## Refactor of the `execpolicy` crate
To illustrate why we need this refactor, consider an agent attempting to
run `apple | rm -rf ./`. Suppose `apple` is allowed by `execpolicy`.
Before this PR, `execpolicy` would consider `apple` and `pear` and only
render one rule match: `Allow`. We would skip any heuristics checks on
`rm -rf ./` and immediately approve `apple | rm -rf ./` to run.
To fix this, we now thread a `fallback` evaluation function into
`execpolicy` that runs when no `execpolicy` rules match a given command.
In our example, we would run `fallback` on `rm -rf ./` and prevent
`apple | rm -rf ./` from being run without approval.
this PR enables TUI to approve commands and add their prefixes to an
allowlist:
<img width="708" height="605" alt="Screenshot 2025-11-21 at 4 18 07 PM"
src="https://github.com/user-attachments/assets/56a19893-4553-4770-a881-becf79eeda32"
/>
note: we only show the option to whitelist the command when
1) command is not multi-part (e.g `git add -A && git commit -m 'hello
world'`)
2) command is not already matched by an existing rule
This PR moves `ModelsFamily` to `openai_models`. It also propagates
`ModelsManager` to session services and use it to drive model family. We
also make `derive_default_model_family` private because it's a step
towards what we want: one place that gives model configuration.
This is a second step at having one source of truth for models
information and config: `ModelsManager`.
Next steps would be to remove `ModelsFamily` from config. That's massive
because it's being used in 41 occasions mostly pre launching `codex`.
Also, we need to make `find_family_for_model` private. It's also big
because it's being used in 21 occasions ~ all tests.
- This PR treats the `ModelsManager` like `AuthManager` and propagate it
into the tui, replacing the `builtin_model_presets`
- We are also decreasing the visibility of `builtin_model_presets`
based on https://github.com/openai/codex/pull/7552
- Introduce `openai_models` in `/core`
- Move `PRESETS` under it
- Move `ModelPreset`, `ModelUpgrade`, `ReasoningEffortPreset`,
`ReasoningEffortPreset`, and `ReasoningEffortPreset` to `protocol`
- Introduce `Op::ListModels` and `EventMsg::AvailableModels`
Next steps:
- migrate `app-server` and `tui` to use the introduced Operation
No integration test for now because it would make them flaky. Tracking
it in my todos to add some once we have a clock based system for
integration tests
### Summary
Added `mcp/servers/list` which is equivalent to `/mcp` slash command in
CLI for response. This will be used in VSCE MCP settings to show log in
status, available tools etc.
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
# 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.
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`.
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.
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)
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>
- 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
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.
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.
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.
pakrym-oaiandGitHub
zhao-oaiandGitHub
Ahmed IbrahimandGitHub
Michael BolinandGitHub
xl-openaiandGitHub
jif-oaiandGitHub
Shijie RaoandGitHub
Jeremy RoseandGitHub
Robby HeandGitHub
liamandGitHub
Steve MostovoyandGitHub
Dylan HurdandGitHub
Ali TowaijiandGitHub
Job ChongandGitHub
Eric TrautandGitHub
Clifford ResselandGitHub
Josh McKinneyandGitHub
Gabriel PealandGitHub
Matthew ZengandGitHub