I'm seeing two tests fail intermittently in CI. This PR attempts to
address (or at least mitigate) the flakiness.
* summarize_context_three_requests_and_instructions - The test snapshots
server.received_requests() immediately after observing TaskComplete.
Because the OpenAI /v1/responses call is streamed, the HTTP request can
still be draining when that event fires, so wiremock occasionally
reports only two captured requests. Fix is to wait for async activity to
complete.
* archive_conversation_moves_rollout_into_archived_directory - times out
on a slow CI run. Mitigation is to increase timeout value from 10s to
20s.
Implements:
```
thread/list
thread/start
thread/resume
thread/archive
```
along with their integration tests. These are relatively light wrappers
around the existing core logic, and changes to core logic are minimal.
However, an improvement made for developer ergonomics:
- `thread/start` and `thread/resume` automatically attaches a
conversation listener internally, so clients don't have to make a
separate `AddConversationListener` call like they do today.
For consistency, also updated `model/list` and `feedback/upload` (naming
conventions, list API params).
Currently, when the access token expires, we attempt to use the refresh
token to acquire a new access token. This works most of the time.
However, there are situations where the refresh token is expired,
exhausted (already used to perform a refresh), or revoked. In those
cases, the current logic treats the error as transient and attempts to
retry it repeatedly.
This PR changes the token refresh logic to differentiate between
permanent and transient errors. It also changes callers to treat the
permanent errors as fatal rather than retrying them. And it provides
better error messages to users so they understand how to address the
problem. These error messages should also help us further understand why
we're seeing examples of refresh token exhaustion.
Here is the error message in the CLI. The same text appears within the
extension.
<img width="863" height="38" alt="image"
src="https://github.com/user-attachments/assets/7ffc0d08-ebf0-4900-b9a9-265064202f4f"
/>
I also correct the spelling of "Re-connecting", which shouldn't have a
hyphen in it.
Testing: I manually tested these code paths by adding temporary code to
programmatically cause my refresh token to be exhausted (by calling the
token refresh endpoint in a tight loop more than 50 times). I then
simulated an access token expiration, which caused the token refresh
logic to be invoked. I confirmed that the updated logic properly handled
the error condition.
Note: We earlier discussed the idea of forcefully logging out the user
at the point where token refresh failed. I made several attempts to do
this, and all of them resulted in a bad UX. It's important to surface
this error to users in a way that explains the problem and tells them
that they need to log in again. We also previously discussed deleting
the auth.json file when this condition is detected. That also creates
problems because it effectively changes the auth status from logged in
to logged out, and this causes odd failures and inconsistent UX. I think
it's therefore better not to delete auth.json in this case. If the user
closes the CLI or VSCE and starts it again, we properly detect that the
access token is expired and the refresh token is "dead", and we force
the user to go through the login flow at that time.
This should address aspects of #6191, #5679, and #5505
This is just a refactor of `conversation_history` file by breaking it up
into multiple smaller ones with helper. This refactor will help us move
more functionality related to context management here. in a clean way.
Previously it was not possible for codex to run commands as the init
process (pid 1) in linux. Commands run in containers tend to see their
own pid as 1. See https://github.com/openai/codex/issues/4198
This pr implements the solution mentioned in that issue.
Co-authored-by: Eric Traut <etraut@openai.com>
The deprecation message is currently a bit confusing. Users may not
understand what is `[features].x`. I updated the docs and the
deprecation message for more guidance.
---------
Co-authored-by: Gabriel Peal <gpeal@users.noreply.github.com>
V2 for `account/updated` and `account/logout` for app server. correspond
to old `authStatusChange` and `LogoutChatGpt` respectively. Followup PRs
will make other v2 endpoints call `account/updated` instead of
`authStatusChange` too.
## Problem
The `is_api_message` function in `conversation_history.rs` had a
misalignment between its documentation and implementation:
- **Comment stated**: "Anything that is not a system message or
'reasoning' message is considered an API message"
- **Code behavior**: Was returning `true` for `ResponseItem::Reasoning`,
meaning reasoning messages were incorrectly treated as API messages
This inconsistency could lead to reasoning messages being persisted in
conversation history when they should be filtered out.
## Root Cause
Investigation revealed that reasoning messages are explicitly excluded
throughout the codebase:
1. **Chat completions API** (lines 267-272 in `chat_completions.rs`)
omits reasoning from conversation history:
```rust
ResponseItem::Reasoning { .. } | ResponseItem::Other => {
// Omit these items from the conversation history.
continue;
}
```
2. **Existing tests** like `drops_reasoning_when_last_role_is_user` and
`ignores_reasoning_before_last_user` validate that reasoning should be
excluded from API payloads
## Solution
Fixed the `is_api_message` function to align with its documentation and
the rest of the codebase:
```rust
// Before: Reasoning was incorrectly returning true
ResponseItem::Reasoning { .. } | ResponseItem::WebSearchCall { .. } => true,
// After: Reasoning correctly returns false
ResponseItem::WebSearchCall { .. } => true,
ResponseItem::Reasoning { .. } | ResponseItem::Other => false,
```
## Testing
- Enhanced existing test to verify reasoning messages are properly
filtered out
- All 264 core tests pass, including 8 chat completions tests that
validate reasoning behavior
- No regressions introduced
This ensures reasoning messages are consistently excluded from API
message processing across the entire codebase.
This fixes bug #6121.
The `input_messages` field passed to the notify handler is currently
empty because the logic is incorrectly including the OutputText rather
than InputText. I've fixed that and added proper filtering to remove
messages associated with AGENTS.md and other context injected by the
harness.
Testing: I wrote a notify handler and verified that the user prompt is
correctly passed through to the handler.
Hi OpenAI Codex team, currently "Visit chatgpt.com/codex/settings/usage
for up-to-date information on rate limits and credits" message in status
card and error messages. For now, without the "https://" prefix, the
link cannot be clicked directly from most terminals or chat interfaces.
<img width="636" height="127" alt="Screenshot 2025-11-02 at 22 47 06"
src="https://github.com/user-attachments/assets/5ea11e8b-fb74-451c-85dc-f4d492b2678b"
/>
---
The fix is intent to improve this issue:
- It makes the link clickable in terminals that support it, hence better
accessibility
- It follows standard URL formatting practices
- It maintains consistency with other links in the application (like the
existing "https://openai.com/chatgpt/pricing" links)
Thank you!
When I enable `experimental_sandbox_command_assessment`, I get an
incorrect deprecation warning: "experimental_sandbox_command_assessment
is deprecated. Use experimental_sandbox_command_assessment instead."
This PR fixes this error.
* Removed sandbox risk categories; feedback indicates that these are not
that useful and "less is more"
* Tweaked the assessment prompt to generate terser answers
* Fixed bug in orchestrator that prevents this feature from being
exposed in the extension
Fixes#4161
Currently Codex uses a regex to parse the "Please try again in 1.898s"
OpenAI-style rate limit message, so that it can wait the correct
duration before retrying. Azure OpenAI returns a different error that
looks like "Rate limit exceeded. Try again in 35 seconds."
This PR extends the regex and parsing code to match in a more fuzzy
manner, handling anything matching the pattern "try again in
\<duration>\<unit>".
## Summary
Can never have enough tests on this code path - checking that json
inside a shell call is deserialized correctly.
## Tests
- [x] These are tests 😎
we are seeing [reports](https://github.com/openai/codex/issues/6004) of
users having verbosity in their config.toml and facing issues.
gpt-5-codex doesn't accept other values rather than medium for
verbosity.
Fixes a Markdown parsing issue where a list item used `*` without a
following space (`*Line ranges ...`). Per CommonMark, a space after the
list marker is required. Updated to `* Line ranges ...` so the guideline
renders as a standalone bullet. This change improves readability and
prevents mis-parsing in renderers.
Co-authored-by: Eric Traut <etraut@openai.com>
- Added the new codex-windows-sandbox crate that builds both a library
entry point (run_windows_sandbox_capture) and a CLI executable to launch
commands inside a Windows restricted-token sandbox, including ACL
management, capability SID provisioning, network lockdown, and output
capture
(windows-sandbox-rs/src/lib.rs:167, windows-sandbox-rs/src/main.rs:54).
- Introduced the experimental WindowsSandbox feature flag and wiring so
Windows builds can opt into the sandbox:
SandboxType::WindowsRestrictedToken, the in-process execution path, and
platform sandbox selection now honor the flag (core/src/features.rs:47,
core/src/config.rs:1224, core/src/safety.rs:19,
core/src/sandboxing/mod.rs:69, core/src/exec.rs:79,
core/src/exec.rs:172).
- Updated workspace metadata to include the new crate and its
Windows-specific dependencies so the core crate can link against it
(codex-rs/
Cargo.toml:91, core/Cargo.toml:86).
- Added a PowerShell bootstrap script that installs the Windows
toolchain, required CLI utilities, and builds the workspace to ease
development
on the platform (scripts/setup-windows.ps1:1).
- Landed a Python smoke-test suite that exercises
read-only/workspace-write policies, ACL behavior, and network denial for
the Windows sandbox
binary (windows-sandbox-rs/sandbox_smoketests.py:1).
# Summary
This PR is related to the Issue #3978 and contains a fix to the seatbelt
profile for macOS that allows to run java/jdk tooling from the sandbox.
I have found that the included change is the minimum change to make it
run on my machine.
There is a unit test added by codex when making this fix. I wonder if it
is useful since you need java installed on the target machine for it to
be relevant. I can remove it it is better.
Fixes#3978
There's still some debate about whether we want to expose
`tools.view_image` or `feature.view_image` so those are left unchanged
for now, but this old `include_view_image_tool` config is good-to-go.
Also updated the doc to reflect that `view_image` tool is now by default
true.
## Summary
Duplicates the tests in `apply_patch_cli.rs`, but tests the freeform
apply_patch tool as opposed to the function call path. The good news is
that all the tests pass with zero logical tests, with the exception of
the heredoc, which doesn't really make sense in the freeform tool
context anyway.
@jif-oai since you wrote the original tests in #5557, I'd love your
opinion on the right way to DRY these test cases between the two. Happy
to set up a more sophisticated harness, but didn't want to go down the
rabbit hole until we agreed on the right pattern
## Testing
- [x] These are tests
Adds AgentMessageContentDelta, ReasoningContentDelta,
ReasoningRawContentDelta item streaming events while maintaining
compatibility for old events.
---------
Co-authored-by: Owen Lin <owen@openai.com>
In this PR, I am exploring migrating task kind to an invocation of
Codex. The main reason would be getting rid off multiple
`ConversationHistory` state and streamlining our context/history
management.
This approach depends on opening a channel between the sub-codex and
codex. This channel is responsible for forwarding `interactive`
(`approvals`) and `non-interactive` events. The `task` is responsible
for handling those events.
This opens the door for implementing `codex as a tool`, replacing
`compact` and `review`, and potentially subagents.
One consideration is this code is very similar to `app-server` specially
in the approval part. If in the future we wanted an interactive
`sub-codex` we should consider using `codex-mcp`
The goal is to have a single place where we actually write files
In a follow-up PR, will move everything config related in a dedicated
module and move the helpers in a dedicated file
We currently have nested enums when sending raw response items in the
app-server protocol. This makes downstream schemas confusing because we
need to embed `type`-discriminated enums within each other.
This PR adds a small wrapper around the response item so we can keep the
schemas separate
feature: Add "!cmd" user shell execution
This change lets users run local shell commands directly from the TUI by
prefixing their input with ! (e.g. !ls). Output is truncated to keep the
exec cell usable, and Ctrl-C cleanly
interrupts long-running commands (e.g. !sleep 10000).
**Summary of changes**
- Route Op::RunUserShellCommand through a dedicated UserShellCommandTask
(core/src/tasks/user_shell.rs), keeping the task logic out of codex.rs.
- Reuse the existing tool router: the task constructs a ToolCall for the
local_shell tool and relies on ShellHandler, so no manual MCP tool
lookup is required.
- Emit exec lifecycle events (ExecCommandBegin/ExecCommandEnd) so the
TUI can show command metadata, live output, and exit status.
**End-to-end flow**
**TUI handling**
1. ChatWidget::submit_user_message (TUI) intercepts messages starting
with !.
2. Non-empty commands dispatch Op::RunUserShellCommand { command };
empty commands surface a help hint.
3. No UserInput items are created, so nothing is enqueued for the model.
**Core submission loop**
4. The submission loop routes the op to handlers::run_user_shell_command
(core/src/codex.rs).
5. A fresh TurnContext is created and Session::spawn_user_shell_command
enqueues UserShellCommandTask.
**Task execution**
6. UserShellCommandTask::run emits TaskStartedEvent, formats the
command, and prepares a ToolCall targeting local_shell.
7. ToolCallRuntime::handle_tool_call dispatches to ShellHandler.
**Shell tool runtime**
8. ShellHandler::run_exec_like launches the process via the unified exec
runtime, honoring sandbox and shell policies, and emits
ExecCommandBegin/End.
9. Stdout/stderr are captured for the UI, but the task does not turn the
resulting ToolOutput into a model response.
**Completion**
10. After ExecCommandEnd, the task finishes without an assistant
message; the session marks it complete and the exec cell displays the
final output.
**Conversation context**
- The command and its output never enter the conversation history or the
model prompt; the flow is local-only.
- Only exec/task events are emitted for UI rendering.
**Demo video**
https://github.com/user-attachments/assets/fcd114b0-4304-4448-a367-a04c43e0b996
Found that the VS Code Codex extension throws “Error starting
conversation” when initializing a conversation with Git for Windows’
bash on PATH.
Debugging showed the bash-detection logic did not return as expected;
this change makes it reliable in that scenario.
Possibly related to issue #2841.
