Agent wouldn't "see" attached images and would instead try to use the
view_file tool:
<img width="1516" height="504" alt="image"
src="https://github.com/user-attachments/assets/68a705bb-f962-4fc1-9087-e932a6859b12"
/>
In this PR, we wrap image content items in XML tags with the name of
each image (now just a numbered name like `[Image #1]`), so that the
model can understand inline image references (based on name). We also
put the image content items above the user message which the model seems
to prefer (maybe it's more used to definitions being before references).
We also tweak the view_file tool description which seemed to help a bit
Results on a simple eval set of images:
Before
<img width="980" height="310" alt="image"
src="https://github.com/user-attachments/assets/ba838651-2565-4684-a12e-81a36641bf86"
/>
After
<img width="918" height="322" alt="image"
src="https://github.com/user-attachments/assets/10a81951-7ee6-415e-a27e-e7a3fd0aee6f"
/>
```json
[
{
"id": "single_describe",
"prompt": "Describe the attached image in one sentence.",
"images": ["image_a.png"]
},
{
"id": "single_color",
"prompt": "What is the dominant color in the image? Answer with a single color word.",
"images": ["image_b.png"]
},
{
"id": "orientation_check",
"prompt": "Is the image portrait or landscape? Answer in one sentence.",
"images": ["image_c.png"]
},
{
"id": "detail_request",
"prompt": "Look closely at the image and call out any small details you notice.",
"images": ["image_d.png"]
},
{
"id": "two_images_compare",
"prompt": "I attached two images. Are they the same or different? Briefly explain.",
"images": ["image_a.png", "image_b.png"]
},
{
"id": "two_images_captions",
"prompt": "Provide a short caption for each image (Image 1, Image 2).",
"images": ["image_c.png", "image_d.png"]
},
{
"id": "multi_image_rank",
"prompt": "Rank the attached images from most colorful to least colorful.",
"images": ["image_a.png", "image_b.png", "image_c.png"]
},
{
"id": "multi_image_choice",
"prompt": "Which image looks more vibrant? Answer with 'Image 1' or 'Image 2'.",
"images": ["image_b.png", "image_d.png"]
}
]
```
As explained in `codex-rs/core/BUILD.bazel`, including the repo's own
`AGENTS.md` is a hack to get some tests passing. We should fix this
properly, but I wanted to put stake in the ground ASAP to get `just
bazel-remote-test` working and then add a job to `bazel.yml` to ensure
it keeps working.
This PR configures Codex CLI so it can be built with
[Bazel](https://bazel.build) in addition to Cargo. The `.bazelrc`
includes configuration so that remote builds can be done using
[BuildBuddy](https://www.buildbuddy.io).
If you are familiar with Bazel, things should work as you expect, e.g.,
run `bazel test //... --keep-going` to run all the tests in the repo,
but we have also added some new aliases in the `justfile` for
convenience:
- `just bazel-test` to run tests locally
- `just bazel-remote-test` to run tests remotely (currently, the remote
build is for x86_64 Linux regardless of your host platform). Note we are
currently seeing the following test failures in the remote build, so we
still need to figure out what is happening here:
```
failures:
suite::compact::manual_compact_twice_preserves_latest_user_messages
suite::compact_resume_fork::compact_resume_after_second_compaction_preserves_history
suite::compact_resume_fork::compact_resume_and_fork_preserve_model_history_view
```
- `just build-for-release` to build release binaries for all
platforms/architectures remotely
To setup remote execution:
- [Create a buildbuddy account](https://app.buildbuddy.io/) (OpenAI
employees should also request org access at
https://openai.buildbuddy.io/join/ with their `@openai.com` email
address.)
- [Copy your API key](https://app.buildbuddy.io/docs/setup/) to
`~/.bazelrc` (add the line `build
--remote_header=x-buildbuddy-api-key=YOUR_KEY`)
- Use `--config=remote` in your `bazel` invocations (or add `common
--config=remote` to your `~/.bazelrc`, or use the `just` commands)
## CI
In terms of CI, this PR introduces `.github/workflows/bazel.yml`, which
uses Bazel to run the tests _locally_ on Mac and Linux GitHub runners
(we are working on supporting Windows, but that is not ready yet). Note
that the failures we are seeing in `just bazel-remote-test` do not occur
on these GitHub CI jobs, so everything in `.github/workflows/bazel.yml`
is green right now.
The `bazel.yml` uses extra config in `.github/workflows/ci.bazelrc` so
that macOS CI jobs build _remotely_ on Linux hosts (using the
`docker://docker.io/mbolin491/codex-bazel` Docker image declared in the
root `BUILD.bazel`) using cross-compilation to build the macOS
artifacts. Then these artifacts are downloaded locally to GitHub's macOS
runner so the tests can be executed natively. This is the relevant
config that enables this:
```
common:macos --config=remote
common:macos --strategy=remote
common:macos --strategy=TestRunner=darwin-sandbox,local
```
Because of the remote caching benefits we get from BuildBuddy, these new
CI jobs can be extremely fast! For example, consider these two jobs that
ran all the tests on Linux x86_64:
- Bazel 1m37s
https://github.com/openai/codex/actions/runs/20861063212/job/59940545209?pr=8875
- Cargo 9m20s
https://github.com/openai/codex/actions/runs/20861063192/job/59940559592?pr=8875
For now, we will continue to run both the Bazel and Cargo jobs for PRs,
but once we add support for Windows and running Clippy, we should be
able to cutover to using Bazel exclusively for PRs, which should still
speed things up considerably. We will probably continue to run the Cargo
jobs post-merge for commits that land on `main` as a sanity check.
Release builds will also continue to be done by Cargo for now.
Earlier attempt at this PR: https://github.com/openai/codex/pull/8832
Earlier attempt to add support for Buck2, now abandoned:
https://github.com/openai/codex/pull/8504
---------
Co-authored-by: David Zbarsky <dzbarsky@gmail.com>
Co-authored-by: Michael Bolin <mbolin@openai.com>
Fixes#2558
Codex uses alternate screen mode (CSI 1049) which, per xterm spec,
doesn't support scrollback. Zellij follows this strictly, so users can't
scroll back through output.
**Changes:**
- Add `tui.alternate_screen` config: `auto` (default), `always`, `never`
- Add `--no-alt-screen` CLI flag
- Auto-detect Zellij and skip alt screen (uses existing `ZELLIJ` env var
detection)
**Usage:**
```bash
# CLI flag
codex --no-alt-screen
# Or in config.toml
[tui]
alternate_screen = "never"
```
With default `auto` mode, Zellij users get working scrollback without
any config changes.
---------
Co-authored-by: Josh McKinney <joshka@openai.com>
Some enterprises do not want their users to be able to `/feedback`.
<img width="395" height="325" alt="image"
src="https://github.com/user-attachments/assets/2dae9c0b-20c3-4a15-bcd3-0187857ebbd8"
/>
Adds to `config.toml`:
```toml
[feedback]
enabled = false
```
I've deliberately decided to:
1. leave other references to `/feedback` (e.g. in the interrupt message,
tips of the day) unchanged. I think we should continue to promote the
feature even if it is not usable currently.
2. leave the `/feedback` menu item selectable and display an error
saying it's disabled, rather than remove the menu item (which I believe
would raise more questions).
but happy to discuss these.
This will be followed by a change to requirements.toml that admins can
use to force the value of feedback.enabled.
**Motivation**
The `originator` header is important for codex-backend’s Responses API
proxy because it identifies the real end client (codex cli, codex vscode
extension, codex exec, future IDEs) and is used to categorize requests
by client for our enterprise compliance API.
Today the `originator` header is set by either:
- the `CODEX_INTERNAL_ORIGINATOR_OVERRIDE` env var (our VSCode extension
does this)
- calling `set_default_originator()` which sets a global immutable
singleton (`codex exec` does this)
For `codex app-server`, we want the `initialize` JSON-RPC request to set
that header because it is a natural place to do so. Example:
```json
{
"method": "initialize",
"id": 0,
"params": {
"clientInfo": {
"name": "codex_vscode",
"title": "Codex VS Code Extension",
"version": "0.1.0"
}
}
}
```
and when app-server receives that request, it can call
`set_default_originator()`. This is a much more natural interface than
asking third party developers to set an env var.
One hiccup is that `originator()` reads the global singleton and locks
in the value, preventing a later `set_default_originator()` call from
setting it. This would be fine but is brittle, since any codepath that
calls `originator()` before app-server can process an `initialize`
JSON-RPC call would prevent app-server from setting it. This was
actually the case with OTEL initialization which runs on boot, but I
also saw this behavior in certain tests.
Instead, what we now do is:
- [unchanged] If `CODEX_INTERNAL_ORIGINATOR_OVERRIDE` env var is set,
`originator()` would return that value and `set_default_originator()`
with some other value does NOT override it.
- [new] If no env var is set, `originator()` would return the default
value which is `codex_cli_rs` UNTIL `set_default_originator()` is called
once, in which case it is set to the new value and becomes immutable.
Later calls to `set_default_originator()` returns
`SetOriginatorError::AlreadyInitialized`.
**Other notes**
- I updated `codex_core::otel_init::build_provider` to accepts a service
name override, and app-server sends a hardcoded `codex_app_server`
service name to distinguish it from `codex_cli_rs` used by default (e.g.
TUI).
**Next steps**
- Update VSCE to set the proper value for `clientInfo.name` on
`initialize` and drop the `CODEX_INTERNAL_ORIGINATOR_OVERRIDE` env var.
- Delete support for `CODEX_INTERNAL_ORIGINATOR_OVERRIDE` in codex-rs.
Handle null tool arguments in the MCP resource handler so optional
resource tools accept null without failing, preserving normal JSON
parsing for non-null payloads and improving robustness when models emit
null; this avoids spurious argument parse errors for list/read MCP
resource calls.
Elevated Sandbox NUX:
* prompt for elevated sandbox setup when agent mode is selected (via
/approvals or at startup)
* prompt for degraded sandbox if elevated setup is declined or fails
* introduce /elevate-sandbox command to upgrade from degraded
experience.
This seems to be necessary to get the Bazel builds on ARM Linux to go
green on https://github.com/openai/codex/pull/8875.
I don't feel great about timeout-whack-a-mole, but we're still learning
here...
I have seen this test flake out sometimes when running the macOS build
using Bazel in CI: https://github.com/openai/codex/pull/8875. Perhaps
Bazel runs with greater parallelism, inducing a heavier load, causing an
issue?
Historically we started with a CodexAuth that knew how to refresh it's
own tokens and then added AuthManager that did a different kind of
refresh (re-reading from disk).
I don't think it makes sense for both `CodexAuth` and `AuthManager` to
be mutable and contain behaviors.
Move all refresh logic into `AuthManager` and keep `CodexAuth` as a data
object.
This updates core shell environment policy handling to match Windows
case-insensitive variable names and adds a Windows-only regression test,
so Path/TEMP are no longer dropped when inherit=core.
**Before:**
```
Error loading configuration: value `Never` is not in the allowed set [OnRequest]
```
**After:**
```
Error loading configuration: invalid value for `approval_policy`: `Never` is not in the
allowed set [OnRequest] (set by MDM com.openai.codex:requirements_toml_base64)
```
Done by introducing a new struct `ConfigRequirementsWithSources` onto
which we `merge_unset_fields` now. Also introduces a pair of requirement
value and its `RequirementSource` (inspired by `ConfigLayerSource`):
```rust
pub struct Sourced<T> {
pub value: T,
pub source: RequirementSource,
}
```
## Summary
- avoid setting a new process group when stdio is inherited (keeps child
in foreground PG)
- keep process-group isolation when stdio is redirected so killpg
cleanup still works
- prevents macOS job-control SIGTTIN stops that look like hangs after
output
## Testing
- `cargo build -p codex-cli`
- `GIT_CONFIG_GLOBAL=/dev/null GIT_CONFIG_NOSYSTEM=1
CARGO_BIN_EXE_codex=/Users/denis/Code/codex/codex-rs/target/debug/codex
/opt/homebrew/bin/timeout 30m cargo test -p codex-core -p codex-exec`
## Context
This fixes macOS sandbox hangs for commands like `elixir -v` / `erl
-noshell`, where the child was moved into a new process group while
still attached to the controlling TTY. See issue #8690.
## Authorship & collaboration
- This change and analysis were authored by **Codex** (AI coding agent).
- Human collaborator: @seeekr provided repro environment, context, and
review guidance.
- CLI used: `codex-cli 0.77.0`.
- Model: `gpt-5.2-codex (xhigh)`.
Co-authored-by: Eric Traut <etraut@openai.com>
Include project-level AGENTS.md and skills in /review sessions so the
review sub-agent uses the same instruction pipeline as standard runs,
keeping reviewer context aligned with normal sessions.
Sort list_dir entries before applying offset/limit so pagination matches
the displayed order, update pagination/truncation expectations, and add
coverage for sorted pagination. This ensures stable, predictable
directory pages when list_dir is enabled.
Add metrics capabilities to Codex. The `README.md` is up to date.
This will not be merged with the metrics before this PR of course:
https://github.com/openai/codex/pull/8350
To support Bazelification in https://github.com/openai/codex/pull/8875,
this PR introduces a new `find_resource!` macro that we use in place of
our existing logic in tests that looks for resources relative to the
compile-time `CARGO_MANIFEST_DIR` env var.
To make this work, we plan to add the following to all `rust_library()`
and `rust_test()` Bazel rules in the project:
```
rustc_env = {
"BAZEL_PACKAGE": native.package_name(),
},
```
Our new `find_resource!` macro reads this value via
`option_env!("BAZEL_PACKAGE")` so that the Bazel package _of the code
using `find_resource!`_ is injected into the code expanded from the
macro. (If `find_resource()` were a function, then
`option_env!("BAZEL_PACKAGE")` would always be
`codex-rs/utils/cargo-bin`, which is not what we want.)
Note we only consider the `BAZEL_PACKAGE` value when the `RUNFILES_DIR`
environment variable is set at runtime, indicating that the test is
being run by Bazel. In this case, we have to concatenate the runtime
`RUNFILES_DIR` with the compile-time `BAZEL_PACKAGE` value to build the
path to the resource.
In testing this change, I discovered one funky edge case in
`codex-rs/exec-server/tests/common/lib.rs` where we have to _normalize_
(but not canonicalize!) the result from `find_resource!` because the
path contains a `common/..` component that does not exist on disk when
the test is run under Bazel, so it must be semantically normalized using
the [`path-absolutize`](https://crates.io/crates/path-absolutize) crate
before it is passed to `dotslash fetch`.
Because this new behavior may be non-obvious, this PR also updates
`AGENTS.md` to make humans/Codex aware that this API is preferred.
jif-oaiandGitHub
charley-oaiandGitHub
Michael BolinandGitHub
gt-oaiandGitHub
Ahmed IbrahimandGitHub
pakrym-oaiandGitHub
Owen LinandGitHub
Matthew ZengandGitHub
Thibault SottiauxandGitHub
iceweasel-oaiandGitHub
Anton PanasenkoandGitHub
