto,
})
.await?;
wait_for_event(&codex, |event| matches!(event, EventMsg::TaskComplete(_))).await;
let first_body = invalid_image_mock.single_request().body_json();
assert!(
find_image_message(&first_body).is_some(),
"initial request should include the uploaded image"
);
let second_request = completion_mock.single_request();
let second_body = second_request.body_json();
assert!(
find_image_message(&second_body).is_none(),
"second request should replace the invalid image"
);
let user_texts = second_request.message_input_texts("user");
assert!(user_texts.iter().any(|text| text == "Invalid image"));
Ok(())
}

View File

@@ -22,6 +22,7 @@ rustPlatform.buildRustPackage (_: {
cargoLock.outputHashes = {
"ratatui-0.29.0" = "sha256-HBvT5c8GsiCxMffNjJGLmHnvG77A6cqEL+1ARurBXho=";
"crossterm-0.28.1" = "sha256-6qCtfSMuXACKFb9ATID39XyFDIEMFDmbx6SSmNe+728=";
"rmcp-0.9.0" = "sha256-0iPrpf0Ha/facO3p5e0hUKHBqGp/iS+C+OdS+pRKMOU=";
};
meta = with lib; {

View File

@@ -78,6 +78,7 @@ mod stopwatch;
const CODEX_EXECVE_WRAPPER_EXE_NAME: &str = "codex-execve-wrapper";
#[derive(Parser)]
#[clap(version)]
struct McpServerCli {
/// Executable to delegate execve(2) calls to in Bash.
#[arg(long = "execve")]

View File

@@ -91,6 +91,9 @@ pub struct Cli {
pub enum Command {
/// Resume a previous session by id or pick the most recent with --last.
Resume(ResumeArgs),
/// Run a code review against the current repository.
Review(ReviewArgs),
}
#[derive(Parser, Debug)]
@@ -109,6 +112,41 @@ pub struct ResumeArgs {
pub prompt: Option<String>,
}
#[derive(Parser, Debug)]
pub struct ReviewArgs {
/// Review staged, unstaged, and untracked changes.
#[arg(
long = "uncommitted",
default_value_t = false,
conflicts_with_all = ["base", "commit", "prompt"]
)]
pub uncommitted: bool,
/// Review changes against the given base branch.
#[arg(
long = "base",
value_name = "BRANCH",
conflicts_with_all = ["uncommitted", "commit", "prompt"]
)]
pub base: Option<String>,
/// Review the changes introduced by a commit.
#[arg(
long = "commit",
value_name = "SHA",
conflicts_with_all = ["uncommitted", "base", "prompt"]
)]
pub commit: Option<String>,
/// Optional commit title to display in the review summary.
#[arg(long = "title", value_name = "TITLE", requires = "commit")]
pub commit_title: Option<String>,
/// Custom review instructions. If `-` is used, read from stdin.
#[arg(value_name = "PROMPT", value_hint = clap::ValueHint::Other)]
pub prompt: Option<String>,
}
#[derive(Debug, Clone, Copy, Default, PartialEq, Eq, ValueEnum)]
#[value(rename_all = "kebab-case")]
pub enum Color {

View File

@@ -583,6 +583,7 @@ impl EventProcessor for EventProcessorWithHumanOutput {
| EventMsg::ReasoningContentDelta(_)
| EventMsg::ReasoningRawContentDelta(_)
| EventMsg::UndoCompleted(_)
| EventMsg::SaveSessionResponse(_)
| EventMsg::UndoStarted(_) => {}
}
CodexStatus::Running

View File

@@ -11,6 +11,8 @@ pub mod event_processor_with_jsonl_output;
pub mod exec_events;
pub use cli::Cli;
pub use cli::Command;
pub use cli::ReviewArgs;
use codex_common::oss::ensure_oss_provider_ready;
use codex_common::oss::get_default_model_for_oss_provider;
use codex_core::AuthManager;
@@ -29,6 +31,8 @@ use codex_core::protocol::AskForApproval;
use codex_core::protocol::Event;
use codex_core::protocol::EventMsg;
use codex_core::protocol::Op;
use codex_core::protocol::ReviewRequest;
use codex_core::protocol::ReviewTarget;
use codex_core::protocol::SessionSource;
use codex_protocol::approvals::ElicitationAction;
use codex_protocol::config_types::SandboxMode;
@@ -53,6 +57,16 @@ use crate::event_processor::EventProcessor;
use codex_core::default_client::set_default_originator;
use codex_core::find_conversation_path_by_id_str;
enum InitialOperation {
UserTurn {
items: Vec<UserInput>,
output_schema: Option<Value>,
},
Review {
review_request: ReviewRequest,
},
}
pub async fn run_main(cli: Cli, codex_linux_sandbox_exe: Option<PathBuf>) -> anyhow::Result<()> {
if let Err(err) = set_default_originator("codex_exec".to_string()) {
tracing::warn!(?err, "Failed to set codex exec originator override {err:?}");
@@ -79,64 +93,6 @@ pub async fn run_main(cli: Cli, codex_linux_sandbox_exe: Option<PathBuf>) -> any
config_overrides,
} = cli;
// Determine the prompt source (parent or subcommand) and read from stdin if needed.
let prompt_arg = match &command {
// Allow prompt before the subcommand by falling back to the parent-level prompt
// when the Resume subcommand did not provide its own prompt.
Some(ExecCommand::Resume(args)) => {
let resume_prompt = args
.prompt
.clone()
// When using `resume --last <PROMPT>`, clap still parses the first positional
// as `session_id`. Reinterpret it as the prompt so the flag works with JSON mode.
.or_else(|| {
if args.last {
args.session_id.clone()
} else {
None
}
});
resume_prompt.or(prompt)
}
None => prompt,
};
let prompt = match prompt_arg {
Some(p) if p != "-" => p,
// Either `-` was passed or no positional arg.
maybe_dash => {
// When no arg (None) **and** stdin is a TTY, bail out early unless the
// user explicitly forced reading via `-`.
let force_stdin = matches!(maybe_dash.as_deref(), Some("-"));
if std::io::stdin().is_terminal() && !force_stdin {
eprintln!(
"No prompt provided. Either specify one as an argument or pipe the prompt into stdin."
);
std::process::exit(1);
}
// Ensure the user knows we are waiting on stdin, as they may
// have gotten into this state by mistake. If so, and they are not
// writing to stdin, Codex will hang indefinitely, so this should
// help them debug in that case.
if !force_stdin {
eprintln!("Reading prompt from stdin...");
}
let mut buffer = String::new();
if let Err(e) = std::io::stdin().read_to_string(&mut buffer) {
eprintln!("Failed to read prompt from stdin: {e}");
std::process::exit(1);
} else if buffer.trim().is_empty() {
eprintln!("No prompt provided via stdin.");
std::process::exit(1);
}
buffer
}
};
let output_schema = load_output_schema(output_schema_path);
let (stdout_with_ansi, stderr_with_ansi) = match color {
cli::Color::Always => (true, true),
cli::Color::Never => (false, false),
@@ -329,8 +285,8 @@ pub async fn run_main(cli: Cli, codex_linux_sandbox_exe: Option<PathBuf>) -> any
conversation_id: _,
conversation,
session_configured,
} = if let Some(ExecCommand::Resume(args)) = command {
let resume_path = resolve_resume_path(&config, &args).await?;
} = if let Some(ExecCommand::Resume(args)) = command.as_ref() {
let resume_path = resolve_resume_path(&config, args).await?;
if let Some(path) = resume_path {
conversation_manager
@@ -346,9 +302,64 @@ pub async fn run_main(cli: Cli, codex_linux_sandbox_exe: Option<PathBuf>) -> any
.new_conversation(config.clone())
.await?
};
// Print the effective configuration and prompt so users can see what Codex
let (initial_operation, prompt_summary) = match (command, prompt, images) {
(Some(ExecCommand::Review(review_cli)), _, _) => {
let review_request = build_review_request(review_cli)?;
let summary = codex_core::review_prompts::user_facing_hint(&review_request.target);
(InitialOperation::Review { review_request }, summary)
}
(Some(ExecCommand::Resume(args)), root_prompt, imgs) => {
let prompt_arg = args
.prompt
.clone()
.or_else(|| {
if args.last {
args.session_id.clone()
} else {
None
}
})
.or(root_prompt);
let prompt_text = resolve_prompt(prompt_arg);
let mut items: Vec<UserInput> = imgs
.into_iter()
.map(|path| UserInput::LocalImage { path })
.collect();
items.push(UserInput::Text {
text: prompt_text.clone(),
});
let output_schema = load_output_schema(output_schema_path.clone());
(
InitialOperation::UserTurn {
items,
output_schema,
},
prompt_text,
)
}
(None, root_prompt, imgs) => {
let prompt_text = resolve_prompt(root_prompt);
let mut items: Vec<UserInput> = imgs
.into_iter()
.map(|path| UserInput::LocalImage { path })
.collect();
items.push(UserInput::Text {
text: prompt_text.clone(),
});
let output_schema = load_output_schema(output_schema_path);
(
InitialOperation::UserTurn {
items,
output_schema,
},
prompt_text,
)
}
};
// Print the effective configuration and initial request so users can see what Codex
// is using.
event_processor.print_config_summary(&config, &prompt, &session_configured);
event_processor.print_config_summary(&config, &prompt_summary, &session_configured);
info!("Codex initialized with event: {session_configured:?}");
@@ -391,25 +402,32 @@ pub async fn run_main(cli: Cli, codex_linux_sandbox_exe: Option<PathBuf>) -> any
});
}
// Package images and prompt into a single user input turn.
let mut items: Vec<UserInput> = images
.into_iter()
.map(|path| UserInput::LocalImage { path })
.collect();
items.push(UserInput::Text { text: prompt });
let initial_prompt_task_id = conversation
.submit(Op::UserTurn {
match initial_operation {
InitialOperation::UserTurn {
items,
cwd: default_cwd,
approval_policy: default_approval_policy,
sandbox_policy: default_sandbox_policy,
model: default_model,
effort: default_effort,
summary: default_summary,
final_output_json_schema: output_schema,
})
.await?;
info!("Sent prompt with event ID: {initial_prompt_task_id}");
output_schema,
} => {
let task_id = conversation
.submit(Op::UserTurn {
items,
cwd: default_cwd,
approval_policy: default_approval_policy,
sandbox_policy: default_sandbox_policy,
model: default_model,
effort: default_effort,
summary: default_summary,
final_output_json_schema: output_schema,
})
.await?;
info!("Sent prompt with event ID: {task_id}");
task_id
}
InitialOperation::Review { review_request } => {
let task_id = conversation.submit(Op::Review { review_request }).await?;
info!("Sent review request with event ID: {task_id}");
task_id
}
};
// Run the loop until the task is complete.
// Track whether a fatal error was reported by the server so we can
@@ -503,3 +521,130 @@ fn load_output_schema(path: Option<PathBuf>) -> Option<Value> {
}
}
}
fn resolve_prompt(prompt_arg: Option<String>) -> String {
match prompt_arg {
Some(p) if p != "-" => p,
maybe_dash => {
let force_stdin = matches!(maybe_dash.as_deref(), Some("-"));
if std::io::stdin().is_terminal() && !force_stdin {
eprintln!(
"No prompt provided. Either specify one as an argument or pipe the prompt into stdin."
);
std::process::exit(1);
}
if !force_stdin {
eprintln!("Reading prompt from stdin...");
}
let mut buffer = String::new();
if let Err(e) = std::io::stdin().read_to_string(&mut buffer) {
eprintln!("Failed to read prompt from stdin: {e}");
std::process::exit(1);
} else if buffer.trim().is_empty() {
eprintln!("No prompt provided via stdin.");
std::process::exit(1);
}
buffer
}
}
}
fn build_review_request(args: ReviewArgs) -> anyhow::Result<ReviewRequest> {
let target = if args.uncommitted {
ReviewTarget::UncommittedChanges
} else if let Some(branch) = args.base {
ReviewTarget::BaseBranch { branch }
} else if let Some(sha) = args.commit {
ReviewTarget::Commit {
sha,
title: args.commit_title,
}
} else if let Some(prompt_arg) = args.prompt {
let prompt = resolve_prompt(Some(prompt_arg)).trim().to_string();
if prompt.is_empty() {
anyhow::bail!("Review prompt cannot be empty");
}
ReviewTarget::Custom {
instructions: prompt,
}
} else {
anyhow::bail!(
"Specify --uncommitted, --base, --commit, or provide custom review instructions"
);
};
Ok(ReviewRequest {
target,
user_facing_hint: None,
})
}
#[cfg(test)]
mod tests {
use super::*;
use pretty_assertions::assert_eq;
#[test]
fn builds_uncommitted_review_request() {
let request = build_review_request(ReviewArgs {
uncommitted: true,
base: None,
commit: None,
commit_title: None,
prompt: None,
})
.expect("builds uncommitted review request");
let expected = ReviewRequest {
target: ReviewTarget::UncommittedChanges,
user_facing_hint: None,
};
assert_eq!(request, expected);
}
#[test]
fn builds_commit_review_request_with_title() {
let request = build_review_request(ReviewArgs {
uncommitted: false,
base: None,
commit: Some("123456789".to_string()),
commit_title: Some("Add review command".to_string()),
prompt: None,
})
.expect("builds commit review request");
let expected = ReviewRequest {
target: ReviewTarget::Commit {
sha: "123456789".to_string(),
title: Some("Add review command".to_string()),
},
user_facing_hint: None,
};
assert_eq!(request, expected);
}
#[test]
fn builds_custom_review_request_trims_prompt() {
let request = build_review_request(ReviewArgs {
uncommitted: false,
base: None,
commit: None,
commit_title: None,
prompt: Some(" custom review instructions ".to_string()),
})
.expect("builds custom review request");
let expected = ReviewRequest {
target: ReviewTarget::Custom {
instructions: "custom review instructions".to_string(),
},
user_facing_hint: None,
};
assert_eq!(request, expected);
}
}

View File

@@ -637,6 +637,7 @@ fn exec_command_end_success_produces_completed_command_item() {
"c1",
EventMsg::ExecCommandBegin(ExecCommandBeginEvent {
call_id: "1".to_string(),
process_id: None,
turn_id: "turn-1".to_string(),
command: command.clone(),
cwd: cwd.clone(),
@@ -666,6 +667,7 @@ fn exec_command_end_success_produces_completed_command_item() {
"c2",
EventMsg::ExecCommandEnd(ExecCommandEndEvent {
call_id: "1".to_string(),
process_id: None,
turn_id: "turn-1".to_string(),
command,
cwd,
@@ -709,6 +711,7 @@ fn exec_command_end_failure_produces_failed_command_item() {
"c1",
EventMsg::ExecCommandBegin(ExecCommandBeginEvent {
call_id: "2".to_string(),
process_id: None,
turn_id: "turn-1".to_string(),
command: command.clone(),
cwd: cwd.clone(),
@@ -737,6 +740,7 @@ fn exec_command_end_failure_produces_failed_command_item() {
"c2",
EventMsg::ExecCommandEnd(ExecCommandEndEvent {
call_id: "2".to_string(),
process_id: None,
turn_id: "turn-1".to_string(),
command,
cwd,
@@ -777,6 +781,7 @@ fn exec_command_end_without_begin_is_ignored() {
"c1",
EventMsg::ExecCommandEnd(ExecCommandEndEvent {
call_id: "no-begin".to_string(),
process_id: None,
turn_id: "turn-1".to_string(),
command: Vec::new(),
cwd: PathBuf::from("."),

View File

@@ -28,3 +28,4 @@ thiserror = { workspace = true }
[dev-dependencies]
pretty_assertions = { workspace = true }
tempfile = { workspace = true }

View File

@@ -0,0 +1,226 @@
use std::fs::OpenOptions;
use std::io::Read;
use std::io::Seek;
use std::io::SeekFrom;
use std::io::Write;
use std::path::Path;
use std::path::PathBuf;
use serde_json;
use thiserror::Error;
#[derive(Debug, Error)]
pub enum AmendError {
#[error("prefix rule requires at least one token")]
EmptyPrefix,
#[error("policy path has no parent: {path}")]
MissingParent { path: PathBuf },
#[error("failed to create policy directory {dir}: {source}")]
CreatePolicyDir {
dir: PathBuf,
source: std::io::Error,
},
#[error("failed to format prefix tokens: {source}")]
SerializePrefix { source: serde_json::Error },
#[error("failed to open policy file {path}: {source}")]
OpenPolicyFile {
path: PathBuf,
source: std::io::Error,
},
#[error("failed to write to policy file {path}: {source}")]
WritePolicyFile {
path: PathBuf,
source: std::io::Error,
},
#[error("failed to lock policy file {path}: {source}")]
LockPolicyFile {
path: PathBuf,
source: std::io::Error,
},
#[error("failed to seek policy file {path}: {source}")]
SeekPolicyFile {
path: PathBuf,
source: std::io::Error,
},
#[error("failed to read policy file {path}: {source}")]
ReadPolicyFile {
path: PathBuf,
source: std::io::Error,
},
#[error("failed to read metadata for policy file {path}: {source}")]
PolicyMetadata {
path: PathBuf,
source: std::io::Error,
},
}
/// Note this thread uses advisory file locking and performs blocking I/O, so it should be used with
/// [`tokio::task::spawn_blocking`] when called from an async context.
pub fn blocking_append_allow_prefix_rule(
policy_path: &Path,
prefix: &[String],
) -> Result<(), AmendError> {
if prefix.is_empty() {
return Err(AmendError::EmptyPrefix);
}
let tokens = prefix
.iter()
.map(serde_json::to_string)
.collect::<Result<Vec<_>, _>>()
.map_err(|source| AmendError::SerializePrefix { source })?;
let pattern = format!("[{}]", tokens.join(", "));
let rule = format!(r#"prefix_rule(pattern={pattern}, decision="allow")"#);
let dir = policy_path
.parent()
.ok_or_else(|| AmendError::MissingParent {
path: policy_path.to_path_buf(),
})?;
match std::fs::create_dir(dir) {
Ok(()) => {}
Err(ref source) if source.kind() == std::io::ErrorKind::AlreadyExists => {}
Err(source) => {
return Err(AmendError::CreatePolicyDir {
dir: dir.to_path_buf(),
source,
});
}
}
append_locked_line(policy_path, &rule)
}
fn append_locked_line(policy_path: &Path, line: &str) -> Result<(), AmendError> {
let mut file = OpenOptions::new()
.create(true)
.read(true)
.append(true)
.open(policy_path)
.map_err(|source| AmendError::OpenPolicyFile {
path: policy_path.to_path_buf(),
source,
})?;
file.lock().map_err(|source| AmendError::LockPolicyFile {
path: policy_path.to_path_buf(),
source,
})?;
let len = file
.metadata()
.map_err(|source| AmendError::PolicyMetadata {
path: policy_path.to_path_buf(),
source,
})?
.len();
// Ensure file ends in a newline before appending.
if len > 0 {
file.seek(SeekFrom::End(-1))
.map_err(|source| AmendError::SeekPolicyFile {
path: policy_path.to_path_buf(),
source,
})?;
let mut last = [0; 1];
file.read_exact(&mut last)
.map_err(|source| AmendError::ReadPolicyFile {
path: policy_path.to_path_buf(),
source,
})?;
if last[0] != b'\n' {
file.write_all(b"\n")
.map_err(|source| AmendError::WritePolicyFile {
path: policy_path.to_path_buf(),
source,
})?;
}
}
file.write_all(format!("{line}\n").as_bytes())
.map_err(|source| AmendError::WritePolicyFile {
path: policy_path.to_path_buf(),
source,
})?;
Ok(())
}
#[cfg(test)]
mod tests {
use super::*;
use pretty_assertions::assert_eq;
use tempfile::tempdir;
#[test]
fn appends_rule_and_creates_directories() {
let tmp = tempdir().expect("create temp dir");
let policy_path = tmp.path().join("policy").join("default.codexpolicy");
blocking_append_allow_prefix_rule(
&policy_path,
&[String::from("echo"), String::from("Hello, world!")],
)
.expect("append rule");
let contents =
std::fs::read_to_string(&policy_path).expect("default.codexpolicy should exist");
assert_eq!(
contents,
r#"prefix_rule(pattern=["echo", "Hello, world!"], decision="allow")
"#
);
}
#[test]
fn appends_rule_without_duplicate_newline() {
let tmp = tempdir().expect("create temp dir");
let policy_path = tmp.path().join("policy").join("default.codexpolicy");
std::fs::create_dir_all(policy_path.parent().unwrap()).expect("create policy dir");
std::fs::write(
&policy_path,
r#"prefix_rule(pattern=["ls"], decision="allow")
"#,
)
.expect("write seed rule");
blocking_append_allow_prefix_rule(
&policy_path,
&[String::from("echo"), String::from("Hello, world!")],
)
.expect("append rule");
let contents = std::fs::read_to_string(&policy_path).expect("read policy");
assert_eq!(
contents,
r#"prefix_rule(pattern=["ls"], decision="allow")
prefix_rule(pattern=["echo", "Hello, world!"], decision="allow")
"#
);
}
#[test]
fn inserts_newline_when_missing_before_append() {
let tmp = tempdir().expect("create temp dir");
let policy_path = tmp.path().join("policy").join("default.codexpolicy");
std::fs::create_dir_all(policy_path.parent().unwrap()).expect("create policy dir");
std::fs::write(
&policy_path,
r#"prefix_rule(pattern=["ls"], decision="allow")"#,
)
.expect("write seed rule without newline");
blocking_append_allow_prefix_rule(
&policy_path,
&[String::from("echo"), String::from("Hello, world!")],
)
.expect("append rule");
let contents = std::fs::read_to_string(&policy_path).expect("read policy");
assert_eq!(
contents,
r#"prefix_rule(pattern=["ls"], decision="allow")
prefix_rule(pattern=["echo", "Hello, world!"], decision="allow")
"#
);
}
}

View File

@@ -1,3 +1,4 @@
pub mod amend;
pub mod decision;
pub mod error;
pub mod execpolicycheck;
@@ -5,6 +6,8 @@ pub mod parser;
pub mod policy;
pub mod rule;
pub use amend::AmendError;
pub use amend::blocking_append_allow_prefix_rule;
pub use decision::Decision;
pub use error::Error;
pub use error::Result;

View File

@@ -1,9 +1,15 @@
use crate::decision::Decision;
use crate::error::Error;
use crate::error::Result;
use crate::rule::PatternToken;
use crate::rule::PrefixPattern;
use crate::rule::PrefixRule;
use crate::rule::RuleMatch;
use crate::rule::RuleRef;
use multimap::MultiMap;
use serde::Deserialize;
use serde::Serialize;
use std::sync::Arc;
#[derive(Clone, Debug)]
pub struct Policy {
@@ -23,6 +29,27 @@ impl Policy {
&self.rules_by_program
}
pub fn add_prefix_rule(&mut self, prefix: &[String], decision: Decision) -> Result<()> {
let (first_token, rest) = prefix
.split_first()
.ok_or_else(|| Error::InvalidPattern("prefix cannot be empty".to_string()))?;
let rule: RuleRef = Arc::new(PrefixRule {
pattern: PrefixPattern {
first: Arc::from(first_token.as_str()),
rest: rest
.iter()
.map(|token| PatternToken::Single(token.clone()))
.collect::<Vec<_>>()
.into(),
},
decision,
});
self.rules_by_program.insert(first_token.clone(), rule);
Ok(())
}
pub fn check(&self, cmd: &[String]) -> Evaluation {
let rules = match cmd.first() {
Some(first) => match self.rules_by_program.get_vec(first) {

View File

@@ -1,8 +1,12 @@
use std::any::Any;
use std::sync::Arc;
use anyhow::Context;
use anyhow::Result;
use codex_execpolicy::Decision;
use codex_execpolicy::Error;
use codex_execpolicy::Evaluation;
use codex_execpolicy::Policy;
use codex_execpolicy::PolicyParser;
use codex_execpolicy::RuleMatch;
use codex_execpolicy::RuleRef;
@@ -35,16 +39,14 @@ fn rule_snapshots(rules: &[RuleRef]) -> Vec<RuleSnapshot> {
}
#[test]
fn basic_match() {
fn basic_match() -> Result<()> {
let policy_src = r#"
prefix_rule(
pattern = ["git", "status"],
)
"#;
let mut parser = PolicyParser::new();
parser
.parse("test.codexpolicy", policy_src)
.expect("parse policy");
parser.parse("test.codexpolicy", policy_src)?;
let policy = parser.build();
let cmd = tokens(&["git", "status"]);
let evaluation = policy.check(&cmd);
@@ -58,10 +60,54 @@ prefix_rule(
},
evaluation
);
Ok(())
}
#[test]
fn parses_multiple_policy_files() {
fn add_prefix_rule_extends_policy() -> Result<()> {
let mut policy = Policy::empty();
policy.add_prefix_rule(&tokens(&["ls", "-l"]), Decision::Prompt)?;
let rules = rule_snapshots(policy.rules().get_vec("ls").context("missing ls rules")?);
assert_eq!(
vec![RuleSnapshot::Prefix(PrefixRule {
pattern: PrefixPattern {
first: Arc::from("ls"),
rest: vec![PatternToken::Single(String::from("-l"))].into(),
},
decision: Decision::Prompt,
})],
rules
);
let evaluation = policy.check(&tokens(&["ls", "-l", "/tmp"]));
assert_eq!(
Evaluation::Match {
decision: Decision::Prompt,
matched_rules: vec![RuleMatch::PrefixRuleMatch {
matched_prefix: tokens(&["ls", "-l"]),
decision: Decision::Prompt,
}],
},
evaluation
);
Ok(())
}
#[test]
fn add_prefix_rule_rejects_empty_prefix() -> Result<()> {
let mut policy = Policy::empty();
let result = policy.add_prefix_rule(&[], Decision::Allow);
match result.unwrap_err() {
Error::InvalidPattern(message) => assert_eq!(message, "prefix cannot be empty"),
other => panic!("expected InvalidPattern(..), got {other:?}"),
}
Ok(())
}
#[test]
fn parses_multiple_policy_files() -> Result<()> {
let first_policy = r#"
prefix_rule(
pattern = ["git"],
@@ -75,15 +121,11 @@ prefix_rule(
)
"#;
let mut parser = PolicyParser::new();
parser
.parse("first.codexpolicy", first_policy)
.expect("parse policy");
parser
.parse("second.codexpolicy", second_policy)
.expect("parse policy");
parser.parse("first.codexpolicy", first_policy)?;
parser.parse("second.codexpolicy", second_policy)?;
let policy = parser.build();
let git_rules = rule_snapshots(policy.rules().get_vec("git").expect("git rules"));
let git_rules = rule_snapshots(policy.rules().get_vec("git").context("missing git rules")?);
assert_eq!(
vec![
RuleSnapshot::Prefix(PrefixRule {
@@ -133,23 +175,27 @@ prefix_rule(
},
commit_eval
);
Ok(())
}
#[test]
fn only_first_token_alias_expands_to_multiple_rules() {
fn only_first_token_alias_expands_to_multiple_rules() -> Result<()> {
let policy_src = r#"
prefix_rule(
pattern = [["bash", "sh"], ["-c", "-l"]],
)
"#;
let mut parser = PolicyParser::new();
parser
.parse("test.codexpolicy", policy_src)
.expect("parse policy");
parser.parse("test.codexpolicy", policy_src)?;
let policy = parser.build();
let bash_rules = rule_snapshots(policy.rules().get_vec("bash").expect("bash rules"));
let sh_rules = rule_snapshots(policy.rules().get_vec("sh").expect("sh rules"));
let bash_rules = rule_snapshots(
policy
.rules()
.get_vec("bash")
.context("missing bash rules")?,
);
let sh_rules = rule_snapshots(policy.rules().get_vec("sh").context("missing sh rules")?);
assert_eq!(
vec![RuleSnapshot::Prefix(PrefixRule {
pattern: PrefixPattern {
@@ -194,22 +240,21 @@ prefix_rule(
},
sh_eval
);
Ok(())
}
#[test]
fn tail_aliases_are_not_cartesian_expanded() {
fn tail_aliases_are_not_cartesian_expanded() -> Result<()> {
let policy_src = r#"
prefix_rule(
pattern = ["npm", ["i", "install"], ["--legacy-peer-deps", "--no-save"]],
)
"#;
let mut parser = PolicyParser::new();
parser
.parse("test.codexpolicy", policy_src)
.expect("parse policy");
parser.parse("test.codexpolicy", policy_src)?;
let policy = parser.build();
let rules = rule_snapshots(policy.rules().get_vec("npm").expect("npm rules"));
let rules = rule_snapshots(policy.rules().get_vec("npm").context("missing npm rules")?);
assert_eq!(
vec![RuleSnapshot::Prefix(PrefixRule {
pattern: PrefixPattern {
@@ -251,10 +296,11 @@ prefix_rule(
},
npm_install
);
Ok(())
}
#[test]
fn match_and_not_match_examples_are_enforced() {
fn match_and_not_match_examples_are_enforced() -> Result<()> {
let policy_src = r#"
prefix_rule(
pattern = ["git", "status"],
@@ -266,9 +312,7 @@ prefix_rule(
)
"#;
let mut parser = PolicyParser::new();
parser
.parse("test.codexpolicy", policy_src)
.expect("parse policy");
parser.parse("test.codexpolicy", policy_src)?;
let policy = parser.build();
let match_eval = policy.check(&tokens(&["git", "status"]));
assert_eq!(
@@ -289,10 +333,11 @@ prefix_rule(
"status",
]));
assert_eq!(Evaluation::NoMatch {}, no_match_eval);
Ok(())
}
#[test]
fn strictest_decision_wins_across_matches() {
fn strictest_decision_wins_across_matches() -> Result<()> {
let policy_src = r#"
prefix_rule(
pattern = ["git"],
@@ -304,9 +349,7 @@ prefix_rule(
)
"#;
let mut parser = PolicyParser::new();
parser
.parse("test.codexpolicy", policy_src)
.expect("parse policy");
parser.parse("test.codexpolicy", policy_src)?;
let policy = parser.build();
let commit = policy.check(&tokens(&["git", "commit", "-m", "hi"]));
@@ -326,10 +369,11 @@ prefix_rule(
},
commit
);
Ok(())
}
#[test]
fn strictest_decision_across_multiple_commands() {
fn strictest_decision_across_multiple_commands() -> Result<()> {
let policy_src = r#"
prefix_rule(
pattern = ["git"],
@@ -341,9 +385,7 @@ prefix_rule(
)
"#;
let mut parser = PolicyParser::new();
parser
.parse("test.codexpolicy", policy_src)
.expect("parse policy");
parser.parse("test.codexpolicy", policy_src)?;
let policy = parser.build();
let commands = vec![
@@ -372,4 +414,5 @@ prefix_rule(
},
evaluation
);
Ok(())
}

View File

@@ -14,7 +14,7 @@ codex-core = { path = "../core" }
reqwest = { version = "0.12", features = ["json", "stream"] }
serde_json = "1"
tokio = { version = "1", features = ["rt"] }
tracing = { version = "0.1.41", features = ["log"] }
tracing = { version = "0.1.43", features = ["log"] }
which = "6.0"
[dev-dependencies]

View File

@@ -8,11 +8,10 @@ use std::time::Instant;
use crate::pkce::PkceCodes;
use crate::server::ServerOptions;
use std::io::Write;
use std::io::{self};
use std::io;
const ANSI_YELLOW: &str = "\x1b[93m";
const ANSI_BOLD: &str = "\x1b[1m";
const ANSI_BLUE: &str = "\x1b[94m";
const ANSI_GRAY: &str = "\x1b[90m";
const ANSI_RESET: &str = "\x1b[0m";
#[derive(Deserialize)]
@@ -138,14 +137,16 @@ async fn poll_for_token(
}
}
fn print_colored_warning_device_code() {
let mut stdout = io::stdout().lock();
let _ = write!(
stdout,
"{ANSI_YELLOW}{ANSI_BOLD}Only use device code authentication when browser login is not available.{ANSI_RESET}{ANSI_YELLOW}\n\
{ANSI_BOLD}Keep the code secret; do not share it.{ANSI_RESET}{ANSI_RESET}\n\n"
fn print_device_code_prompt(code: &str) {
println!(
"\nWelcome to Codex [v{ANSI_GRAY}{version}{ANSI_RESET}]\n{ANSI_GRAY}OpenAI's command-line coding agent{ANSI_RESET}\n\
\nFollow these steps to sign in with ChatGPT using device code authorization:\n\
\n1. Open this link in your browser\n {ANSI_BLUE}https://auth.openai.com/codex/device{ANSI_RESET}\n\
\n2. Enter this one-time code {ANSI_GRAY}(expires in 15 minutes){ANSI_RESET}\n {ANSI_BLUE}{code}{ANSI_RESET}\n\
\n{ANSI_GRAY}Device codes are a common phishing target. Never share this code.{ANSI_RESET}\n",
version = env!("CARGO_PKG_VERSION"),
code = code
);
let _ = stdout.flush();
}
/// Full device code login flow.
@@ -153,13 +154,9 @@ pub async fn run_device_code_login(opts: ServerOptions) -> std::io::Result<()> {
let client = reqwest::Client::new();
let base_url = opts.issuer.trim_end_matches('/');
let api_base_url = format!("{}/api/accounts", opts.issuer.trim_end_matches('/'));
print_colored_warning_device_code();
let uc = request_user_code(&client, &api_base_url, &opts.client_id).await?;
println!(
"To authenticate:\n 1. Open in your browser: {ANSI_BOLD}https://auth.openai.com/codex/device{ANSI_RESET}\n 2. Enter the one-time code below within 15 minutes:\n\n {ANSI_BOLD}{}{ANSI_RESET}\n",
uc.user_code
);
print_device_code_prompt(&uc.user_code);
let code_resp = poll_for_token(
&client,

View File

@@ -307,6 +307,7 @@ async fn run_codex_tool_session_inner(
| EventMsg::UndoCompleted(_)
| EventMsg::ExitedReviewMode(_)
| EventMsg::ContextCompacted(_)
| EventMsg::SaveSessionResponse(_)
| EventMsg::DeprecationNotice(_) => {
// For now, we do not do anything extra for these
// events. Note that

View File

@@ -38,6 +38,13 @@ SERVER_NOTIFICATION_TYPE_NAMES: list[str] = []
# order to compile without warnings.
LARGE_ENUMS = {"ServerResult"}
# some types need setting a default value for `r#type`
# ref: [#7417](https://github.com/openai/codex/pull/7417)
default_type_values: dict[str, str] = {
"ToolInputSchema": "object",
"ToolOutputSchema": "object",
}
def main() -> int:
parser = argparse.ArgumentParser(
@@ -351,6 +358,14 @@ class StructField:
out.append(f" pub {self.name}: {self.type_name},\n")
def append_serde_attr(existing: str | None, fragment: str) -> str:
if existing is None:
return f"#[serde({fragment})]"
assert existing.startswith("#[serde(") and existing.endswith(")]"), existing
body = existing[len("#[serde(") : -2]
return f"#[serde({body}, {fragment})]"
def define_struct(
name: str,
properties: dict[str, Any],
@@ -359,6 +374,14 @@ def define_struct(
) -> list[str]:
out: list[str] = []
type_default_fn: str | None = None
if name in default_type_values:
snake_name = to_snake_case(name) or name
type_default_fn = f"{snake_name}_type_default_str"
out.append(f"fn {type_default_fn}() -> String {{\n")
out.append(f' "{default_type_values[name]}".to_string()\n')
out.append("}\n\n")
fields: list[StructField] = []
for prop_name, prop in properties.items():
if prop_name == "_meta":
@@ -380,6 +403,10 @@ def define_struct(
if is_optional:
prop_type = f"Option<{prop_type}>"
rs_prop = rust_prop_name(prop_name, is_optional)
if prop_name == "type" and type_default_fn:
rs_prop.serde = append_serde_attr(rs_prop.serde, f'default = "{type_default_fn}"')
if prop_type.startswith("&'static str"):
fields.append(StructField("const", rs_prop.name, prop_type, rs_prop.serde, rs_prop.ts))
else:

View File

@@ -1474,6 +1474,10 @@ pub struct Tool {
pub title: Option<String>,
}
fn tool_output_schema_type_default_str() -> String {
"object".to_string()
}
/// An optional JSON Schema object defining the structure of the tool's output returned in
/// the structuredContent field of a CallToolResult.
#[derive(Debug, Clone, PartialEq, Deserialize, Serialize, JsonSchema, TS)]
@@ -1484,9 +1488,14 @@ pub struct ToolOutputSchema {
#[serde(default, skip_serializing_if = "Option::is_none")]
#[ts(optional)]
pub required: Option<Vec<String>>,
#[serde(default = "tool_output_schema_type_default_str")]
pub r#type: String, // &'static str = "object"
}
fn tool_input_schema_type_default_str() -> String {
"object".to_string()
}
/// A JSON Schema object defining the expected parameters for the tool.
#[derive(Debug, Clone, PartialEq, Deserialize, Serialize, JsonSchema, TS)]
pub struct ToolInputSchema {
@@ -1496,6 +1505,7 @@ pub struct ToolInputSchema {
#[serde(default, skip_serializing_if = "Option::is_none")]
#[ts(optional)]
pub required: Option<Vec<String>>,
#[serde(default = "tool_input_schema_type_default_str")]
pub r#type: String, // &'static str = "object"
}

View File

@@ -132,6 +132,7 @@ pub enum ResponseItem {
GhostSnapshot {
ghost_commit: GhostCommit,
},
#[serde(alias = "compaction")]
CompactionSummary {
encrypted_content: String,
},
@@ -537,6 +538,7 @@ mod tests {
use anyhow::Result;
use mcp_types::ImageContent;
use mcp_types::TextContent;
use pretty_assertions::assert_eq;
use tempfile::tempdir;
#[test]
@@ -650,6 +652,21 @@ mod tests {
Ok(())
}
#[test]
fn deserializes_compaction_alias() -> Result<()> {
let json = r#"{"type":"compaction","encrypted_content":"abc"}"#;
let item: ResponseItem = serde_json::from_str(json)?;
assert_eq!(
item,
ResponseItem::CompactionSummary {
encrypted_content: "abc".into(),
}
);
Ok(())
}
#[test]
fn roundtrips_web_search_call_actions() -> Result<()> {
let cases = vec![

View File

@@ -105,6 +105,9 @@ pub enum Op {
final_output_json_schema: Option<Value>,
},
/// Persist the current session under a user-provided name.
SaveSession { name: String },
/// Override parts of the persistent turn context for subsequent turns.
///
/// All fields are optional; when omitted, the existing value is preserved.
@@ -531,6 +534,9 @@ pub enum EventMsg {
BackgroundEvent(BackgroundEventEvent),
/// Result of a save-session request.
SaveSessionResponse(SaveSessionResponseEvent),
UndoStarted(UndoStartedEvent),
UndoCompleted(UndoCompletedEvent),
@@ -1088,6 +1094,13 @@ impl InitialHistory {
}
}
pub fn without_session_meta(&self) -> Vec<RolloutItem> {
self.get_rollout_items()
.into_iter()
.filter(|item| !matches!(item, RolloutItem::SessionMeta(_)))
.collect()
}
pub fn get_event_msgs(&self) -> Option<Vec<EventMsg>> {
match self {
InitialHistory::New => None,
@@ -1171,6 +1184,8 @@ pub struct SessionMeta {
#[serde(default)]
pub source: SessionSource,
pub model_provider: Option<String>,
#[serde(skip_serializing_if = "Option::is_none")]
pub name: Option<String>,
}
impl Default for SessionMeta {
@@ -1184,6 +1199,7 @@ impl Default for SessionMeta {
instructions: None,
source: SessionSource::default(),
model_provider: None,
name: None,
}
}
}
@@ -1256,13 +1272,47 @@ pub struct GitInfo {
pub repository_url: Option<String>,
}
#[derive(Debug, Clone, Copy, Deserialize, Serialize, PartialEq, Eq, JsonSchema, TS)]
#[serde(rename_all = "snake_case")]
pub enum ReviewDelivery {
Inline,
Detached,
}
#[derive(Serialize, Deserialize, Clone, Debug, PartialEq, JsonSchema, TS)]
#[serde(tag = "type", rename_all = "camelCase")]
#[ts(tag = "type")]
pub enum ReviewTarget {
/// Review the working tree: staged, unstaged, and untracked files.
UncommittedChanges,
/// Review changes between the current branch and the given base branch.
#[serde(rename_all = "camelCase")]
#[ts(rename_all = "camelCase")]
BaseBranch { branch: String },
/// Review the changes introduced by a specific commit.
#[serde(rename_all = "camelCase")]
#[ts(rename_all = "camelCase")]
Commit {
sha: String,
/// Optional human-readable label (e.g., commit subject) for UIs.
title: Option<String>,
},
/// Arbitrary instructions provided by the user.
#[serde(rename_all = "camelCase")]
#[ts(rename_all = "camelCase")]
Custom { instructions: String },
}
#[derive(Debug, Clone, Deserialize, Serialize, PartialEq, JsonSchema, TS)]
/// Review request sent to the review session.
pub struct ReviewRequest {
pub prompt: String,
pub user_facing_hint: String,
#[serde(default)]
pub append_to_original_thread: bool,
pub target: ReviewTarget,
#[serde(skip_serializing_if = "Option::is_none")]
#[ts(optional)]
pub user_facing_hint: Option<String>,
}
/// Structured review result produced by a child review session.
@@ -1328,6 +1378,10 @@ impl Default for ExecCommandSource {
pub struct ExecCommandBeginEvent {
/// Identifier so this can be paired with the ExecCommandEnd event.
pub call_id: String,
/// Identifier for the underlying PTY process (when available).
#[serde(default, skip_serializing_if = "Option::is_none")]
#[ts(optional)]
pub process_id: Option<String>,
/// Turn ID that this command belongs to.
pub turn_id: String,
/// The command to be executed.
@@ -1348,6 +1402,10 @@ pub struct ExecCommandBeginEvent {
pub struct ExecCommandEndEvent {
/// Identifier for the ExecCommandBegin that finished.
pub call_id: String,
/// Identifier for the underlying PTY process (when available).
#[serde(default, skip_serializing_if = "Option::is_none")]
#[ts(optional)]
pub process_id: Option<String>,
/// Turn ID that this command belongs to.
pub turn_id: String,
/// The command that was executed.
@@ -1447,6 +1505,13 @@ pub struct StreamInfoEvent {
pub message: String,
}
#[derive(Debug, Clone, Deserialize, Serialize, JsonSchema, TS)]
pub struct SaveSessionResponseEvent {
pub name: String,
pub rollout_path: PathBuf,
pub conversation_id: ConversationId,
}
#[derive(Debug, Clone, Deserialize, Serialize, JsonSchema, TS)]
pub struct PatchApplyBeginEvent {
/// Identifier so this can be paired with the PatchApplyEnd event.

View File

@@ -91,6 +91,7 @@ unicode-width = { workspace = true }
url = { workspace = true }
codex-windows-sandbox = { workspace = true }
tokio-util = { workspace = true, features = ["time"] }
[target.'cfg(unix)'.dependencies]
libc = { workspace = true }

View File

@@ -7,6 +7,8 @@ use crate::diff_render::DiffSummary;
use crate::exec_command::strip_bash_lc_and_escape;
use crate::file_search::FileSearchManager;
use crate::history_cell::HistoryCell;
#[cfg(not(debug_assertions))]
use crate::history_cell::UpdateAvailableHistoryCell;
use crate::model_migration::ModelMigrationOutcome;
use crate::model_migration::migration_copy_for_config;
use crate::model_migration::run_model_migration_prompt;
@@ -14,6 +16,8 @@ use crate::pager_overlay::Overlay;
use crate::render::highlight::highlight_bash_to_lines;
use crate::render::renderable::Renderable;
use crate::resume_picker::ResumeSelection;
use crate::skill_error_prompt::SkillErrorPromptOutcome;
use crate::skill_error_prompt::run_skill_error_prompt;
use crate::tui;
use crate::tui::TuiEvent;
use crate::update_action::UpdateAction;
@@ -36,6 +40,7 @@ use codex_core::protocol::Op;
use codex_core::protocol::SessionSource;
use codex_core::protocol::TokenUsage;
use codex_core::protocol_config_types::ReasoningEffort as ReasoningEffortConfig;
use codex_core::skills::load_skills;
use codex_protocol::ConversationId;
use color_eyre::eyre::Result;
use color_eyre::eyre::WrapErr;
@@ -55,9 +60,6 @@ use std::time::Duration;
use tokio::select;
use tokio::sync::mpsc::unbounded_channel;
#[cfg(not(debug_assertions))]
use crate::history_cell::UpdateAvailableHistoryCell;
const GPT_5_1_MIGRATION_AUTH_MODES: [AuthMode; 2] = [AuthMode::ChatGPT, AuthMode::ApiKey];
const GPT_5_1_CODEX_MIGRATION_AUTH_MODES: [AuthMode; 1] = [AuthMode::ChatGPT];
@@ -249,6 +251,7 @@ impl App {
initial_images: Vec<PathBuf>,
resume_selection: ResumeSelection,
feedback: codex_feedback::CodexFeedback,
is_first_run: bool,
) -> Result<AppExitInfo> {
use tokio_stream::StreamExt;
let (app_event_tx, mut app_event_rx) = unbounded_channel();
@@ -267,6 +270,20 @@ impl App {
SessionSource::Cli,
));
let skills_outcome = load_skills(&config);
if !skills_outcome.errors.is_empty() {
match run_skill_error_prompt(tui, &skills_outcome.errors).await {
SkillErrorPromptOutcome::Exit => {
return Ok(AppExitInfo {
token_usage: TokenUsage::default(),
conversation_id: None,
update_action: None,
});
}
SkillErrorPromptOutcome::Continue => {}
}
}
let enhanced_keys_supported = tui.enhanced_keys_supported();
let mut chat_widget = match resume_selection {
@@ -280,6 +297,7 @@ impl App {
enhanced_keys_supported,
auth_manager: auth_manager.clone(),
feedback: feedback.clone(),
is_first_run,
};
ChatWidget::new(init, conversation_manager.clone())
}
@@ -303,6 +321,29 @@ impl App {
enhanced_keys_supported,
auth_manager: auth_manager.clone(),
feedback: feedback.clone(),
is_first_run,
};
ChatWidget::new_from_existing(
init,
resumed.conversation,
resumed.session_configured,
)
}
ResumeSelection::Fork(path) => {
let resumed = conversation_manager
.fork_from_rollout(config.clone(), path.clone(), auth_manager.clone())
.await
.wrap_err_with(|| format!("Failed to fork session from {}", path.display()))?;
let init = crate::chatwidget::ChatWidgetInit {
config: config.clone(),
frame_requester: tui.frame_requester(),
app_event_tx: app_event_tx.clone(),
initial_prompt: initial_prompt.clone(),
initial_images: initial_images.clone(),
enhanced_keys_supported,
auth_manager: auth_manager.clone(),
feedback: feedback.clone(),
is_first_run: false,
};
ChatWidget::new_from_existing(
init,
@@ -456,6 +497,7 @@ impl App {
enhanced_keys_supported: self.enhanced_keys_supported,
auth_manager: self.auth_manager.clone(),
feedback: self.feedback.clone(),
is_first_run: false,
};
self.chat_widget = ChatWidget::new(init, self.server.clone());
if let Some(summary) = summary {

View File

@@ -347,6 +347,7 @@ impl App {
enhanced_keys_supported: self.enhanced_keys_supported,
auth_manager: self.auth_manager.clone(),
feedback: self.feedback.clone(),
is_first_run: false,
};
self.chat_widget =
crate::chatwidget::ChatWidget::new_from_existing(init, conv, session_configured);

View File

@@ -69,7 +69,10 @@ const LARGE_PASTE_CHAR_THRESHOLD: usize = 1000;
#[derive(Debug, PartialEq)]
pub enum InputResult {
Submitted(String),
Command(SlashCommand),
Command {
command: SlashCommand,
args: Option<String>,
},
None,
}
@@ -101,6 +104,7 @@ pub(crate) struct ChatComposer {
dismissed_file_popup_token: Option<String>,
current_file_query: Option<String>,
pending_pastes: Vec<(String, String)>,
large_paste_counters: HashMap<usize, usize>,
has_focus: bool,
attached_images: Vec<AttachedImage>,
placeholder_text: String,
@@ -113,6 +117,7 @@ pub(crate) struct ChatComposer {
footer_mode: FooterMode,
footer_hint_override: Option<Vec<(String, String)>>,
context_window_percent: Option<i64>,
context_window_used_tokens: Option<i64>,
}
/// Popup state at most one can be visible at any time.
@@ -146,6 +151,7 @@ impl ChatComposer {
dismissed_file_popup_token: None,
current_file_query: None,
pending_pastes: Vec::new(),
large_paste_counters: HashMap::new(),
has_focus: has_input_focus,
attached_images: Vec::new(),
placeholder_text,
@@ -156,6 +162,7 @@ impl ChatComposer {
footer_mode: FooterMode::ShortcutSummary,
footer_hint_override: None,
context_window_percent: None,
context_window_used_tokens: None,
};
// Apply configuration via the setter to keep side-effects centralized.
this.set_disable_paste_burst(disable_paste_burst);
@@ -220,7 +227,7 @@ impl ChatComposer {
pub fn handle_paste(&mut self, pasted: String) -> bool {
let char_count = pasted.chars().count();
if char_count > LARGE_PASTE_CHAR_THRESHOLD {
let placeholder = format!("[Pasted Content {char_count} chars]");
let placeholder = self.next_large_paste_placeholder(char_count);
self.textarea.insert_element(&placeholder);
self.pending_pastes.push((placeholder, pasted));
} else if char_count > 1 && self.handle_paste_image_path(pasted.clone()) {
@@ -333,6 +340,19 @@ impl ChatComposer {
PasteBurst::recommended_flush_delay()
}
fn command_args_from_line(line: &str, command: SlashCommand) -> Option<String> {
if let Some((name, rest)) = parse_slash_name(line)
&& name == command.command()
{
let trimmed = rest.trim();
if trimmed.is_empty() {
return None;
}
return Some(trimmed.to_string());
}
None
}
/// Integrate results from an asynchronous file search.
pub(crate) fn on_file_search_result(&mut self, query: String, matches: Vec<FileMatch>) {
// Only apply if user is still editing a token starting with `query`.
@@ -360,6 +380,17 @@ impl ChatComposer {
self.set_has_focus(has_focus);
}
fn next_large_paste_placeholder(&mut self, char_count: usize) -> String {
let base = format!("[Pasted Content {char_count} chars]");
let next_suffix = self.large_paste_counters.entry(char_count).or_insert(0);
*next_suffix += 1;
if *next_suffix == 1 {
base
} else {
format!("{base} #{next_suffix}")
}
}
pub(crate) fn insert_str(&mut self, text: &str) {
self.textarea.insert_str(text);
self.sync_command_popup();
@@ -505,8 +536,9 @@ impl ChatComposer {
if let Some(sel) = popup.selected_item() {
match sel {
CommandItem::Builtin(cmd) => {
let args = Self::command_args_from_line(first_line, cmd);
self.textarea.set_text("");
return (InputResult::Command(cmd), true);
return (InputResult::Command { command: cmd, args }, true);
}
CommandItem::UserPrompt(idx) => {
if let Some(prompt) = popup.prompt(idx) {
@@ -920,22 +952,21 @@ impl ChatComposer {
modifiers: KeyModifiers::NONE,
..
} => {
// If the first line is a bare built-in slash command (no args),
// dispatch it even when the slash popup isn't visible. This preserves
// the workflow: type a prefix ("/di"), press Tab to complete to
// "/diff ", then press Enter to run it. Tab moves the cursor beyond
// the '/name' token and our caret-based heuristic hides the popup,
// but Enter should still dispatch the command rather than submit
// literal text.
// If the first line is a built-in slash command, dispatch it even when
// the slash popup isn't visible. This preserves the workflow:
// type a prefix ("/di"), press Tab to complete to "/diff ", then press
// Enter to run it. Tab moves the cursor beyond the '/name' token and
// our caret-based heuristic hides the popup, but Enter should still
// dispatch the command rather than submit literal text.
let first_line = self.textarea.text().lines().next().unwrap_or("");
if let Some((name, rest)) = parse_slash_name(first_line)
&& rest.is_empty()
if let Some((name, _rest)) = parse_slash_name(first_line)
&& let Some((_n, cmd)) = built_in_slash_commands()
.into_iter()
.find(|(n, _)| *n == name)
{
let args = Self::command_args_from_line(first_line, cmd);
self.textarea.set_text("");
return (InputResult::Command(cmd), true);
return (InputResult::Command { command: cmd, args }, true);
}
// If we're in a paste-like burst capture, treat Enter as part of the burst
// and accumulate it rather than submitting or inserting immediately.
@@ -1387,6 +1418,7 @@ impl ChatComposer {
use_shift_enter_hint: self.use_shift_enter_hint,
is_task_running: self.is_task_running,
context_window_percent: self.context_window_percent,
context_window_used_tokens: self.context_window_used_tokens,
}
}
@@ -1517,10 +1549,13 @@ impl ChatComposer {
self.is_task_running = running;
}
pub(crate) fn set_context_window_percent(&mut self, percent: Option<i64>) {
if self.context_window_percent != percent {
self.context_window_percent = percent;
pub(crate) fn set_context_window(&mut self, percent: Option<i64>, used_tokens: Option<i64>) {
if self.context_window_percent == percent && self.context_window_used_tokens == used_tokens
{
return;
}
self.context_window_percent = percent;
self.context_window_used_tokens = used_tokens;
}
pub(crate) fn set_esc_backtrack_hint(&mut self, show: bool) {
@@ -2395,8 +2430,9 @@ mod tests {
// When a slash command is dispatched, the composer should return a
// Command result (not submit literal text) and clear its textarea.
match result {
InputResult::Command(cmd) => {
InputResult::Command { command: cmd, args } => {
assert_eq!(cmd.command(), "init");
assert!(args.is_none());
}
InputResult::Submitted(text) => {
panic!("expected command dispatch, but composer submitted literal text: {text}")
@@ -2470,7 +2506,10 @@ mod tests {
let (result, _needs_redraw) =
composer.handle_key_event(KeyEvent::new(KeyCode::Enter, KeyModifiers::NONE));
match result {
InputResult::Command(cmd) => assert_eq!(cmd.command(), "diff"),
InputResult::Command { command: cmd, args } => {
assert_eq!(cmd.command(), "diff");
assert!(args.is_none());
}
InputResult::Submitted(text) => {
panic!("expected command dispatch after Tab completion, got literal submit: {text}")
}
@@ -2479,6 +2518,42 @@ mod tests {
assert!(composer.textarea.is_empty());
}
#[test]
fn slash_command_with_args_dispatches_and_preserves_args() {
use crossterm::event::KeyCode;
use crossterm::event::KeyEvent;
use crossterm::event::KeyModifiers;
let (tx, _rx) = unbounded_channel::<AppEvent>();
let sender = AppEventSender::new(tx);
let mut composer = ChatComposer::new(
true,
sender,
false,
"Ask Codex to do anything".to_string(),
false,
);
composer.textarea.set_text("/save feature-one");
let (result, _needs_redraw) =
composer.handle_key_event(KeyEvent::new(KeyCode::Enter, KeyModifiers::NONE));
match result {
InputResult::Command { command: cmd, args } => {
assert_eq!(cmd, SlashCommand::Save);
assert_eq!(args.as_deref(), Some("feature-one"));
}
InputResult::Submitted(text) => {
panic!(
"expected slash command dispatch, but composer submitted literal text: {text}"
)
}
InputResult::None => panic!("expected Command result for '/save feature-one'"),
}
assert!(composer.textarea.is_empty());
}
#[test]
fn slash_mention_dispatches_command_and_inserts_at() {
use crossterm::event::KeyCode;
@@ -2501,8 +2576,9 @@ mod tests {
composer.handle_key_event(KeyEvent::new(KeyCode::Enter, KeyModifiers::NONE));
match result {
InputResult::Command(cmd) => {
InputResult::Command { command: cmd, args } => {
assert_eq!(cmd.command(), "mention");
assert!(args.is_none());
}
InputResult::Submitted(text) => {
panic!("expected command dispatch, but composer submitted literal text: {text}")
@@ -2665,6 +2741,83 @@ mod tests {
);
}
#[test]
fn deleting_duplicate_length_pastes_removes_only_target() {
use crossterm::event::KeyCode;
use crossterm::event::KeyEvent;
use crossterm::event::KeyModifiers;
let (tx, _rx) = unbounded_channel::<AppEvent>();
let sender = AppEventSender::new(tx);
let mut composer = ChatComposer::new(
true,
sender,
false,
"Ask Codex to do anything".to_string(),
false,
);
let paste = "x".repeat(LARGE_PASTE_CHAR_THRESHOLD + 4);
let placeholder_base = format!("[Pasted Content {} chars]", paste.chars().count());
let placeholder_second = format!("{placeholder_base} #2");
composer.handle_paste(paste.clone());
composer.handle_paste(paste.clone());
assert_eq!(
composer.textarea.text(),
format!("{placeholder_base}{placeholder_second}")
);
assert_eq!(composer.pending_pastes.len(), 2);
composer.textarea.set_cursor(composer.textarea.text().len());
composer.handle_key_event(KeyEvent::new(KeyCode::Backspace, KeyModifiers::NONE));
assert_eq!(composer.textarea.text(), placeholder_base);
assert_eq!(composer.pending_pastes.len(), 1);
assert_eq!(composer.pending_pastes[0].0, placeholder_base);
assert_eq!(composer.pending_pastes[0].1, paste);
}
#[test]
fn large_paste_numbering_does_not_reuse_after_deletion() {
use crossterm::event::KeyCode;
use crossterm::event::KeyEvent;
use crossterm::event::KeyModifiers;
let (tx, _rx) = unbounded_channel::<AppEvent>();
let sender = AppEventSender::new(tx);
let mut composer = ChatComposer::new(
true,
sender,
false,
"Ask Codex to do anything".to_string(),
false,
);
let paste = "x".repeat(LARGE_PASTE_CHAR_THRESHOLD + 4);
let base = format!("[Pasted Content {} chars]", paste.chars().count());
let second = format!("{base} #2");
let third = format!("{base} #3");
composer.handle_paste(paste.clone());
composer.handle_paste(paste.clone());
assert_eq!(composer.textarea.text(), format!("{base}{second}"));
composer.textarea.set_cursor(base.len());
composer.handle_key_event(KeyEvent::new(KeyCode::Backspace, KeyModifiers::NONE));
assert_eq!(composer.textarea.text(), second);
assert_eq!(composer.pending_pastes.len(), 1);
assert_eq!(composer.pending_pastes[0].0, second);
composer.textarea.set_cursor(composer.textarea.text().len());
composer.handle_paste(paste);
assert_eq!(composer.textarea.text(), format!("{second}{third}"));
assert_eq!(composer.pending_pastes.len(), 2);
assert_eq!(composer.pending_pastes[0].0, second);
assert_eq!(composer.pending_pastes[1].0, third);
}
#[test]
fn test_partial_placeholder_deletion() {
use crossterm::event::KeyCode;

View File

@@ -1,6 +1,7 @@
use crate::key_hint;
use crate::key_hint::KeyBinding;
use crate::render::line_utils::prefix_lines;
use crate::status::format_tokens_compact;
use crate::ui_consts::FOOTER_INDENT_COLS;
use crossterm::event::KeyCode;
use ratatui::buffer::Buffer;
@@ -18,6 +19,7 @@ pub(crate) struct FooterProps {
pub(crate) use_shift_enter_hint: bool,
pub(crate) is_task_running: bool,
pub(crate) context_window_percent: Option<i64>,
pub(crate) context_window_used_tokens: Option<i64>,
}
#[derive(Clone, Copy, Debug, Eq, PartialEq)]
@@ -81,7 +83,10 @@ fn footer_lines(props: FooterProps) -> Vec<Line<'static>> {
is_task_running: props.is_task_running,
})],
FooterMode::ShortcutSummary => {
let mut line = context_window_line(props.context_window_percent);
let mut line = context_window_line(
props.context_window_percent,
props.context_window_used_tokens,
);
line.push_span(" · ".dim());
line.extend(vec![
key_hint::plain(KeyCode::Char('?')).into(),
@@ -94,7 +99,10 @@ fn footer_lines(props: FooterProps) -> Vec<Line<'static>> {
esc_backtrack_hint: props.esc_backtrack_hint,
}),
FooterMode::EscHint => vec![esc_hint_line(props.esc_backtrack_hint)],
FooterMode::ContextOnly => vec![context_window_line(props.context_window_percent)],
FooterMode::ContextOnly => vec![context_window_line(
props.context_window_percent,
props.context_window_used_tokens,
)],
}
}
@@ -221,9 +229,18 @@ fn build_columns(entries: Vec<Line<'static>>) -> Vec<Line<'static>> {
.collect()
}
fn context_window_line(percent: Option<i64>) -> Line<'static> {
let percent = percent.unwrap_or(100).clamp(0, 100);
Line::from(vec![Span::from(format!("{percent}% context left")).dim()])
fn context_window_line(percent: Option<i64>, used_tokens: Option<i64>) -> Line<'static> {
if let Some(percent) = percent {
let percent = percent.clamp(0, 100);
return Line::from(vec![Span::from(format!("{percent}% context left")).dim()]);
}
if let Some(tokens) = used_tokens {
let used_fmt = format_tokens_compact(tokens);
return Line::from(vec![Span::from(format!("{used_fmt} used")).dim()]);
}
Line::from(vec![Span::from("100% context left").dim()])
}
#[derive(Clone, Copy, Debug, Eq, PartialEq)]
@@ -400,6 +417,7 @@ mod tests {
use_shift_enter_hint: false,
is_task_running: false,
context_window_percent: None,
context_window_used_tokens: None,
},
);
@@ -411,6 +429,7 @@ mod tests {
use_shift_enter_hint: true,
is_task_running: false,
context_window_percent: None,
context_window_used_tokens: None,
},
);
@@ -422,6 +441,7 @@ mod tests {
use_shift_enter_hint: false,
is_task_running: false,
context_window_percent: None,
context_window_used_tokens: None,
},
);
@@ -433,6 +453,7 @@ mod tests {
use_shift_enter_hint: false,
is_task_running: true,
context_window_percent: None,
context_window_used_tokens: None,
},
);
@@ -444,6 +465,7 @@ mod tests {
use_shift_enter_hint: false,
is_task_running: false,
context_window_percent: None,
context_window_used_tokens: None,
},
);
@@ -455,6 +477,7 @@ mod tests {
use_shift_enter_hint: false,
is_task_running: false,
context_window_percent: None,
context_window_used_tokens: None,
},
);
@@ -466,6 +489,19 @@ mod tests {
use_shift_enter_hint: false,
is_task_running: true,
context_window_percent: Some(72),
context_window_used_tokens: None,
},
);
snapshot_footer(
"footer_context_tokens_used",
FooterProps {
mode: FooterMode::ShortcutSummary,
esc_backtrack_hint: false,
use_shift_enter_hint: false,
is_task_running: false,
context_window_percent: None,
context_window_used_tokens: Some(123_456),
},
);
}

View File

@@ -76,6 +76,7 @@ pub(crate) struct BottomPane {
/// Queued user messages to show above the composer while a turn is running.
queued_user_messages: QueuedUserMessages,
context_window_percent: Option<i64>,
context_window_used_tokens: Option<i64>,
}
pub(crate) struct BottomPaneParams {
@@ -118,6 +119,7 @@ impl BottomPane {
esc_backtrack_hint: false,
animations_enabled,
context_window_percent: None,
context_window_used_tokens: None,
}
}
@@ -130,6 +132,11 @@ impl BottomPane {
self.context_window_percent
}
#[cfg(test)]
pub(crate) fn context_window_used_tokens(&self) -> Option<i64> {
self.context_window_used_tokens
}
fn active_view(&self) -> Option<&dyn BottomPaneView> {
self.view_stack.last().map(std::convert::AsRef::as_ref)
}
@@ -344,13 +351,16 @@ impl BottomPane {
}
}
pub(crate) fn set_context_window_percent(&mut self, percent: Option<i64>) {
if self.context_window_percent == percent {
pub(crate) fn set_context_window(&mut self, percent: Option<i64>, used_tokens: Option<i64>) {
if self.context_window_percent == percent && self.context_window_used_tokens == used_tokens
{
return;
}
self.context_window_percent = percent;
self.composer.set_context_window_percent(percent);
self.context_window_used_tokens = used_tokens;
self.composer
.set_context_window(percent, self.context_window_used_tokens);
self.request_redraw();
}

View File

@@ -0,0 +1,5 @@
---
source: tui/src/bottom_pane/footer.rs
expression: terminal.backend()
---
" 123K used · ? for shortcuts "

Some files were not shown because too many files have changed in this diff Show More