[codex-mcp] Avoid blocking on optional MCP startup#28407
Closed
bromano-oai wants to merge 1 commit into
Closed
Conversation
bromano-oai
force-pushed
the
codex/cli-8241-7204-nonblocking-mcp-tools
branch
from
June 16, 2026 07:16
37f53a3 to
ff17e5d
Compare
jif-oai
added a commit
that referenced
this pull request
Jun 16, 2026
## Summary - disable the stable Apps feature in the generic `test_codex()` integration-test harness - keep Apps-specific tests explicit: their builders re-enable Apps and point it at a local mock server ## Why Generic tests that use dummy ChatGPT auth were also enabling the host-owned `codex_apps` MCP server. That made unrelated tests contact `chatgpt.com` and wait for MCP startup, causing the Bazel timeouts observed on #28368. The generic harness should be hermetic and should not start an external service that the test did not request. This is test-only; production Apps behavior is unchanged. The broader optional-MCP startup behavior is being handled separately in #28407. ## Testing - `just test -p codex-core -E 'test(pre_sampling_compact_runs_when_comp_hash_changes) | test(model_switch_to_smaller_model_updates_token_context_window) | test(codex_apps_file_params_upload_local_paths_before_mcp_tool_call)'` - `just fix -p codex-core` - `just fmt`
|
Message from Paris's Codex agent: One extra repro that may be relevant here: after upgrading and restarting app-server from 0.143.0-alpha.6 to alpha.14, So I think this may need to cover cold |
wangjiecloud
pushed a commit
to wangjiecloud/codex
that referenced
this pull request
Jun 27, 2026
## Summary - disable the stable Apps feature in the generic `test_codex()` integration-test harness - keep Apps-specific tests explicit: their builders re-enable Apps and point it at a local mock server ## Why Generic tests that use dummy ChatGPT auth were also enabling the host-owned `codex_apps` MCP server. That made unrelated tests contact `chatgpt.com` and wait for MCP startup, causing the Bazel timeouts observed on openai#28368. The generic harness should be hermetic and should not start an external service that the test did not request. This is test-only; production Apps behavior is unchanged. The broader optional-MCP startup behavior is being handled separately in openai#28407. ## Testing - `just test -p codex-core -E 'test(pre_sampling_compact_runs_when_comp_hash_changes) | test(model_switch_to_smaller_model_updates_token_context_window) | test(codex_apps_file_params_upload_local_paths_before_mcp_tool_call)'` - `just fix -p codex-core` - `just fmt`
dkropachev
added a commit
to dkropachev/codex
that referenced
this pull request
Jul 5, 2026
* [codex] Record external agent import results (#28396)
## Summary
- restore `externalAgentConfig/import/progress` notifications while
keeping `externalAgentConfig/import/completed` as the must-deliver event
- persist completed external-agent config imports in state DB by
`importId`, including concrete success/failure details for config,
AGENTS.md, skills, plugins, MCP servers, subagents, hooks, commands, and
sessions
- add `externalAgentConfig/import/readHistories` so clients can recover
persisted import results after missing the live completion notification
- include `errorType` on import failures in protocol
responses/notifications and persisted DB JSON so future code can
classify failures without another wire/storage shape change
## Validation
- `git diff --check`
- `just test -p codex-state external_agent_config_imports`
- `just test -p codex-app-server-protocol`
- `CODEX_SQLITE_HOME=/private/tmp/codex-app-server-sqlite-read-details
just test -p codex-app-server
external_agent_config_import_sends_completion_notification_for_sync_only_import`
Also ran earlier broader checks before publishing:
- `just test -p codex-state`
-
`CODEX_SQLITE_HOME=/private/tmp/codex-app-server-external-agent-test-sqlite
just test -p codex-app-server external_agent_config`
- `just test -p codex-external-agent-migration`
* [codex] expose Bedrock credential source in account/read (#27751)
## Why
`account/read` currently reports only `type: "amazonBedrock"`, so
clients cannot distinguish a Codex-managed Bedrock API key from
credentials supplied by AWS. The app UI needs that distinction to render
the appropriate account state without duplicating provider-auth logic.
Credential-source selection belongs to the Bedrock model provider
because it already owns the precedence between managed Bedrock auth and
the external AWS credential path. This builds on #27443 and #27689.
## What changed
- Added `AmazonBedrockCredentialSource` with `codexManaged` and
`awsManaged` values.
- Included the selected credential source in
`ProviderAccount::AmazonBedrock` and the app-server `Account` response.
- Made `AmazonBedrockModelProvider::account_state()` classify the source
from its managed-auth state.
- Regenerated the app-server JSON and TypeScript schemas.
- Updated app-server account documentation and downstream TUI matches.
`codexManaged` means the provider found a managed Bedrock API key.
`awsManaged` identifies the provider's external AWS credential path; it
does not assert that the AWS credential chain has been validated.
## Testing
- Added model-provider coverage for Codex-managed precedence and
AWS-managed fallback.
- Added app-server protocol serialization coverage for both wire values.
- Added app-server integration coverage for both `account/read`
responses.
- `just test -p codex-protocol -p codex-model-provider -p
codex-app-server-protocol` (497 tests passed).
After rebasing onto #27711, the `codex-app-server` test target compiled
past the image-generation `PathUri` migration. Local linking was then
interrupted by disk exhaustion (`No space left on device`).
* [codex] Compress cold active rollouts (#28338)
## Why
The local rollout compression worker currently scans only
`archived_sessions`, so cold unarchived thread history remains expanded
indefinitely.
## What changed
- Scan `sessions` after `archived_sessions` within the existing worker
runtime budget.
- Update rollout compression coverage to require both cold active and
archived rollouts to be compressed while fresh active rollouts remain
plain.
The worker remains behind the disabled-by-default
`local_thread_store_compression` feature, and the existing seven-day
cold-file threshold is unchanged.
## Validation
- `just test -p codex-rollout` (69 passed)
- `just fmt`
- `git diff --check`
* feat: render typed envelopes for multi-agent v2 messages (#28368)
## Why
Multi-agent v2 messages need a consistent, model-visible envelope that
identifies what kind of interaction occurred, who sent it, and which
agent it targets. Previously, encrypted deliveries exposed only
`encrypted_content`, while child completion used the legacy
`<subagent_notification>` shape. That meant the client could not
consistently present `NEW_TASK`, `MESSAGE`, and `FINAL_ANSWER` using the
same format.
This change adds the routing envelope as plaintext while keeping task
and message payloads encrypted. No new Responses API field is required:
an encrypted delivery is represented as an `input_text` header
immediately followed by its existing `encrypted_content` item.
Every envelope now follows this shape:
```text
Message Type: <NEW_TASK | MESSAGE | FINAL_ANSWER>
Task name: <recipient agent path>
Sender: <author agent path>
Payload:
<message payload>
```
## Message types
### `NEW_TASK`
`NEW_TASK` is used when the recipient should begin a new turn, including
an initial `spawn_agent` task and a later `followup_task`.
For a root agent spawning `/root/worker`, the request contains a
plaintext envelope followed by the encrypted task:
```json
{
"type": "agent_message",
"author": "/root",
"recipient": "/root/worker",
"content": [
{
"type": "input_text",
"text": "Message Type: NEW_TASK\nTask name: /root/worker\nSender: /root\nPayload:\n"
},
{
"type": "encrypted_content",
"encrypted_content": "<encrypted task payload>"
}
]
}
```
Conceptually, the model receives:
```text
Message Type: NEW_TASK
Task name: /root/worker
Sender: /root
Payload:
Review the authentication changes and report any regressions.
```
### `MESSAGE`
`MESSAGE` is used for a queued `send_message` delivery. It communicates
with an existing agent without starting a new turn.
For `/root/worker` reporting progress to the root agent, the request
contains:
```json
{
"type": "agent_message",
"author": "/root/worker",
"recipient": "/root",
"content": [
{
"type": "input_text",
"text": "Message Type: MESSAGE\nTask name: /root\nSender: /root/worker\nPayload:\n"
},
{
"type": "encrypted_content",
"encrypted_content": "<encrypted message payload>"
}
]
}
```
Conceptually, the model receives:
```text
Message Type: MESSAGE
Task name: /root
Sender: /root/worker
Payload:
The protocol tests pass; I am checking the resume path now.
```
### `FINAL_ANSWER`
`FINAL_ANSWER` is emitted when a child agent reaches a terminal state
and reports its result to its parent. Completion payloads are already
available locally, so the complete envelope is represented as plaintext
rather than as a plaintext header plus encrypted content.
For `/root/worker` completing work for the root agent, the request
contains:
```json
{
"type": "agent_message",
"author": "/root/worker",
"recipient": "/root",
"content": [
{
"type": "input_text",
"text": "Message Type: FINAL_ANSWER\nTask name: /root\nSender: /root/worker\nPayload:\nNo regressions found."
}
]
}
```
The model-visible form is:
```text
Message Type: FINAL_ANSWER
Task name: /root
Sender: /root/worker
Payload:
No regressions found.
```
Errored, shut down, and missing agents also use `FINAL_ANSWER`, with a
terminal-status description in the payload.
## What changed
- Render `NEW_TASK` or `MESSAGE` in
`InterAgentCommunication::to_model_input_item`, based on whether the
encrypted delivery starts a turn.
- Replace the multi-agent v2 `<subagent_notification>` completion
payload with a model-visible `FINAL_ANSWER` envelope.
- Document `Task name`, `Sender`, and `Payload` consistently in the
multi-agent developer instructions.
- Prevent local-only history projections from treating an encrypted
message's plaintext header as the complete assistant message.
- Preserve rollout-trace interaction edges when an agent message
contains both plaintext and encrypted content.
Legacy multi-agent behavior remains unchanged.
## Verification
- `just test -p codex-protocol`
- `just test -p codex-rollout-trace`
- `just test -p codex-web-search-extension`
- `just test -p codex-core
encrypted_multi_agent_v2_spawn_sends_agent_message_to_child`
- `just test -p codex-core
plaintext_multi_agent_v2_completion_sends_agent_message`
- `just test -p codex-core
multi_agent_v2_followup_task_completion_notifies_parent_on_every_turn`
- `just test -p codex-core
multi_agent_v2_completion_queues_message_for_direct_parent`
* [tests] Keep Apps out of generic core test harness (#28508)
## Summary
- disable the stable Apps feature in the generic `test_codex()`
integration-test harness
- keep Apps-specific tests explicit: their builders re-enable Apps and
point it at a local mock server
## Why
Generic tests that use dummy ChatGPT auth were also enabling the
host-owned `codex_apps` MCP server. That made unrelated tests contact
`chatgpt.com` and wait for MCP startup, causing the Bazel timeouts
observed on #28368.
The generic harness should be hermetic and should not start an external
service that the test did not request. This is test-only; production
Apps behavior is unchanged. The broader optional-MCP startup behavior is
being handled separately in #28407.
## Testing
- `just test -p codex-core -E
'test(pre_sampling_compact_runs_when_comp_hash_changes) |
test(model_switch_to_smaller_model_updates_token_context_window) |
test(codex_apps_file_params_upload_local_paths_before_mcp_tool_call)'`
- `just fix -p codex-core`
- `just fmt`
* [codex] Clarify plugin load and runtime capability stages (#28472)
## Summary
Plugin loading and auth projection both previously produced
`PluginLoadOutcome`. That made an unfiltered load result look like
runtime-ready capabilities and generated capability summaries before
auth routing had run.
This change keeps loaded plugin records in the cache, applies the
current auth policy in `PluginsManager`, and only then builds
`PluginLoadOutcome` and its summaries. Auth changes still reuse the
cached disk load and re-resolve apps and MCP servers without reloading
plugins.
The updated tests cover cached auth changes and verify that capability
summaries match the effective app/MCP surface.
## Testing
- `just test -p codex-core-plugins`
- `just test -p codex-plugin`
- `just fix -p codex-core-plugins`
* core: surface terminal subagent errors to parent agents (#28375)
## Why
When a subagent exhausts its retries, it emits an `Error`, but the
generic task lifecycle then emits `TurnComplete(None)`. That completion
used to overwrite the subagent's `Errored` status with
`Completed(None)`, so the parent received an empty completion
notification.
This made a failed child look indistinguishable from a child that
completed without an answer. In unattended or long-running multi-agent
work, the root could silently continue without knowing that delegated
work failed or how to restart it.
## Behavior
Before, a terminal stream failure was reduced to an empty completion:
```text
<subagent_notification>
{"agent_path":"/root/worker","status":{"completed":null}}
</subagent_notification>
```
Now the parent receives the actual terminal error, bounded to 1,000
tokens, together with an actionable recovery hint:
```text
<subagent_notification>
{
"agent_path": "/root/worker",
"status": {
"errored": "stream disconnected before completion: stream closed before response.completed"
},
"next_action": "This agent's turn failed. If you still need this agent, use `followup_task` to give it another task."
}
</subagent_notification>
```
The notification remains queue-only: it does not wake the root or replay
the failed request. The root sees it at the next sampling boundary and
can use `followup_task` to start a new turn for that agent.
## What changed
- Added terminal-error precedence to the [agent status
reducer](https://github.com/openai/codex/blob/e95fcfe2bb6a02f1a75650afa20048859f556511/codex-rs/core/src/agent/status.rs#L23-L34),
so a closing `TurnComplete` cannot erase an immediately preceding
`Errored` status.
- Made MultiAgentV2 completion forwarding use the retained session
status instead of re-deriving `Completed(None)` from the final event.
- Extended the [subagent notification
fragment](https://github.com/openai/codex/blob/e95fcfe2bb6a02f1a75650afa20048859f556511/codex-rs/core/src/context/subagent_notification.rs#L6-L60)
with a `next_action` for terminal errors and a hard cap on model-visible
error text.
- Kept successful completions and interrupted turns unchanged.
## Verification
- Added a status-reducer test proving that `Errored` survives the
trailing `TurnComplete`.
- Added an integration test that exhausts a subagent's stream retries
and verifies the exact `agent_message` delivered to the parent,
including the error and `followup_task` guidance.
- Re-ran the existing successful-completion and interrupted-turn
notification tests.
* perf(config): defer remote sandbox hostname lookup (#28542)
## Why
[#18763](https://github.com/openai/codex/pull/18763) added canonical
hostname resolution for `remote_sandbox_config`. Requirements
composition currently performs that synchronous DNS lookup on every
fresh process, even when none of the loaded requirements layers contains
`[[remote_sandbox_config]]`. On hosts with slow local DNS resolution,
this can add several seconds to Codex startup.
## What
- defer hostname resolution until a parsed requirements layer actually
contains `remote_sandbox_config`
- cache the resolver result once per requirements composition,
preserving the existing single-lookup behavior across multiple layers
- keep the existing FQDN resolution and per-layer requirements
precedence unchanged
- cover both the ordinary no-lookup path and the multi-layer
single-lookup path
## How to Test
On a host where local canonical-name resolution is slow:
1. Start Codex without `[[remote_sandbox_config]]` in any managed
requirements layer and confirm startup no longer waits for hostname
resolution.
2. Add a matching `[[remote_sandbox_config]]` entry and confirm its
`allowed_sandbox_modes` still overrides the layer's top-level value.
3. Add remote sandbox entries to multiple requirements layers and
confirm precedence remains unchanged while the hostname is resolved only
once.
Targeted tests:
- `just test -p codex-config hostname_resolver`
- `just test -p codex-config` (181 passed)
* path-uri: clarify invalid host path errors (#28473)
## Why
Ensure a consistent string format when exposing path conversion errors
to the model.
## What
- Render `PathUriParseError::InvalidFileUriPath` as `'$PATH' is invalid
on '$OS'`.
* fix(tui): restore TUI after suspend (#28342)
## Why
On Linux, suspending Codex with `Ctrl+Z` and returning with `fg` can
leave the composer misaligned or inject terminal response bytes such as
focus reports into the prompt. Shell job-control output moves the cursor
while Codex is suspended, and terminal input polling can race with the
responses used to restore the inline viewport.
Fixes #26564.
## What changed
- preserve and restore keyboard reporting without disturbing the parent
terminal stack
- pause terminal event polling while Codex is suspended and flush
buffered input before resuming it
- force crossterm's cached raw-mode state back in sync after the shell
completes its `fg` handoff
- probe the actual post-`fg` cursor position with the tolerant
terminal-response parser, then realign the inline viewport before
redrawing
## How to Test
1. On Linux, start the development TUI with `just c`.
2. Type text into the composer without submitting it.
3. Press `Ctrl+Z`, run any harmless shell command, then run `fg`.
4. Confirm the composer redraws below the shell output, the draft text
is preserved, and no raw escape sequences appear.
5. Repeat the suspend/resume cycle and confirm normal typing still
works.
Targeted tests:
- `cargo test -p codex-tui --lib parses_cursor_position_as_zero_based -j
1`
- `cargo test -p codex-tui --lib tui::event_stream::tests -j 1`
* [codex] exec-server: stream files in chunks (#28354)
## Why
`fs/readFile` buffers the entire file in one response, which makes large
remote reads expensive and prevents callers from applying backpressure.
We need an opt-in streaming path with bounded block sizes while
preserving the existing single-call API for small and sandboxed reads.
## What changed
- Add `ExecServerClient::stream`, returning a named `FileReadStream`
that implements `futures::Stream` and yields immutable 1 MiB byte
blocks.
- Add internal `fs/open`, `fs/readBlock`, and `fs/close` RPCs.
`fs/readBlock` accepts an explicit offset and length.
- Keep unsandboxed files open between block reads, cap open handles per
connection, and clean them up on EOF, error, stream drop, explicit
close, or connection shutdown.
- Reject platform-sandboxed streaming opens instead of turning the
one-shot sandbox helper into a persistent server. Existing `fs/readFile`
behavior is unchanged.
## Testing
- `just test -p codex-exec-server`
- Integration coverage for 1 MiB chunking, exact block-boundary EOF,
sandbox rejection, and continued reads from the opened file after path
replacement.
- Handle-manager coverage for non-sequential offsets, variable block
lengths, the 128-handle limit, and capacity release after close.
* chore: side prompt (#28553)
Fix side bug with prompt
* [codex-app-server-test-client & codex-app-server] Plugin Usage Analytics Smoke Test (#27099)
## This PR
The original [combined remote plugin analytics PR
#26281](https://github.com/openai/codex/pull/26281) mixed reusable
analytics test infrastructure, two manual smoke workflows, a metadata
refactor, and the final identity behavior. This PR establishes a
non-mutating end-to-end plugin smoke workflow before any analytics
identity semantics change.
- Add `plugin-analytics-smoke` to the existing app-server test client.
- Exercise plugin disable, enable, and use through production app-server
RPC paths.
- Isolate config writes in a temporary file and use a loopback Responses
API server.
- Capture analytics without sending them to the production analytics
backend.
- Validate the current local `plugin_id`, names, capability metadata,
thread, turn, and model fields.
This is intentionally a baseline smoke workflow. It does not assert
`remote_plugin_id`; the final PR will update it when that field exists.
Review this PR as the net diff against #27093.
## Testing
- The test-client target compiles successfully.
- The combined reference branch exercised the manual smoke against the
live remote plugin service.
- CI is green across the required platform matrix.
## Split Overview
```text
main
├── #27093 Debug analytics capture
│ └── #27099 Non-mutating plugin smoke ← you are here
│ └── #27100 Remote install/uninstall smoke
└── #27102 Plugin telemetry metadata refactor
After #27093, #27099, #27100, and #27102 merge:
└── Final PR: add remote_plugin_id to plugin analytics
```
Review order and dependencies:
1. [#27093 Add debug-only analytics event
capture](https://github.com/openai/codex/pull/27093) (based on `main`)
2. [#27099 Add a plugin analytics smoke
workflow](https://github.com/openai/codex/pull/27099) **(this PR,
stacked on #27093)**
3. [#27100 Add a remote plugin analytics mutation smoke
workflow](https://github.com/openai/codex/pull/27100) (stacked on this
PR)
4. [#27102 Centralize plugin telemetry metadata
construction](https://github.com/openai/codex/pull/27102) (independent,
based on `main`)
5. Final remote-ID behavior PR (created after PRs 1-4 merge)
The original [#26281](https://github.com/openai/codex/pull/26281)
remains open as the green aggregate reference until the final PR is
published.
* fix(tui): highlight C++ module files (#28554)
## Why
Codex syntax-highlights diffs for conventional C++ extensions such as
`.cpp` and `.cxx`, but C++ module interface files using `.cppm`, `.ixx`,
or `.cxxm` fall back to plain diff coloring. The bundled syntax set
already includes C++, but it does not resolve those module extensions by
itself.
Closes #28223.
## What changed
- map `.cppm`, `.ixx`, and `.cxxm` to the existing `cpp` syntax in
`render/highlight.rs`
- extend alias-resolution coverage for all three module extensions
- verify `.cpp`, `.cppm`, `.ixx`, and `.cxxm` diffs produce
syntax-highlighted RGB spans while unknown extensions retain the plain
fallback
- snapshot the syntax-colored token segmentation for the supported C++
module extensions
## How to Test
1. Ask Codex to create or modify a C++ module interface file using
`.cppm`, `.ixx`, or `.cxxm`.
2. Confirm C++ tokens in the rendered diff receive syntax colors instead
of only the red/green diff treatment.
3. Modify an equivalent `.cpp` file and confirm its existing
highlighting remains unchanged.
4. Modify a file with an unknown extension and confirm it still uses the
plain diff fallback.
Targeted tests:
- `just test -p codex-tui -E
'test(find_syntax_resolves_languages_and_aliases) |
test(cpp_module_extensions_use_cpp_highlighting) |
test(unknown_extension_falls_back_without_syntax_highlighting)'`
* [codex] Warn clearly when code mode output is truncated (#28467)
## Summary
- make `formatted_truncate_text` prepend `Warning: truncated output
(original token count: N)` above the existing `Total output lines`
header
- update direct formatter, unified-exec, user-shell, and code-mode
expectations
- add core unit coverage that runs in Bazel without requiring the
skipped V8-backed code-mode integration suite
## Validation
- `cargo test -p codex-utils-output-truncation -- --nocapture` (17
passed)
- `cargo test -p codex-core --lib
truncated_text_output_starts_with_warning -- --nocapture`
- `cargo test -p codex-core --test all
clamps_model_requested_max_output_tokens_to_policy -- --nocapture` (2
passed)
- `cargo test -p codex-core --test all
unified_exec_formats_large_output_summary -- --nocapture`
- `cargo test -p codex-core --test all
user_shell_command_output_is_truncated_in_history -- --nocapture`
- Bazel CI exercises the shared formatter and downstream integration
expectations
* Add incremental thread history changes
Add ThreadHistoryBuilder APIs for collecting incremental thread item and turn changes while applying rollout items.
Batch handling coalesces repeated changes so callers can get the latest incremental thread item changes for a set of rollout items without rebuilding full history.
* feat(tui): add rate-limit reset redemption to /usage (#28154)
## Why
Codex users can earn personal rate-limit reset credits, but the CLI does
not currently provide a way to view or redeem them. The `/usage` command
restored in #27925 is intended to be the entry point for usage-related
actions, so reset redemption belongs there rather than in a separate
dashed slash command.
Depends on #28143 for the app-server and backend-client reset-credit
APIs.
## What changed
- Turn bare `/usage` into a menu with entries for token activity and
earned rate-limit resets while preserving `/usage daily`, `/usage
weekly`, and `/usage cumulative`.
- Add loading, empty, confirmation, success, retry, and error states
with a caller-generated UUID idempotency key reused across retries of
the same logical reset.
- Show an availability hint only for backend-classified rate-limit
errors with credits available.
- Hide the reset entry for workspace accounts.
## Validation
- `just test -p codex-tui chatwidget::tests::usage` — 19 passed.
- `just fix -p codex-tui` — passed.
- `just fmt` — passed.
- `cargo insta pending-snapshots` from `codex-rs/tui` — no pending
snapshots.
## Examples
<img width="1168" height="304" alt="image"
src="https://github.com/user-attachments/assets/caa4c1e3-e996-494d-ae17-50b521f5dce8"
/>
<img width="908" height="260" alt="image"
src="https://github.com/user-attachments/assets/e38a726b-77cc-4bd0-9ea8-9f3ad21c5768"
/>
### Reset flow
<img width="1509" height="312" alt="image"
src="https://github.com/user-attachments/assets/d987013c-78a5-48a2-ad8d-c61ad267a327"
/>
<img width="585" height="190" alt="image"
src="https://github.com/user-attachments/assets/de32be19-79b9-4a3e-8574-6f1c208c98ae"
/>
<img width="600" height="210" alt="image"
src="https://github.com/user-attachments/assets/88a165cf-796d-4fdc-a7bc-ea89917573da"
/>
<img width="512" height="193" alt="image"
src="https://github.com/user-attachments/assets/d2353998-5aa8-442e-a5f8-3a8a5b832753"
/>
* ci: run code-mode unit tests on all bazel targets (#28562)
## Why
V8 should be stable under Bazel, so the `codex-code-mode` unit tests
should run across the Bazel platform matrix. If these tests prove
unstable, we should fix the tests rather than exclude them from CI.
## What changed
- Remove the explicit `//codex-rs/code-mode:code-mode-unit-tests`
exclusion from the macOS and Linux Bazel test jobs.
- Remove the same exclusion from the native Windows post-merge job.
- Keep the existing Windows gnullvm shard coverage.
## Bazel test coverage
The target contains 26 unit tests. A fresh uncached local Bazel
execution ran all 26 with 0 failures, 0 ignored tests, and 0 filtered
tests.
PR Bazel CI selected the target on every enabled platform and reported a
cached pass:
| Platform | Passing CI job |
| --- | --- |
| macOS aarch64 | [Bazel test
passed](https://github.com/openai/codex/actions/runs/27636617545/job/81725447804)
|
| macOS x86_64 | [Bazel test passed in
2.2s](https://github.com/openai/codex/actions/runs/27636617545/job/81725448008)
|
| Linux GNU | [Bazel test passed in
0.4s](https://github.com/openai/codex/actions/runs/27636617545/job/81725447898)
|
| Linux musl | [Bazel test passed in
0.4s](https://github.com/openai/codex/actions/runs/27636617545/job/81725448117)
|
| Windows gnullvm | [Bazel test passed in shard 4/4 in
1.6s](https://github.com/openai/codex/actions/runs/27636617545/job/81725448166)
|
* [codex] Route MCP file uploads through environment filesystem (#27923)
## Why
Codex Apps tools can mark arguments with `openai/fileParams`, but the
execution path resolved and opened those files directly on the host.
That bypassed the selected turn environment and prevented annotated file
arguments from working with remote environments.
## What changed
- resolve annotated file arguments against the primary turn environment
- read file metadata and contents through that environment's sandboxed
`ExecutorFileSystem`
- reject files over the 512 MiB limit from metadata before reading or
transferring them
- retain the buffered upload-size check as defense in depth
- make the OpenAI upload API accept a filename and buffered contents
instead of owning local filesystem access
- describe the model-visible argument as a path in the primary
environment
This builds on #27927, which added `size` to internal filesystem
metadata.
## Testing
- `just test -p codex-api upload_openai_file_returns_canonical_uri`
- `just test -p codex-mcp
tool_with_model_visible_input_schema_masks_file_params`
- `just test -p codex-core mcp_openai_file`
- `just test -p codex-core
codex_apps_file_params_upload_environment_files_before_mcp_tool_call`
* [codex-app-server-test-client] Plugin Install/Uninstall Analytics Smoke Test (#27100)
## This PR
The original [combined remote plugin analytics PR
#26281](https://github.com/openai/codex/pull/26281) mixed reusable
analytics test infrastructure, two manual smoke workflows, a metadata
refactor, and the final identity behavior. This PR adds the
account-mutating validation workflow separately so its cleanup and
recovery guarantees can be reviewed without the final analytics behavior
change.
- Add a manually invoked remote plugin install/uninstall smoke workflow.
- Require explicit account-mutation confirmation and an initially
uninstalled plugin.
- Validate the current `codex_plugin_installed` contract, where
`plugin_id` is the backend ID.
- Restore and verify the original uninstalled state, with a dedicated
recovery command.
This baseline intentionally does not require `codex_plugin_uninstalled`,
because production does not emit that event yet. The final PR will
update this smoke to require local `plugin_id`, `remote_plugin_id`, and
uninstall emission. Review this PR as the net diff against #27099.
## Testing
- `just test -p codex-app-server-test-client` (3 focused
capture/validation tests passed)
- The live workflow was previously exercised on the green combined
reference branch, and the original uninstalled account state was
restored.
- CI is green across the required platform matrix.
## Split Overview
```text
main
├── #27093 Debug analytics capture
│ └── #27099 Non-mutating plugin smoke
│ └── #27100 Remote install/uninstall smoke ← you are here
└── #27102 Plugin telemetry metadata refactor
After #27093, #27099, #27100, and #27102 merge:
└── Final PR: add remote_plugin_id to plugin analytics
```
Review order and dependencies:
1. [#27093 Add debug-only analytics event
capture](https://github.com/openai/codex/pull/27093) (based on `main`)
2. [#27099 Add a plugin analytics smoke
workflow](https://github.com/openai/codex/pull/27099) (stacked on
#27093)
3. [#27100 Add a remote plugin analytics mutation smoke
workflow](https://github.com/openai/codex/pull/27100) **(this PR,
stacked on #27099)**
4. [#27102 Centralize plugin telemetry metadata
construction](https://github.com/openai/codex/pull/27102) (independent,
based on `main`)
5. Final remote-ID behavior PR (created after PRs 1-4 merge)
The original [#26281](https://github.com/openai/codex/pull/26281)
remains open as the green aggregate reference until the final PR is
published.
* [codex] re-enable absolute workdir integration test (#28581)
## Why
In #28146 I missed the invariant that an absolute `exec_command` workdir
must override the environment cwd. The existing integration test would
have caught that regression, but it was ignored as flaky.
## What
Re-enable `unified_exec_respects_workdir_override`.
## Validation
`just test -p codex-core unified_exec_respects_workdir_override`
* code-mode: extend test coverage to lock in cell lifecycle (#28468)
This PR establishes the intended behavior as an executable contract
before a refactor of the cell runtime begins. It also fixes cases where
a second observer or termination request could replace an existing
response channel and leave the original caller unresolved.
### Behavior codified
- A cell can yield output and subsequently resume to completion.
- A caller can run a cell until it has no immediately runnable work,
receive its accumulated output and outstanding tool-call IDs, and then
resume the same cell when the awaited work is available.
- Each cell admits one active observer:
- a second observer receives an explicit busy error
- the existing observer remains registered and is not displaced
- A natural result (conclusion of the js module) that has already
reached the cell controller wins over a later termination request.
- Otherwise, termination preempts execution and resolves both:
- the active observer, if present
- the caller requesting termination
- Repeated termination requests are rejected while termination is
already in progress.
- Terminal responses are sent only after outstanding callback work has
been handled:
- natural completion drains notifications and cancels outstanding tool
calls
- termination cancels and drains both notification and tool callbacks.
- Cell removal and cell_closed notification happen after callback
cleanup
* [codex] test exec relative additional permissions (#28587)
## Why
Review caught some would-be regressions in changes to unified_exec that
weren't surfaced in CI.
## What
Add coverage for requesting permissions through unified exec when there
are additional permissions. Previously this flow was only tested against
shell_command.
* Clarify model-generated and legacy app path types (#28577)
## Why
`ApiPathString` kind of implies that it can be used anywhere we pull a
path out of JSON, but it's not really appropriate for tool arguments
when the model might generate relative paths.
Prefer `String` for model-generated paths and we can handle the
conversion per feature for now and define a shared abstraction later if
it makes sense.
# What
Rename `ApiPathString` to `AppLegacyPathString` to clarify its role.
Expand the `path-types` skill to tell the model to leave tool args as
bare strings.
* Record invariants for path migration. (#28589)
## Why
Help Codex understand how to execute the migration to support cross-OS
paths.
## What
Expand the path-types skill with our goals and constraints.
* app-server: preserve target-native environment cwd (#28146)
## Why
app-server may run on a different OS from the selected exec-server
environment. Parsing that environment’s cwd with the Codex host’s path
rules prevents thread startup.
## What
Carry environment cwd values as `LegacyAppPathString` at the app-server
boundary and `PathUri` internally. Existing tool-call schemas and
relative-path behavior stay host-native; remaining local-only consumers
convert explicitly and leave follow-up TODOs.
The Wine integration test verifies app-server can start a thread and
complete an ordinary turn with a Windows environment cwd from Linux.
## Validation
- `bazel test //codex-rs/core/tests/remote_env_windows:smoke-test
--test_output=errors`
- focused app-server environment-selection and protocol schema tests
- scoped Clippy for `codex-core` and `codex-app-server-protocol`
* Tell codex about PathUri serde compat. (#28595)
This addresses another wrinkle I keep having to re-prompt codex about
when migrating to cross-OS paths.
* [codex] [1/4] Add recommended plugin endpoint cache (#28399)
Summary
- Add authenticated parsing for `/ps/plugins/suggested?scope=GLOBAL`,
including remote plugin and connector app identities.
- Validate, deduplicate, sort, and cap endpoint candidates before
caching them by backend and account identity.
- Deduplicate concurrent cache misses and warm recommendations from the
existing remote-installed-plugin refresh path used at startup and after
account changes.
- Keep endpoint results model-invisible in this PR; failures and
responses without `enabled: true` resolve to legacy mode.
Stack
- 1/3. Follow-up: #28400 generalizes plugin suggestion presentation
without activating endpoint recommendations.
- Final activation: #27704.
Validation
- `just test -p codex-core-plugins recommended_plugins`
- `just fix -p codex-core-plugins`
- `just fmt`
- `git diff --check`
* [codex] [2/4] Generalize plugin suggestion presentation (#28400)
Summary
- Add list-backed and developer-context presentations for plugin
suggestion candidates.
- Let tool planning, install validation, and request-tool copy follow
the selected presentation.
- Keep every production caller on the existing list-backed presentation,
preserving the current list tool, request schema, connector behavior,
and model-visible copy.
- Leave developer-context presentation latent until the final PR in the
stack.
Stack
- 2/3, based on #28399.
- Follow-up: #27704 activates endpoint recommendations.
Validation
- `just test -p codex-core request_plugin_install`
- `just test -p codex-core spec_plan`
- `just fix -p codex-core`
- `just fmt`
- `git diff --check`
* [codex] [3/4] Activate endpoint plugin recommendations (#27704)
Summary\n- Await endpoint recommendation selection while constructing
each authenticated turn, removing the first-turn cache race.\n- Snapshot
and filter endpoint candidates once per turn, then use that same set for
the bounded contextual user fragment, tool exposure, and exact install
validation.\n- Keep recommendation selection ephemeral: do not persist
recommendation state in or gate resumed threads on prior context.\n-
Hide the legacy list tool in endpoint mode and preserve legacy discovery
unchanged when the endpoint is disabled or unavailable.\n- Keep remote
plugin and connector app identities out of model-visible context and
attach them only to Codex-owned elicitation metadata.\n\nStack\n- 3/4,
based on #28400.\n- Endpoint client and cache: #28399.\n- Generalized
suggestion presentation: #28400.\n- Install-schema follow-up:
#28403.\n\nValidation\n- \n- \n- \n- \n- Full : 2,649 passed and 88
environment-dependent tests failed because this sandbox cannot write ,
nest Seatbelt, or locate auxiliary test binaries.
* core: render remote environment cwd natively (#28152)
## Why
Model-visible `<environment_context>` should match the environment of
the executor, not of the app server.
Stacked on #28146.
## What
- Keep selected environment cwd values as `PathUri` while building
environment context.
- Render cwd text using the path convention represented by the URI, with
the canonical URI as a fallback.
- Preserve compatibility with legacy `TurnContextItem.cwd` values when
reconstructing and diffing context.
- Extend the Wine-backed remote Windows test to assert that the model
sees `powershell` and `C:\windows`.
* [codex] [4/4] Simplify recommended plugin install schema (#28403)
## Summary
- Simplify recommendation-context `request_plugin_install` arguments to
`plugin_id` and `suggest_reason`.
- Derive plugin type and install action from the matched candidate while
preserving Codex-owned elicitation metadata.
- Keep the legacy list-backed schema unchanged and accept resumed calls
that still use `tool_id`.
## Stack
- #28399
- #28400
- #27704
- This PR
## Validation
- `just test -p codex-tools -p codex-core request_plugin_install` (25
passed)
- `just fix -p codex-tools -p codex-core`
- `just fmt`
- `git diff --check`
* PAC 1 - Add system proxy feature config surface (#26706)
## Summary
Introduces the default-off `respect_system_proxy` feature flag used to
gate first-class system PAC/proxy support for Codex-owned native
clients.
With the feature disabled or absent, behavior remains unchanged. This PR
establishes the configuration and managed-requirement surface; proxy
discovery and request routing are implemented by follow-up PRs.
## Configuration
User configuration uses the standard boolean feature form:
```toml
[features]
respect_system_proxy = true
```
Managed feature requirements use the corresponding boolean key. The
effective runtime configuration is exposed as a boolean and defaults to
`false`.
## Implementation
- Registers `respect_system_proxy` as an under-development, default-off
feature.
- Resolves user configuration and managed feature requirements into
`Config.respect_system_proxy`.
- Provides bootstrap resolution for startup paths that must evaluate the
feature before full configuration loading completes.
- Uses the standard feature CLI and config-editing behavior.
- Excludes `features.respect_system_proxy` from project-local
configuration.
- Updates the generated configuration schema.
## End-user behavior
- No networking behavior changes when the feature is absent or disabled.
- Enabling the feature makes the boolean available to the native
proxy-routing implementation in follow-up PRs.
- Repository-local configuration cannot enable the feature.
## Test coverage
Covers scalar configuration and CLI override resolution, managed
requirement constraints, bootstrap resolution, and project-local
filtering.
* Add thread recencyAt for sidebar ordering (#27910)
## Summary
Add a server-owned `recencyAt` timestamp and `recency_at` thread-list
sort key for product recency ordering while preserving the existing
meaning of `updatedAt` as the latest persisted thread mutation.
This is the server-side alternative to #27697. Rather than narrowing
`updatedAt`, clients can sort the sidebar by `recency_at` and continue
treating `updatedAt` as mutation time.
Paired Codex Apps PR:
[openai/openai#1024599](https://github.com/openai/openai/pull/1024599)
## Contract
- `recencyAt` initializes when a thread is created.
- A turn start advances `recencyAt` monotonically.
- Commentary, agent output, tool results, token/accounting updates, turn
completion, archive, unarchive, resume, and generic metadata writes do
not advance it.
- `updatedAt` retains its existing behavior and continues to advance for
persisted thread mutations.
- Current servers populate `recencyAt`; the response field is optional
in generated TypeScript so clients connected to older servers can fall
back to `updatedAt`.
- Filesystem-only fallback uses existing updated/mtime ordering when
SQLite is unavailable.
## Persistence and compatibility
Migration 0038 adds second- and millisecond-precision recency columns,
backfills them from the existing updated timestamp, creates list
indexes, and includes an insert trigger so older binaries writing to a
migrated database seed recency without causing later mutations to
advance it.
Generic metadata upserts preserve existing recency values. Turn-start
updates use a dedicated monotonic touch, and process-local allocation
keeps millisecond cursor values unique. State DB list, search, read,
filtered-list repair, rollout fallback propagation, and app-server
conversions all carry the new field.
## API
`Thread` responses include:
```ts
recencyAt?: number
```
`thread/list` and `thread/search` accept:
```json
{ "sortKey": "recency_at" }
```
Generated TypeScript and JSON schemas are included.
## Validation
- `just test -p codex-state` — 146 passed
- `just test -p codex-rollout` — 69 passed
- `just test -p codex-thread-store` — 81 passed
- `just test -p codex-app-server-protocol` — 231 passed
- Focused app-server list ordering, response mapping, archive/unarchive,
and resume lifecycle tests passed
- Scoped `just fix` for state, rollout, thread-store,
app-server-protocol, and app-server
- `just fmt`
- `git diff --check`
- Independent correctness, simplicity, elegance, security, and
test-quality reviews; actionable ordering, lifecycle, query-projection,
and timestamp-uniqueness findings were addressed
* Revert "Tell codex about PathUri serde compat. (#28595)" (#28627)
This reverts commit bd2a78632665006149d6e7df627c52ebc1b3464e, which
didn't capture all the nuance we need for this migration.
* [codex] Gate remote plugin catalog by auth (#28625)
## Summary
- Treat the remote global plugin catalog as active only when
`remote_plugin` is enabled and the current auth uses the Codex backend.
- Skip the local OpenAI curated marketplace for remote-enabled ChatGPT
users while preserving configured marketplaces.
- Keep the local curated marketplace for API-key users, unauthenticated
fallback, and ChatGPT users with `remote_plugin` disabled.
- Apply the same effective-remote gate to the remote
installed-marketplace cache.
## Root cause
The tool-suggestion discovery path unconditionally included the local
OpenAI curated marketplace. For remote-enabled ChatGPT users, that made
remote discovery additive: Codex parsed every local curated
`plugin.json` before also loading the remote catalog.
## Validation
- `just fmt`
- `cargo build -p codex-cli --bin codex`
- Targeted auth/feature matrix tests pass, including API-key auth with
`remote_plugin` enabled.
- Manual CLI validation confirmed:
- ChatGPT + remote off includes local curated.
- ChatGPT + remote on excludes local curated.
- API-key auth keeps local curated when remote is enabled.
- `just test -p codex-core-plugins`: 235 passed; one unrelated existing
marketplace test failed because it loaded the developer's home
marketplace configuration.
* [codex] core: restore absolute turn context cwd (#28629)
## Why
#28152 jumped the gun on moving the rollout format to store URIs, and
would likely break compat with some features that don't go through the
same types as the core logic.
## What
Make `TurnContextItem.cwd` an `AbsolutePathBuf` again, remove test added
for `PathUri` serialization in rollouts. Also drops a bunch of error
paths that are no longer needed.
* thread-store: fix response fixture compilation (#28642)
## Why
A `codex-thread-store` test fixture still constructs
`ResponseItem::FunctionCallOutput` without its required `metadata`
field, preventing the crate's test targets from compiling on `main`.
## What changed
- Set the fixture's response-item metadata to `None`.
## Testing
- `cargo check -p codex-thread-store --tests`
* [codex] Support object-valued plugin MCP manifests (#28580)
## Summary
This fixes plugin manifest parsing for MCP servers declared as an object
directly in `plugin.json`.
Before this change, Codex modeled `mcpServers` as only a string path,
for example:
```json
{
"name": "counter-sample",
"version": "1.1.1",
"mcpServers": "./.mcp.json"
}
```
Some migrated plugins instead provide the server map directly in the
manifest:
```json
{
"name": "counter-sample",
"version": "1.1.1",
"description": "Plugin that declares MCP servers in the manifest",
"mcpServers": {
"counter": {
"type": "http",
"url": "https://sample.example/counter/mcp"
}
}
}
```
That object form previously failed during install/load with an error
like:
```text
failed to parse plugin manifest: invalid type: map, expected a string
```
## What changed
- Add a manifest representation for `mcpServers` as either
`Path(Resource)` or `Object(map)`.
- Parse `plugin.json` `mcpServers` as either a string path or an object.
- Route object-valued MCP server maps through the existing plugin MCP
config parser instead of adding a second parser.
- Apply existing per-plugin MCP server policy to object-valued MCP
servers the same way as file-backed MCP servers.
- Include object-valued MCP server names in plugin telemetry/capability
metadata.
- Support object-valued MCP config for executor plugins without
requiring a `.mcp.json` filesystem read.
- Update the bundled plugin-creator validator and `plugin-json-spec.md`
so generated-plugin validation accepts the same object-valued shape.
## Compatibility
Existing plugin manifests that use `"mcpServers": "./.mcp.json"`
continue to work. Plugins can now also use the object shape shown above.
## Tests
Added coverage for the new manifest attribute shape at the install,
normal load, telemetry, and executor-provider layers:
- `install_accepts_manifest_mcp_server_objects`
- `load_plugins_loads_manifest_mcp_server_objects`
- `plugin_telemetry_metadata_uses_manifest_mcp_server_objects`
- `reads_manifest_object_config_without_executor_file_system_access`
Also smoke-tested the plugin-creator validator against both supported
forms:
- `mcpServers` as a direct object in `plugin.json`
- `mcpServers` as `"./.mcp.json"` with a companion `.mcp.json`
## Validation
- `just test -p codex-plugin`
- `just test -p codex-core-plugins`
- `just test -p codex-mcp-extension`
- `just bazel-lock-update`
- `just bazel-lock-check`
- `just fmt`
- `git diff --check`
- Focused rename/object-form rerun: `just test -p codex-core-plugins
manager::tests::load_plugins_loads_manifest_mcp_server_objects
manager::tests::plugin_telemetry_metadata_uses_manifest_mcp_server_objects
store::tests::install_accepts_manifest_mcp_server_objects`
- Focused executor rerun: `just test -p codex-mcp-extension
executor_plugin::provider::tests::reads_manifest_object_config_without_executor_file_system_access`
- `python3
codex-rs/skills/src/assets/samples/plugin-creator/scripts/validate_plugin.py
/private/tmp/codex-validator-object`
- `python3
codex-rs/skills/src/assets/samples/plugin-creator/scripts/validate_plugin.py
/private/tmp/codex-validator-path`
* code-mode: move cell state into library actor (#28599)
A code-mode cell is a single JavaScript execution that can produce
output, call tools, wait for asynchronous work, resume, or be
terminated. This PR extracts the existing per-cell run loop into a
dedicated actor that owns the cell’s lifecycle state. It is primarily an
ownership change rather than a new lifecycle contract: existing behavior
now has one clear implementation boundary.
### Architecture
The session service remains responsible for session-wide concerns:
allocating cell IDs, storing shared values, creating cells, and routing
requests to them.
Once a cell is created, its execution state belongs to its actor.
Callers interact with the actor through a handle. The actor receives two
kinds of input: runtime events and control requests.
A single event loop serializes these inputs and applies the lifecycle
rules. It tracks the current observer—the caller waiting for an
update—along with accumulated output, outstanding callbacks, runtime
state, yield deadlines, and termination progress. Observation,
termination, completion, and cleanup therefore have one consistent
owner.
When the runtime has no immediately runnable work and is waiting only on
timers or tool results, the actor can return accumulated output and
information about outstanding tool calls while keeping the cell
available to resume. On completion or termination, it performs the
appropriate callback cleanup before publishing the final result and
removing the cell from the session.
A small host interface connects the actor to session-owned facilities
such as tool dispatch, notifications, stored values, and final cell
removal, keeping those responsibilities outside the actor itself.
### Why
Previously, cell lifecycle state and coordination lived alongside
session management. The actor boundary makes each cell a self-contained
state machine with a single writer, while the service becomes a registry
and adapter around it.
This makes lifecycle behavior easier to reason about and test in
isolation. It also establishes a clean boundary for later changing where
cells run or how they communicate without recreating their lifecycle
rules.
* [codex] Test code-mode variable truncation (#28471)
## Summary
Code mode has two separate truncation points: the nested tool result
returned to JavaScript and the code-mode output later recorded for the
model. These tests now verify those behaviors independently.
- Report whether `result.output` was truncated before printing it.
- Verify omitted or sufficiently large nested limits produce `Variable
truncated: False`, while allowing the printed value to be truncated
downstream.
- Verify an explicit nested limit produces `Variable truncated: True`
when the command output exceeds it.
- Use a token-policy model fixture so downstream truncation is visible
as `…N tokens truncated…`.
- Align the explicit nested-truncation expectation with the warning
header.
This PR changes test coverage only; runtime truncation behavior is
unchanged.
## Validation
- `env -u CODEX_SANDBOX_NETWORK_DISABLED RUST_MIN_STACK=8388608 cargo
test -p codex-core --test all code_mode_exec -- --nocapture` (8 passed)
* Revert thread recencyAt for sidebar ordering (#28655)
## Why
Revert #27910 to remove the newly introduced thread `recencyAt`
persistence and API behavior from `main`.
## What changed
This reverts commit `fac3158c2a783095768076489815f361fa9b0db4`,
including the state migration, thread-store propagation, app-server API
surface, generated schemas, and related tests.
## Validation
Not run before opening; relying on CI for the initial fast signal.
* core: remove redundant TurnContext and Prompt fields (#28638)
## Why
`TurnContext` had accumulated dead fields and cached projections of
values already owned by its per-turn `Config` or `ModelInfo`. Keeping
both copies made ownership unclear and allowed artificial split-brain
states, such as a compatibility hash differing from the model metadata
it came from.
`Prompt` similarly carried a write-only personality after personality
selection had already been materialized into its base instructions.
This makes the canonical owner explicit: configuration-backed values
come from `config`, model-derived values come from `model_info`, and
prompts contain only data consumed by request construction.
## What changed
- Remove the unused `ghost_snapshot`, `codex_self_exe`, and
`thread_source` fields.
- Remove duplicate `comp_hash`, `truncation_policy`, `features`,
`shell_environment_policy`, `codex_linux_sandbox_exe`, `compact_prompt`,
and `tool_mode` fields.
- Read those values directly from `TurnContext::config` or
`TurnContext::model_info` at their consumers.
- Remove the write-only `Prompt::personality` field and its constructor
assignments.
- Preserve review-turn inheritance of the parent turn's shell policy,
Linux sandbox executable, and compact prompt through the review config.
## Testing
- `cargo check -p codex-core --tests`
* [codex] Persist built-in image results reported as generating (#28656)
## Why
#27920 stopped persisting image-generation items unless their status was
`completed`, preventing failed standalone extension items with empty
results from being saved. Built-in image generation can instead emit a
terminal `response.output_item.done` containing a complete base64 PNG
while the item status remains `generating`. In that case, app-server
emits no `savedPath`, so Codex Apps can render the inline image but
cannot expose a file artifact.
## What changed
- Persist image-generation items whenever `result` contains image data.
Failed terminal items still have empty results and remain unpersisted.
- Update the existing built-in image-generation integration test to
cover a terminal `generating` item and verify both `saved_path` and the
written PNG bytes.
## Validation
- Confirmed with a raw built-in websocket trace: the image progressed
through `in_progress`, `generating`, and `partial_image`, then emitted
one `response.output_item.done` with `status: "generating"` and a
complete PNG result.
- `just test -p codex-core builtin_image_generation_call_persisted` is
currently blocked before test execution by a pre-existing compile error
in `thread-store/src/thread_metadata_sync.rs:171`.
* Resume exec-server sessions after disconnect (#28512)
Supersedes #28288 (closed).
## Why
A short WebSocket interruption currently ends every client-side process
handle, even though exec-server keeps the server session and its
processes alive for a short time.
This is especially visible for executor-backed stdio MCP servers: a
temporary connection loss becomes a permanent `Transport closed` error.
The server already has the information needed to resume the session, but
the client opens a fresh session instead of using it.
This change reconnects below the process and MCP layers. Existing
process handles stay valid, missed output is recovered, and the same
server-side processes continue running.
## State machine
One logical `ExecServerClient` stays alive while its underlying RPC
connection changes generations.
```text
transport closes
+------------------------------------------------+
| v
+-------------+ +-------------+
| Connected | | Recovering |
+-------------+ +-------------+
^ |
| session resumed, processes caught up | retryable error
+------------------------------------------------+ loops until deadline
|
| deadline or permanent error
v
+-------------+
| Failed |
+-------------+
```
### `Connected`
- New RPC calls use the current connection.
- Process notifications are published in sequence order.
- A disconnect only starts recovery if it came from the current
connection generation. Late events from older generations cannot replace
the active connection.
### `Recovering`
- New calls wait instead of choosing a half-connected RPC client.
- Existing process handles, wake subscriptions, and event subscriptions
stay open.
- Streaming HTTP response bodies fail immediately because their byte
streams cannot be resumed safely.
- Recovery first waits for process starts that were already in flight. A
start whose result became ambiguous is cleaned up after reconnection
instead of being silently adopted.
- The client reconnects with the learned `session_id`. The server may
briefly report that the old connection is still attached, so that error
is retried until the detach finishes.
- The notification consumer starts before the resume handshake
completes. This prevents a busy process from filling the notification
queue and blocking the initialize response.
- Before installing the new connection, the client catches up every
recoverable process with `process/read`.
### `Failed`
- Recovery stops after 25 seconds or after a permanent error.
- Waiting calls are released with one stable disconnect error.
- Existing process sessions receive a terminal failure instead of
waiting forever.
## Recovering process events
Output, exit, and close events share one sequence. During normal
operation, the client buffers early events until every lower sequence
has been published.
After reconnection, the client reads each process starting after its
last published sequence:
1. Retained output chunks are inserted by sequence number.
2. Exit and close state are reconstructed in their sequence positions.
3. Events already received as live notifications are ignored as
duplicates.
4. Newly contiguous events are published in order.
5. If the server no longer retains enough output to fill a sequence gap,
only that process is terminated and failed. The recovered connection
remains usable for other processes.
The server reports its full next event sequence for unbounded reads,
including exit and close events. Closed processes remain readable for
the same 30-second window used to retain detached sessions.
## Other details
- Detached server sessions are retained for 30 seconds, leaving margin
around the client's 25-second recovery deadline.
- Session attach and detach update the active notification sender under
the same attachment lock, so an old connection cannot clear a newly
attached sender.
- A dedicated error code distinguishes the temporary "session is still
attached" race from permanent initialization errors.
- Process starts are identity-checked on both client and server. Cleanup
from an older start cannot remove a newer process that reused the same
ID.
- Mutating requests that were already in flight when the transport
closed are not replayed, because the client cannot know whether the
server applied them. Requests started after recovery is known wait for
the replacement connection.
- We assume the server/client version stays in sync (on the before/after
this PR)
## User impact
Long-running commands and stdio MCP servers can survive a temporary
exec-server WebSocket interruption without changing process IDs or
losing output produced during the outage.
* Back off registry retries during exec recovery (#28546)
## Why
PR #28512 retries a failed session recovery every 100 ms. Every Noise
recovery attempt first asks the environment registry for a fresh
connection bundle, even when the eventual failure comes from the
WebSocket or initialize handshake. During an outage, that could make
each disconnected client call the registry about 250 times during the
25-second recovery window.
## What changes
All retryable Noise recovery failures now use a separate backoff
schedule:
```text
base: 500 ms -> 1 s -> 2 s -> 4 s -> 5 s maximum
actual: 500-750 ms, 1-1.5 s, 2-3 s, 4-6 s, 5-7.5 s
```
The extra 0-50% is deterministic per-session jitter so disconnected
clients do not retry together. Direct WebSocket recovery keeps the
existing 100 ms retry because it does not re-enter the registry.
* Add join key for MAv2 inter-agent messages (#28561)
## Summary
This keeps inter-agent communication on the existing raw response item
path and adds a join key for MAv2 tool calls.
MAv2 `spawn_agent`, `send_message`, and `followup_task` now stamp the
originating tool call id into `ResponseItemMetadata.source_call_id` on
the raw `ResponseItem::AgentMessage`. App-server clients can join that
raw item back to the existing tool/activity event by call id, while
using the raw agent message's existing sender, receiver, and content
fields.
No new app-server `ThreadItem` or notification type is added.
## Tests
- `just fmt`
- `just write-app-server-schema`
- `just test -p codex-protocol`
- `just test -p codex-app-server-protocol`
- `just test -p codex-core
multi_agent_v2_spawn_returns_path_and_send_message_accepts_relative_path`
- `just test -p codex-core
multi_agent_v2_followup_task_completion_notifies_parent_on_every_turn`
- `just fix -p codex-protocol`
- `just fix -p codex-app-server-protocol`
- `just fix -p codex-core`
* app-server: keep the model cache warm (#28699)
## Why
The app server is long-lived, but its shared model cache otherwise
refreshes only when a caller needs it. Once the five-minute cache
expires, starting a thread or calling `model/list` can wait for
`/models` on the request path.
Refresh the cache in the background before it expires so foreground
callers normally use fresh local state.
## What changed
- Start an app-server worker that refreshes models immediately and then
every three minutes using the existing models-manager API.
- Hold only a weak reference to the models manager between refreshes, so
the worker does not extend its lifetime.
- Stop scheduling refreshes when the app-server lifecycle handle is shut
down or dropped. A refresh already in progress is allowed to finish.
- Adjust affected app-server test fixtures to distinguish the background
`/models` probe from the connection they are testing.
The existing models-manager cache, refresh strategies, auth handling,
ETag behavior, and concurrency semantics are unchanged.
## Testing
-
`models_refresh_worker::tests::refreshes_immediately_periodically_and_stops_when_dropped`
-
`suite::v2::remote_control::listen_off_honors_persisted_remote_control_enable`
-
`suite::v2::attestation::attestation_generate_round_trip_adds_header_to_responses_websocket_handshake`
* Replace SkillsManager with SkillsService (#28705)
## Why
Host skill discovery was still exposed as a manager even though it is a
process-owned service shared by sessions, the app-server catalog, and
file-watcher invalidation. The skills extension also consumed an ad hoc
loaded-skills wrapper instead of a named immutable snapshot.
## What changed
- replace `SkillsManager` with concrete `SkillsService`
- make the service cache and return immutable `HostSkillsSnapshot`
values
- migrate the skills extension host provider to the snapshot boundary
- migrate app-server catalog, watcher, and invalidation paths to the
service
This keeps the service limited to host discovery, caching, roots, and
invalidation. Catalog rendering and invocation remain extension
responsibilities for the next stacked change.
* [ez][codex-rs] Support apps._default.default_tools_approval_mode (#27965)
[from codex]
## Summary
- add `default_tools_approval_mode` to `[apps._default]` and expose it
through app-server v2 `config/read`
- apply it after managed, per-tool, and per-app approval settings,
before the built-in `auto` fallback
- document the precedence, regenerate config/app-server schemas, and add
unit plus end-to-end approval coverage
## Configuration
```toml
[apps._default]
default_tools_approval_mode = "prompt"
```
The effective precedence is managed requirements, tool-specific
`approval_mode`, app-specific `default_tools_approval_mode`,
`apps._default.default_tools_approval_mode`, then `auto`.
## Test plan
- `just write-config-schema`
- `just write-app-server-schema`
- `just write-app-server-schema --experimental`
- `just test -p codex-core app_tool_policy`
- `just test -p codex-core mcp_turn_metadata`
- `just test -p codex-config`
- `just test -p codex-app-server-protocol`
- `just test -p codex-app-server config_read_includes_apps`
- `just fix -p codex-config -p codex-core -p codex-app-server-protocol
-p codex-app-server`
- `just fmt`
* Run fs helper through Windows sandbox wrapper (#28359)
## Why
This is the final PR in the Windows fs-helper sandbox stack and contains
the actual bug fix.
The exec-server filesystem helper is a direct-spawn path: it asks
`SandboxManager` for a `SandboxExecRequest`, then launches the returned
argv itself. That works on macOS and Linux because the transformed argv
is already a self-contained sandbox wrapper. On Windows, the transformed
request carried `WindowsRestrictedToken` metadata, but the direct-spawn
fs-helper runner still launched the helper argv directly.
That means Windows filesystem built-ins backed by the fs-helper could
run with the parent Codex process permissions instead of the configured
Windows sandbox. This PR makes the direct-spawn transform produce a
self-contained Windows wrapper argv before fs-helper launches it.
## What Changed
- Added `SandboxManager::transform_for_direct_spawn()` for callers that
launch the returned argv themselves.
- Wrapped Windows restricted-token direct-spawn requests with `codex.exe
--run-as-windows-sandbox` and then marked the outer request as
unsandboxed, matching the macOS/Linux wrapper argv shape.
- Updated `exec-server/src/fs_sandbox.rs` to use the direct-spawn
transform for fs-helper launches.
- Materialized the inner `codex.exe --codex-run-as-fs-helper` executable
into `.sandbox-bin` so the sandboxed user can run it.
- Carrie…
dkropachev
pushed a commit
to dkropachev/codex
that referenced
this pull request
Jul 5, 2026
## Summary - disable the stable Apps feature in the generic `test_codex()` integration-test harness - keep Apps-specific tests explicit: their builders re-enable Apps and point it at a local mock server ## Why Generic tests that use dummy ChatGPT auth were also enabling the host-owned `codex_apps` MCP server. That made unrelated tests contact `chatgpt.com` and wait for MCP startup, causing the Bazel timeouts observed on openai#28368. The generic harness should be hermetic and should not start an external service that the test did not request. This is test-only; production Apps behavior is unchanged. The broader optional-MCP startup behavior is being handled separately in openai#28407. ## Testing - `just test -p codex-core -E 'test(pre_sampling_compact_runs_when_comp_hash_changes) | test(model_switch_to_smaller_model_updates_token_context_window) | test(codex_apps_file_params_upload_local_paths_before_mcp_tool_call)'` - `just fix -p codex-core` - `just fmt`
dkropachev
added a commit
to dkropachev/codex
that referenced
this pull request
Jul 6, 2026
* [codex] Record external agent import results (#28396)
## Summary
- restore `externalAgentConfig/import/progress` notifications while
keeping `externalAgentConfig/import/completed` as the must-deliver event
- persist completed external-agent config imports in state DB by
`importId`, including concrete success/failure details for config,
AGENTS.md, skills, plugins, MCP servers, subagents, hooks, commands, and
sessions
- add `externalAgentConfig/import/readHistories` so clients can recover
persisted import results after missing the live completion notification
- include `errorType` on import failures in protocol
responses/notifications and persisted DB JSON so future code can
classify failures without another wire/storage shape change
## Validation
- `git diff --check`
- `just test -p codex-state external_agent_config_imports`
- `just test -p codex-app-server-protocol`
- `CODEX_SQLITE_HOME=/private/tmp/codex-app-server-sqlite-read-details
just test -p codex-app-server
external_agent_config_import_sends_completion_notification_for_sync_only_import`
Also ran earlier broader checks before publishing:
- `just test -p codex-state`
-
`CODEX_SQLITE_HOME=/private/tmp/codex-app-server-external-agent-test-sqlite
just test -p codex-app-server external_agent_config`
- `just test -p codex-external-agent-migration`
* [codex] expose Bedrock credential source in account/read (#27751)
## Why
`account/read` currently reports only `type: "amazonBedrock"`, so
clients cannot distinguish a Codex-managed Bedrock API key from
credentials supplied by AWS. The app UI needs that distinction to render
the appropriate account state without duplicating provider-auth logic.
Credential-source selection belongs to the Bedrock model provider
because it already owns the precedence between managed Bedrock auth and
the external AWS credential path. This builds on #27443 and #27689.
## What changed
- Added `AmazonBedrockCredentialSource` with `codexManaged` and
`awsManaged` values.
- Included the selected credential source in
`ProviderAccount::AmazonBedrock` and the app-server `Account` response.
- Made `AmazonBedrockModelProvider::account_state()` classify the source
from its managed-auth state.
- Regenerated the app-server JSON and TypeScript schemas.
- Updated app-server account documentation and downstream TUI matches.
`codexManaged` means the provider found a managed Bedrock API key.
`awsManaged` identifies the provider's external AWS credential path; it
does not assert that the AWS credential chain has been validated.
## Testing
- Added model-provider coverage for Codex-managed precedence and
AWS-managed fallback.
- Added app-server protocol serialization coverage for both wire values.
- Added app-server integration coverage for both `account/read`
responses.
- `just test -p codex-protocol -p codex-model-provider -p
codex-app-server-protocol` (497 tests passed).
After rebasing onto #27711, the `codex-app-server` test target compiled
past the image-generation `PathUri` migration. Local linking was then
interrupted by disk exhaustion (`No space left on device`).
* [codex] Compress cold active rollouts (#28338)
## Why
The local rollout compression worker currently scans only
`archived_sessions`, so cold unarchived thread history remains expanded
indefinitely.
## What changed
- Scan `sessions` after `archived_sessions` within the existing worker
runtime budget.
- Update rollout compression coverage to require both cold active and
archived rollouts to be compressed while fresh active rollouts remain
plain.
The worker remains behind the disabled-by-default
`local_thread_store_compression` feature, and the existing seven-day
cold-file threshold is unchanged.
## Validation
- `just test -p codex-rollout` (69 passed)
- `just fmt`
- `git diff --check`
* feat: render typed envelopes for multi-agent v2 messages (#28368)
## Why
Multi-agent v2 messages need a consistent, model-visible envelope that
identifies what kind of interaction occurred, who sent it, and which
agent it targets. Previously, encrypted deliveries exposed only
`encrypted_content`, while child completion used the legacy
`<subagent_notification>` shape. That meant the client could not
consistently present `NEW_TASK`, `MESSAGE`, and `FINAL_ANSWER` using the
same format.
This change adds the routing envelope as plaintext while keeping task
and message payloads encrypted. No new Responses API field is required:
an encrypted delivery is represented as an `input_text` header
immediately followed by its existing `encrypted_content` item.
Every envelope now follows this shape:
```text
Message Type: <NEW_TASK | MESSAGE | FINAL_ANSWER>
Task name: <recipient agent path>
Sender: <author agent path>
Payload:
<message payload>
```
## Message types
### `NEW_TASK`
`NEW_TASK` is used when the recipient should begin a new turn, including
an initial `spawn_agent` task and a later `followup_task`.
For a root agent spawning `/root/worker`, the request contains a
plaintext envelope followed by the encrypted task:
```json
{
"type": "agent_message",
"author": "/root",
"recipient": "/root/worker",
"content": [
{
"type": "input_text",
"text": "Message Type: NEW_TASK\nTask name: /root/worker\nSender: /root\nPayload:\n"
},
{
"type": "encrypted_content",
"encrypted_content": "<encrypted task payload>"
}
]
}
```
Conceptually, the model receives:
```text
Message Type: NEW_TASK
Task name: /root/worker
Sender: /root
Payload:
Review the authentication changes and report any regressions.
```
### `MESSAGE`
`MESSAGE` is used for a queued `send_message` delivery. It communicates
with an existing agent without starting a new turn.
For `/root/worker` reporting progress to the root agent, the request
contains:
```json
{
"type": "agent_message",
"author": "/root/worker",
"recipient": "/root",
"content": [
{
"type": "input_text",
"text": "Message Type: MESSAGE\nTask name: /root\nSender: /root/worker\nPayload:\n"
},
{
"type": "encrypted_content",
"encrypted_content": "<encrypted message payload>"
}
]
}
```
Conceptually, the model receives:
```text
Message Type: MESSAGE
Task name: /root
Sender: /root/worker
Payload:
The protocol tests pass; I am checking the resume path now.
```
### `FINAL_ANSWER`
`FINAL_ANSWER` is emitted when a child agent reaches a terminal state
and reports its result to its parent. Completion payloads are already
available locally, so the complete envelope is represented as plaintext
rather than as a plaintext header plus encrypted content.
For `/root/worker` completing work for the root agent, the request
contains:
```json
{
"type": "agent_message",
"author": "/root/worker",
"recipient": "/root",
"content": [
{
"type": "input_text",
"text": "Message Type: FINAL_ANSWER\nTask name: /root\nSender: /root/worker\nPayload:\nNo regressions found."
}
]
}
```
The model-visible form is:
```text
Message Type: FINAL_ANSWER
Task name: /root
Sender: /root/worker
Payload:
No regressions found.
```
Errored, shut down, and missing agents also use `FINAL_ANSWER`, with a
terminal-status description in the payload.
## What changed
- Render `NEW_TASK` or `MESSAGE` in
`InterAgentCommunication::to_model_input_item`, based on whether the
encrypted delivery starts a turn.
- Replace the multi-agent v2 `<subagent_notification>` completion
payload with a model-visible `FINAL_ANSWER` envelope.
- Document `Task name`, `Sender`, and `Payload` consistently in the
multi-agent developer instructions.
- Prevent local-only history projections from treating an encrypted
message's plaintext header as the complete assistant message.
- Preserve rollout-trace interaction edges when an agent message
contains both plaintext and encrypted content.
Legacy multi-agent behavior remains unchanged.
## Verification
- `just test -p codex-protocol`
- `just test -p codex-rollout-trace`
- `just test -p codex-web-search-extension`
- `just test -p codex-core
encrypted_multi_agent_v2_spawn_sends_agent_message_to_child`
- `just test -p codex-core
plaintext_multi_agent_v2_completion_sends_agent_message`
- `just test -p codex-core
multi_agent_v2_followup_task_completion_notifies_parent_on_every_turn`
- `just test -p codex-core
multi_agent_v2_completion_queues_message_for_direct_parent`
* [tests] Keep Apps out of generic core test harness (#28508)
## Summary
- disable the stable Apps feature in the generic `test_codex()`
integration-test harness
- keep Apps-specific tests explicit: their builders re-enable Apps and
point it at a local mock server
## Why
Generic tests that use dummy ChatGPT auth were also enabling the
host-owned `codex_apps` MCP server. That made unrelated tests contact
`chatgpt.com` and wait for MCP startup, causing the Bazel timeouts
observed on #28368.
The generic harness should be hermetic and should not start an external
service that the test did not request. This is test-only; production
Apps behavior is unchanged. The broader optional-MCP startup behavior is
being handled separately in #28407.
## Testing
- `just test -p codex-core -E
'test(pre_sampling_compact_runs_when_comp_hash_changes) |
test(model_switch_to_smaller_model_updates_token_context_window) |
test(codex_apps_file_params_upload_local_paths_before_mcp_tool_call)'`
- `just fix -p codex-core`
- `just fmt`
* [codex] Clarify plugin load and runtime capability stages (#28472)
## Summary
Plugin loading and auth projection both previously produced
`PluginLoadOutcome`. That made an unfiltered load result look like
runtime-ready capabilities and generated capability summaries before
auth routing had run.
This change keeps loaded plugin records in the cache, applies the
current auth policy in `PluginsManager`, and only then builds
`PluginLoadOutcome` and its summaries. Auth changes still reuse the
cached disk load and re-resolve apps and MCP servers without reloading
plugins.
The updated tests cover cached auth changes and verify that capability
summaries match the effective app/MCP surface.
## Testing
- `just test -p codex-core-plugins`
- `just test -p codex-plugin`
- `just fix -p codex-core-plugins`
* core: surface terminal subagent errors to parent agents (#28375)
## Why
When a subagent exhausts its retries, it emits an `Error`, but the
generic task lifecycle then emits `TurnComplete(None)`. That completion
used to overwrite the subagent's `Errored` status with
`Completed(None)`, so the parent received an empty completion
notification.
This made a failed child look indistinguishable from a child that
completed without an answer. In unattended or long-running multi-agent
work, the root could silently continue without knowing that delegated
work failed or how to restart it.
## Behavior
Before, a terminal stream failure was reduced to an empty completion:
```text
<subagent_notification>
{"agent_path":"/root/worker","status":{"completed":null}}
</subagent_notification>
```
Now the parent receives the actual terminal error, bounded to 1,000
tokens, together with an actionable recovery hint:
```text
<subagent_notification>
{
"agent_path": "/root/worker",
"status": {
"errored": "stream disconnected before completion: stream closed before response.completed"
},
"next_action": "This agent's turn failed. If you still need this agent, use `followup_task` to give it another task."
}
</subagent_notification>
```
The notification remains queue-only: it does not wake the root or replay
the failed request. The root sees it at the next sampling boundary and
can use `followup_task` to start a new turn for that agent.
## What changed
- Added terminal-error precedence to the [agent status
reducer](https://github.com/openai/codex/blob/e95fcfe2bb6a02f1a75650afa20048859f556511/codex-rs/core/src/agent/status.rs#L23-L34),
so a closing `TurnComplete` cannot erase an immediately preceding
`Errored` status.
- Made MultiAgentV2 completion forwarding use the retained session
status instead of re-deriving `Completed(None)` from the final event.
- Extended the [subagent notification
fragment](https://github.com/openai/codex/blob/e95fcfe2bb6a02f1a75650afa20048859f556511/codex-rs/core/src/context/subagent_notification.rs#L6-L60)
with a `next_action` for terminal errors and a hard cap on model-visible
error text.
- Kept successful completions and interrupted turns unchanged.
## Verification
- Added a status-reducer test proving that `Errored` survives the
trailing `TurnComplete`.
- Added an integration test that exhausts a subagent's stream retries
and verifies the exact `agent_message` delivered to the parent,
including the error and `followup_task` guidance.
- Re-ran the existing successful-completion and interrupted-turn
notification tests.
* perf(config): defer remote sandbox hostname lookup (#28542)
## Why
[#18763](https://github.com/openai/codex/pull/18763) added canonical
hostname resolution for `remote_sandbox_config`. Requirements
composition currently performs that synchronous DNS lookup on every
fresh process, even when none of the loaded requirements layers contains
`[[remote_sandbox_config]]`. On hosts with slow local DNS resolution,
this can add several seconds to Codex startup.
## What
- defer hostname resolution until a parsed requirements layer actually
contains `remote_sandbox_config`
- cache the resolver result once per requirements composition,
preserving the existing single-lookup behavior across multiple layers
- keep the existing FQDN resolution and per-layer requirements
precedence unchanged
- cover both the ordinary no-lookup path and the multi-layer
single-lookup path
## How to Test
On a host where local canonical-name resolution is slow:
1. Start Codex without `[[remote_sandbox_config]]` in any managed
requirements layer and confirm startup no longer waits for hostname
resolution.
2. Add a matching `[[remote_sandbox_config]]` entry and confirm its
`allowed_sandbox_modes` still overrides the layer's top-level value.
3. Add remote sandbox entries to multiple requirements layers and
confirm precedence remains unchanged while the hostname is resolved only
once.
Targeted tests:
- `just test -p codex-config hostname_resolver`
- `just test -p codex-config` (181 passed)
* path-uri: clarify invalid host path errors (#28473)
## Why
Ensure a consistent string format when exposing path conversion errors
to the model.
## What
- Render `PathUriParseError::InvalidFileUriPath` as `'$PATH' is invalid
on '$OS'`.
* fix(tui): restore TUI after suspend (#28342)
## Why
On Linux, suspending Codex with `Ctrl+Z` and returning with `fg` can
leave the composer misaligned or inject terminal response bytes such as
focus reports into the prompt. Shell job-control output moves the cursor
while Codex is suspended, and terminal input polling can race with the
responses used to restore the inline viewport.
Fixes #26564.
## What changed
- preserve and restore keyboard reporting without disturbing the parent
terminal stack
- pause terminal event polling while Codex is suspended and flush
buffered input before resuming it
- force crossterm's cached raw-mode state back in sync after the shell
completes its `fg` handoff
- probe the actual post-`fg` cursor position with the tolerant
terminal-response parser, then realign the inline viewport before
redrawing
## How to Test
1. On Linux, start the development TUI with `just c`.
2. Type text into the composer without submitting it.
3. Press `Ctrl+Z`, run any harmless shell command, then run `fg`.
4. Confirm the composer redraws below the shell output, the draft text
is preserved, and no raw escape sequences appear.
5. Repeat the suspend/resume cycle and confirm normal typing still
works.
Targeted tests:
- `cargo test -p codex-tui --lib parses_cursor_position_as_zero_based -j
1`
- `cargo test -p codex-tui --lib tui::event_stream::tests -j 1`
* [codex] exec-server: stream files in chunks (#28354)
## Why
`fs/readFile` buffers the entire file in one response, which makes large
remote reads expensive and prevents callers from applying backpressure.
We need an opt-in streaming path with bounded block sizes while
preserving the existing single-call API for small and sandboxed reads.
## What changed
- Add `ExecServerClient::stream`, returning a named `FileReadStream`
that implements `futures::Stream` and yields immutable 1 MiB byte
blocks.
- Add internal `fs/open`, `fs/readBlock`, and `fs/close` RPCs.
`fs/readBlock` accepts an explicit offset and length.
- Keep unsandboxed files open between block reads, cap open handles per
connection, and clean them up on EOF, error, stream drop, explicit
close, or connection shutdown.
- Reject platform-sandboxed streaming opens instead of turning the
one-shot sandbox helper into a persistent server. Existing `fs/readFile`
behavior is unchanged.
## Testing
- `just test -p codex-exec-server`
- Integration coverage for 1 MiB chunking, exact block-boundary EOF,
sandbox rejection, and continued reads from the opened file after path
replacement.
- Handle-manager coverage for non-sequential offsets, variable block
lengths, the 128-handle limit, and capacity release after close.
* chore: side prompt (#28553)
Fix side bug with prompt
* [codex-app-server-test-client & codex-app-server] Plugin Usage Analytics Smoke Test (#27099)
## This PR
The original [combined remote plugin analytics PR
#26281](https://github.com/openai/codex/pull/26281) mixed reusable
analytics test infrastructure, two manual smoke workflows, a metadata
refactor, and the final identity behavior. This PR establishes a
non-mutating end-to-end plugin smoke workflow before any analytics
identity semantics change.
- Add `plugin-analytics-smoke` to the existing app-server test client.
- Exercise plugin disable, enable, and use through production app-server
RPC paths.
- Isolate config writes in a temporary file and use a loopback Responses
API server.
- Capture analytics without sending them to the production analytics
backend.
- Validate the current local `plugin_id`, names, capability metadata,
thread, turn, and model fields.
This is intentionally a baseline smoke workflow. It does not assert
`remote_plugin_id`; the final PR will update it when that field exists.
Review this PR as the net diff against #27093.
## Testing
- The test-client target compiles successfully.
- The combined reference branch exercised the manual smoke against the
live remote plugin service.
- CI is green across the required platform matrix.
## Split Overview
```text
main
├── #27093 Debug analytics capture
│ └── #27099 Non-mutating plugin smoke ← you are here
│ └── #27100 Remote install/uninstall smoke
└── #27102 Plugin telemetry metadata refactor
After #27093, #27099, #27100, and #27102 merge:
└── Final PR: add remote_plugin_id to plugin analytics
```
Review order and dependencies:
1. [#27093 Add debug-only analytics event
capture](https://github.com/openai/codex/pull/27093) (based on `main`)
2. [#27099 Add a plugin analytics smoke
workflow](https://github.com/openai/codex/pull/27099) **(this PR,
stacked on #27093)**
3. [#27100 Add a remote plugin analytics mutation smoke
workflow](https://github.com/openai/codex/pull/27100) (stacked on this
PR)
4. [#27102 Centralize plugin telemetry metadata
construction](https://github.com/openai/codex/pull/27102) (independent,
based on `main`)
5. Final remote-ID behavior PR (created after PRs 1-4 merge)
The original [#26281](https://github.com/openai/codex/pull/26281)
remains open as the green aggregate reference until the final PR is
published.
* fix(tui): highlight C++ module files (#28554)
## Why
Codex syntax-highlights diffs for conventional C++ extensions such as
`.cpp` and `.cxx`, but C++ module interface files using `.cppm`, `.ixx`,
or `.cxxm` fall back to plain diff coloring. The bundled syntax set
already includes C++, but it does not resolve those module extensions by
itself.
Closes #28223.
## What changed
- map `.cppm`, `.ixx`, and `.cxxm` to the existing `cpp` syntax in
`render/highlight.rs`
- extend alias-resolution coverage for all three module extensions
- verify `.cpp`, `.cppm`, `.ixx`, and `.cxxm` diffs produce
syntax-highlighted RGB spans while unknown extensions retain the plain
fallback
- snapshot the syntax-colored token segmentation for the supported C++
module extensions
## How to Test
1. Ask Codex to create or modify a C++ module interface file using
`.cppm`, `.ixx`, or `.cxxm`.
2. Confirm C++ tokens in the rendered diff receive syntax colors instead
of only the red/green diff treatment.
3. Modify an equivalent `.cpp` file and confirm its existing
highlighting remains unchanged.
4. Modify a file with an unknown extension and confirm it still uses the
plain diff fallback.
Targeted tests:
- `just test -p codex-tui -E
'test(find_syntax_resolves_languages_and_aliases) |
test(cpp_module_extensions_use_cpp_highlighting) |
test(unknown_extension_falls_back_without_syntax_highlighting)'`
* [codex] Warn clearly when code mode output is truncated (#28467)
## Summary
- make `formatted_truncate_text` prepend `Warning: truncated output
(original token count: N)` above the existing `Total output lines`
header
- update direct formatter, unified-exec, user-shell, and code-mode
expectations
- add core unit coverage that runs in Bazel without requiring the
skipped V8-backed code-mode integration suite
## Validation
- `cargo test -p codex-utils-output-truncation -- --nocapture` (17
passed)
- `cargo test -p codex-core --lib
truncated_text_output_starts_with_warning -- --nocapture`
- `cargo test -p codex-core --test all
clamps_model_requested_max_output_tokens_to_policy -- --nocapture` (2
passed)
- `cargo test -p codex-core --test all
unified_exec_formats_large_output_summary -- --nocapture`
- `cargo test -p codex-core --test all
user_shell_command_output_is_truncated_in_history -- --nocapture`
- Bazel CI exercises the shared formatter and downstream integration
expectations
* Add incremental thread history changes
Add ThreadHistoryBuilder APIs for collecting incremental thread item and turn changes while applying rollout items.
Batch handling coalesces repeated changes so callers can get the latest incremental thread item changes for a set of rollout items without rebuilding full history.
* feat(tui): add rate-limit reset redemption to /usage (#28154)
## Why
Codex users can earn personal rate-limit reset credits, but the CLI does
not currently provide a way to view or redeem them. The `/usage` command
restored in #27925 is intended to be the entry point for usage-related
actions, so reset redemption belongs there rather than in a separate
dashed slash command.
Depends on #28143 for the app-server and backend-client reset-credit
APIs.
## What changed
- Turn bare `/usage` into a menu with entries for token activity and
earned rate-limit resets while preserving `/usage daily`, `/usage
weekly`, and `/usage cumulative`.
- Add loading, empty, confirmation, success, retry, and error states
with a caller-generated UUID idempotency key reused across retries of
the same logical reset.
- Show an availability hint only for backend-classified rate-limit
errors with credits available.
- Hide the reset entry for workspace accounts.
## Validation
- `just test -p codex-tui chatwidget::tests::usage` — 19 passed.
- `just fix -p codex-tui` — passed.
- `just fmt` — passed.
- `cargo insta pending-snapshots` from `codex-rs/tui` — no pending
snapshots.
## Examples
<img width="1168" height="304" alt="image"
src="https://github.com/user-attachments/assets/caa4c1e3-e996-494d-ae17-50b521f5dce8"
/>
<img width="908" height="260" alt="image"
src="https://github.com/user-attachments/assets/e38a726b-77cc-4bd0-9ea8-9f3ad21c5768"
/>
### Reset flow
<img width="1509" height="312" alt="image"
src="https://github.com/user-attachments/assets/d987013c-78a5-48a2-ad8d-c61ad267a327"
/>
<img width="585" height="190" alt="image"
src="https://github.com/user-attachments/assets/de32be19-79b9-4a3e-8574-6f1c208c98ae"
/>
<img width="600" height="210" alt="image"
src="https://github.com/user-attachments/assets/88a165cf-796d-4fdc-a7bc-ea89917573da"
/>
<img width="512" height="193" alt="image"
src="https://github.com/user-attachments/assets/d2353998-5aa8-442e-a5f8-3a8a5b832753"
/>
* ci: run code-mode unit tests on all bazel targets (#28562)
## Why
V8 should be stable under Bazel, so the `codex-code-mode` unit tests
should run across the Bazel platform matrix. If these tests prove
unstable, we should fix the tests rather than exclude them from CI.
## What changed
- Remove the explicit `//codex-rs/code-mode:code-mode-unit-tests`
exclusion from the macOS and Linux Bazel test jobs.
- Remove the same exclusion from the native Windows post-merge job.
- Keep the existing Windows gnullvm shard coverage.
## Bazel test coverage
The target contains 26 unit tests. A fresh uncached local Bazel
execution ran all 26 with 0 failures, 0 ignored tests, and 0 filtered
tests.
PR Bazel CI selected the target on every enabled platform and reported a
cached pass:
| Platform | Passing CI job |
| --- | --- |
| macOS aarch64 | [Bazel test
passed](https://github.com/openai/codex/actions/runs/27636617545/job/81725447804)
|
| macOS x86_64 | [Bazel test passed in
2.2s](https://github.com/openai/codex/actions/runs/27636617545/job/81725448008)
|
| Linux GNU | [Bazel test passed in
0.4s](https://github.com/openai/codex/actions/runs/27636617545/job/81725447898)
|
| Linux musl | [Bazel test passed in
0.4s](https://github.com/openai/codex/actions/runs/27636617545/job/81725448117)
|
| Windows gnullvm | [Bazel test passed in shard 4/4 in
1.6s](https://github.com/openai/codex/actions/runs/27636617545/job/81725448166)
|
* [codex] Route MCP file uploads through environment filesystem (#27923)
## Why
Codex Apps tools can mark arguments with `openai/fileParams`, but the
execution path resolved and opened those files directly on the host.
That bypassed the selected turn environment and prevented annotated file
arguments from working with remote environments.
## What changed
- resolve annotated file arguments against the primary turn environment
- read file metadata and contents through that environment's sandboxed
`ExecutorFileSystem`
- reject files over the 512 MiB limit from metadata before reading or
transferring them
- retain the buffered upload-size check as defense in depth
- make the OpenAI upload API accept a filename and buffered contents
instead of owning local filesystem access
- describe the model-visible argument as a path in the primary
environment
This builds on #27927, which added `size` to internal filesystem
metadata.
## Testing
- `just test -p codex-api upload_openai_file_returns_canonical_uri`
- `just test -p codex-mcp
tool_with_model_visible_input_schema_masks_file_params`
- `just test -p codex-core mcp_openai_file`
- `just test -p codex-core
codex_apps_file_params_upload_environment_files_before_mcp_tool_call`
* [codex-app-server-test-client] Plugin Install/Uninstall Analytics Smoke Test (#27100)
## This PR
The original [combined remote plugin analytics PR
#26281](https://github.com/openai/codex/pull/26281) mixed reusable
analytics test infrastructure, two manual smoke workflows, a metadata
refactor, and the final identity behavior. This PR adds the
account-mutating validation workflow separately so its cleanup and
recovery guarantees can be reviewed without the final analytics behavior
change.
- Add a manually invoked remote plugin install/uninstall smoke workflow.
- Require explicit account-mutation confirmation and an initially
uninstalled plugin.
- Validate the current `codex_plugin_installed` contract, where
`plugin_id` is the backend ID.
- Restore and verify the original uninstalled state, with a dedicated
recovery command.
This baseline intentionally does not require `codex_plugin_uninstalled`,
because production does not emit that event yet. The final PR will
update this smoke to require local `plugin_id`, `remote_plugin_id`, and
uninstall emission. Review this PR as the net diff against #27099.
## Testing
- `just test -p codex-app-server-test-client` (3 focused
capture/validation tests passed)
- The live workflow was previously exercised on the green combined
reference branch, and the original uninstalled account state was
restored.
- CI is green across the required platform matrix.
## Split Overview
```text
main
├── #27093 Debug analytics capture
│ └── #27099 Non-mutating plugin smoke
│ └── #27100 Remote install/uninstall smoke ← you are here
└── #27102 Plugin telemetry metadata refactor
After #27093, #27099, #27100, and #27102 merge:
└── Final PR: add remote_plugin_id to plugin analytics
```
Review order and dependencies:
1. [#27093 Add debug-only analytics event
capture](https://github.com/openai/codex/pull/27093) (based on `main`)
2. [#27099 Add a plugin analytics smoke
workflow](https://github.com/openai/codex/pull/27099) (stacked on
#27093)
3. [#27100 Add a remote plugin analytics mutation smoke
workflow](https://github.com/openai/codex/pull/27100) **(this PR,
stacked on #27099)**
4. [#27102 Centralize plugin telemetry metadata
construction](https://github.com/openai/codex/pull/27102) (independent,
based on `main`)
5. Final remote-ID behavior PR (created after PRs 1-4 merge)
The original [#26281](https://github.com/openai/codex/pull/26281)
remains open as the green aggregate reference until the final PR is
published.
* [codex] re-enable absolute workdir integration test (#28581)
## Why
In #28146 I missed the invariant that an absolute `exec_command` workdir
must override the environment cwd. The existing integration test would
have caught that regression, but it was ignored as flaky.
## What
Re-enable `unified_exec_respects_workdir_override`.
## Validation
`just test -p codex-core unified_exec_respects_workdir_override`
* code-mode: extend test coverage to lock in cell lifecycle (#28468)
This PR establishes the intended behavior as an executable contract
before a refactor of the cell runtime begins. It also fixes cases where
a second observer or termination request could replace an existing
response channel and leave the original caller unresolved.
### Behavior codified
- A cell can yield output and subsequently resume to completion.
- A caller can run a cell until it has no immediately runnable work,
receive its accumulated output and outstanding tool-call IDs, and then
resume the same cell when the awaited work is available.
- Each cell admits one active observer:
- a second observer receives an explicit busy error
- the existing observer remains registered and is not displaced
- A natural result (conclusion of the js module) that has already
reached the cell controller wins over a later termination request.
- Otherwise, termination preempts execution and resolves both:
- the active observer, if present
- the caller requesting termination
- Repeated termination requests are rejected while termination is
already in progress.
- Terminal responses are sent only after outstanding callback work has
been handled:
- natural completion drains notifications and cancels outstanding tool
calls
- termination cancels and drains both notification and tool callbacks.
- Cell removal and cell_closed notification happen after callback
cleanup
* [codex] test exec relative additional permissions (#28587)
## Why
Review caught some would-be regressions in changes to unified_exec that
weren't surfaced in CI.
## What
Add coverage for requesting permissions through unified exec when there
are additional permissions. Previously this flow was only tested against
shell_command.
* Clarify model-generated and legacy app path types (#28577)
## Why
`ApiPathString` kind of implies that it can be used anywhere we pull a
path out of JSON, but it's not really appropriate for tool arguments
when the model might generate relative paths.
Prefer `String` for model-generated paths and we can handle the
conversion per feature for now and define a shared abstraction later if
it makes sense.
# What
Rename `ApiPathString` to `AppLegacyPathString` to clarify its role.
Expand the `path-types` skill to tell the model to leave tool args as
bare strings.
* Record invariants for path migration. (#28589)
## Why
Help Codex understand how to execute the migration to support cross-OS
paths.
## What
Expand the path-types skill with our goals and constraints.
* app-server: preserve target-native environment cwd (#28146)
## Why
app-server may run on a different OS from the selected exec-server
environment. Parsing that environment’s cwd with the Codex host’s path
rules prevents thread startup.
## What
Carry environment cwd values as `LegacyAppPathString` at the app-server
boundary and `PathUri` internally. Existing tool-call schemas and
relative-path behavior stay host-native; remaining local-only consumers
convert explicitly and leave follow-up TODOs.
The Wine integration test verifies app-server can start a thread and
complete an ordinary turn with a Windows environment cwd from Linux.
## Validation
- `bazel test //codex-rs/core/tests/remote_env_windows:smoke-test
--test_output=errors`
- focused app-server environment-selection and protocol schema tests
- scoped Clippy for `codex-core` and `codex-app-server-protocol`
* Tell codex about PathUri serde compat. (#28595)
This addresses another wrinkle I keep having to re-prompt codex about
when migrating to cross-OS paths.
* [codex] [1/4] Add recommended plugin endpoint cache (#28399)
Summary
- Add authenticated parsing for `/ps/plugins/suggested?scope=GLOBAL`,
including remote plugin and connector app identities.
- Validate, deduplicate, sort, and cap endpoint candidates before
caching them by backend and account identity.
- Deduplicate concurrent cache misses and warm recommendations from the
existing remote-installed-plugin refresh path used at startup and after
account changes.
- Keep endpoint results model-invisible in this PR; failures and
responses without `enabled: true` resolve to legacy mode.
Stack
- 1/3. Follow-up: #28400 generalizes plugin suggestion presentation
without activating endpoint recommendations.
- Final activation: #27704.
Validation
- `just test -p codex-core-plugins recommended_plugins`
- `just fix -p codex-core-plugins`
- `just fmt`
- `git diff --check`
* [codex] [2/4] Generalize plugin suggestion presentation (#28400)
Summary
- Add list-backed and developer-context presentations for plugin
suggestion candidates.
- Let tool planning, install validation, and request-tool copy follow
the selected presentation.
- Keep every production caller on the existing list-backed presentation,
preserving the current list tool, request schema, connector behavior,
and model-visible copy.
- Leave developer-context presentation latent until the final PR in the
stack.
Stack
- 2/3, based on #28399.
- Follow-up: #27704 activates endpoint recommendations.
Validation
- `just test -p codex-core request_plugin_install`
- `just test -p codex-core spec_plan`
- `just fix -p codex-core`
- `just fmt`
- `git diff --check`
* [codex] [3/4] Activate endpoint plugin recommendations (#27704)
Summary\n- Await endpoint recommendation selection while constructing
each authenticated turn, removing the first-turn cache race.\n- Snapshot
and filter endpoint candidates once per turn, then use that same set for
the bounded contextual user fragment, tool exposure, and exact install
validation.\n- Keep recommendation selection ephemeral: do not persist
recommendation state in or gate resumed threads on prior context.\n-
Hide the legacy list tool in endpoint mode and preserve legacy discovery
unchanged when the endpoint is disabled or unavailable.\n- Keep remote
plugin and connector app identities out of model-visible context and
attach them only to Codex-owned elicitation metadata.\n\nStack\n- 3/4,
based on #28400.\n- Endpoint client and cache: #28399.\n- Generalized
suggestion presentation: #28400.\n- Install-schema follow-up:
#28403.\n\nValidation\n- \n- \n- \n- \n- Full : 2,649 passed and 88
environment-dependent tests failed because this sandbox cannot write ,
nest Seatbelt, or locate auxiliary test binaries.
* core: render remote environment cwd natively (#28152)
## Why
Model-visible `<environment_context>` should match the environment of
the executor, not of the app server.
Stacked on #28146.
## What
- Keep selected environment cwd values as `PathUri` while building
environment context.
- Render cwd text using the path convention represented by the URI, with
the canonical URI as a fallback.
- Preserve compatibility with legacy `TurnContextItem.cwd` values when
reconstructing and diffing context.
- Extend the Wine-backed remote Windows test to assert that the model
sees `powershell` and `C:\windows`.
* [codex] [4/4] Simplify recommended plugin install schema (#28403)
## Summary
- Simplify recommendation-context `request_plugin_install` arguments to
`plugin_id` and `suggest_reason`.
- Derive plugin type and install action from the matched candidate while
preserving Codex-owned elicitation metadata.
- Keep the legacy list-backed schema unchanged and accept resumed calls
that still use `tool_id`.
## Stack
- #28399
- #28400
- #27704
- This PR
## Validation
- `just test -p codex-tools -p codex-core request_plugin_install` (25
passed)
- `just fix -p codex-tools -p codex-core`
- `just fmt`
- `git diff --check`
* PAC 1 - Add system proxy feature config surface (#26706)
## Summary
Introduces the default-off `respect_system_proxy` feature flag used to
gate first-class system PAC/proxy support for Codex-owned native
clients.
With the feature disabled or absent, behavior remains unchanged. This PR
establishes the configuration and managed-requirement surface; proxy
discovery and request routing are implemented by follow-up PRs.
## Configuration
User configuration uses the standard boolean feature form:
```toml
[features]
respect_system_proxy = true
```
Managed feature requirements use the corresponding boolean key. The
effective runtime configuration is exposed as a boolean and defaults to
`false`.
## Implementation
- Registers `respect_system_proxy` as an under-development, default-off
feature.
- Resolves user configuration and managed feature requirements into
`Config.respect_system_proxy`.
- Provides bootstrap resolution for startup paths that must evaluate the
feature before full configuration loading completes.
- Uses the standard feature CLI and config-editing behavior.
- Excludes `features.respect_system_proxy` from project-local
configuration.
- Updates the generated configuration schema.
## End-user behavior
- No networking behavior changes when the feature is absent or disabled.
- Enabling the feature makes the boolean available to the native
proxy-routing implementation in follow-up PRs.
- Repository-local configuration cannot enable the feature.
## Test coverage
Covers scalar configuration and CLI override resolution, managed
requirement constraints, bootstrap resolution, and project-local
filtering.
* Add thread recencyAt for sidebar ordering (#27910)
## Summary
Add a server-owned `recencyAt` timestamp and `recency_at` thread-list
sort key for product recency ordering while preserving the existing
meaning of `updatedAt` as the latest persisted thread mutation.
This is the server-side alternative to #27697. Rather than narrowing
`updatedAt`, clients can sort the sidebar by `recency_at` and continue
treating `updatedAt` as mutation time.
Paired Codex Apps PR:
[openai/openai#1024599](https://github.com/openai/openai/pull/1024599)
## Contract
- `recencyAt` initializes when a thread is created.
- A turn start advances `recencyAt` monotonically.
- Commentary, agent output, tool results, token/accounting updates, turn
completion, archive, unarchive, resume, and generic metadata writes do
not advance it.
- `updatedAt` retains its existing behavior and continues to advance for
persisted thread mutations.
- Current servers populate `recencyAt`; the response field is optional
in generated TypeScript so clients connected to older servers can fall
back to `updatedAt`.
- Filesystem-only fallback uses existing updated/mtime ordering when
SQLite is unavailable.
## Persistence and compatibility
Migration 0038 adds second- and millisecond-precision recency columns,
backfills them from the existing updated timestamp, creates list
indexes, and includes an insert trigger so older binaries writing to a
migrated database seed recency without causing later mutations to
advance it.
Generic metadata upserts preserve existing recency values. Turn-start
updates use a dedicated monotonic touch, and process-local allocation
keeps millisecond cursor values unique. State DB list, search, read,
filtered-list repair, rollout fallback propagation, and app-server
conversions all carry the new field.
## API
`Thread` responses include:
```ts
recencyAt?: number
```
`thread/list` and `thread/search` accept:
```json
{ "sortKey": "recency_at" }
```
Generated TypeScript and JSON schemas are included.
## Validation
- `just test -p codex-state` — 146 passed
- `just test -p codex-rollout` — 69 passed
- `just test -p codex-thread-store` — 81 passed
- `just test -p codex-app-server-protocol` — 231 passed
- Focused app-server list ordering, response mapping, archive/unarchive,
and resume lifecycle tests passed
- Scoped `just fix` for state, rollout, thread-store,
app-server-protocol, and app-server
- `just fmt`
- `git diff --check`
- Independent correctness, simplicity, elegance, security, and
test-quality reviews; actionable ordering, lifecycle, query-projection,
and timestamp-uniqueness findings were addressed
* Revert "Tell codex about PathUri serde compat. (#28595)" (#28627)
This reverts commit bd2a78632665006149d6e7df627c52ebc1b3464e, which
didn't capture all the nuance we need for this migration.
* [codex] Gate remote plugin catalog by auth (#28625)
## Summary
- Treat the remote global plugin catalog as active only when
`remote_plugin` is enabled and the current auth uses the Codex backend.
- Skip the local OpenAI curated marketplace for remote-enabled ChatGPT
users while preserving configured marketplaces.
- Keep the local curated marketplace for API-key users, unauthenticated
fallback, and ChatGPT users with `remote_plugin` disabled.
- Apply the same effective-remote gate to the remote
installed-marketplace cache.
## Root cause
The tool-suggestion discovery path unconditionally included the local
OpenAI curated marketplace. For remote-enabled ChatGPT users, that made
remote discovery additive: Codex parsed every local curated
`plugin.json` before also loading the remote catalog.
## Validation
- `just fmt`
- `cargo build -p codex-cli --bin codex`
- Targeted auth/feature matrix tests pass, including API-key auth with
`remote_plugin` enabled.
- Manual CLI validation confirmed:
- ChatGPT + remote off includes local curated.
- ChatGPT + remote on excludes local curated.
- API-key auth keeps local curated when remote is enabled.
- `just test -p codex-core-plugins`: 235 passed; one unrelated existing
marketplace test failed because it loaded the developer's home
marketplace configuration.
* [codex] core: restore absolute turn context cwd (#28629)
## Why
#28152 jumped the gun on moving the rollout format to store URIs, and
would likely break compat with some features that don't go through the
same types as the core logic.
## What
Make `TurnContextItem.cwd` an `AbsolutePathBuf` again, remove test added
for `PathUri` serialization in rollouts. Also drops a bunch of error
paths that are no longer needed.
* thread-store: fix response fixture compilation (#28642)
## Why
A `codex-thread-store` test fixture still constructs
`ResponseItem::FunctionCallOutput` without its required `metadata`
field, preventing the crate's test targets from compiling on `main`.
## What changed
- Set the fixture's response-item metadata to `None`.
## Testing
- `cargo check -p codex-thread-store --tests`
* [codex] Support object-valued plugin MCP manifests (#28580)
## Summary
This fixes plugin manifest parsing for MCP servers declared as an object
directly in `plugin.json`.
Before this change, Codex modeled `mcpServers` as only a string path,
for example:
```json
{
"name": "counter-sample",
"version": "1.1.1",
"mcpServers": "./.mcp.json"
}
```
Some migrated plugins instead provide the server map directly in the
manifest:
```json
{
"name": "counter-sample",
"version": "1.1.1",
"description": "Plugin that declares MCP servers in the manifest",
"mcpServers": {
"counter": {
"type": "http",
"url": "https://sample.example/counter/mcp"
}
}
}
```
That object form previously failed during install/load with an error
like:
```text
failed to parse plugin manifest: invalid type: map, expected a string
```
## What changed
- Add a manifest representation for `mcpServers` as either
`Path(Resource)` or `Object(map)`.
- Parse `plugin.json` `mcpServers` as either a string path or an object.
- Route object-valued MCP server maps through the existing plugin MCP
config parser instead of adding a second parser.
- Apply existing per-plugin MCP server policy to object-valued MCP
servers the same way as file-backed MCP servers.
- Include object-valued MCP server names in plugin telemetry/capability
metadata.
- Support object-valued MCP config for executor plugins without
requiring a `.mcp.json` filesystem read.
- Update the bundled plugin-creator validator and `plugin-json-spec.md`
so generated-plugin validation accepts the same object-valued shape.
## Compatibility
Existing plugin manifests that use `"mcpServers": "./.mcp.json"`
continue to work. Plugins can now also use the object shape shown above.
## Tests
Added coverage for the new manifest attribute shape at the install,
normal load, telemetry, and executor-provider layers:
- `install_accepts_manifest_mcp_server_objects`
- `load_plugins_loads_manifest_mcp_server_objects`
- `plugin_telemetry_metadata_uses_manifest_mcp_server_objects`
- `reads_manifest_object_config_without_executor_file_system_access`
Also smoke-tested the plugin-creator validator against both supported
forms:
- `mcpServers` as a direct object in `plugin.json`
- `mcpServers` as `"./.mcp.json"` with a companion `.mcp.json`
## Validation
- `just test -p codex-plugin`
- `just test -p codex-core-plugins`
- `just test -p codex-mcp-extension`
- `just bazel-lock-update`
- `just bazel-lock-check`
- `just fmt`
- `git diff --check`
- Focused rename/object-form rerun: `just test -p codex-core-plugins
manager::tests::load_plugins_loads_manifest_mcp_server_objects
manager::tests::plugin_telemetry_metadata_uses_manifest_mcp_server_objects
store::tests::install_accepts_manifest_mcp_server_objects`
- Focused executor rerun: `just test -p codex-mcp-extension
executor_plugin::provider::tests::reads_manifest_object_config_without_executor_file_system_access`
- `python3
codex-rs/skills/src/assets/samples/plugin-creator/scripts/validate_plugin.py
/private/tmp/codex-validator-object`
- `python3
codex-rs/skills/src/assets/samples/plugin-creator/scripts/validate_plugin.py
/private/tmp/codex-validator-path`
* code-mode: move cell state into library actor (#28599)
A code-mode cell is a single JavaScript execution that can produce
output, call tools, wait for asynchronous work, resume, or be
terminated. This PR extracts the existing per-cell run loop into a
dedicated actor that owns the cell’s lifecycle state. It is primarily an
ownership change rather than a new lifecycle contract: existing behavior
now has one clear implementation boundary.
### Architecture
The session service remains responsible for session-wide concerns:
allocating cell IDs, storing shared values, creating cells, and routing
requests to them.
Once a cell is created, its execution state belongs to its actor.
Callers interact with the actor through a handle. The actor receives two
kinds of input: runtime events and control requests.
A single event loop serializes these inputs and applies the lifecycle
rules. It tracks the current observer—the caller waiting for an
update—along with accumulated output, outstanding callbacks, runtime
state, yield deadlines, and termination progress. Observation,
termination, completion, and cleanup therefore have one consistent
owner.
When the runtime has no immediately runnable work and is waiting only on
timers or tool results, the actor can return accumulated output and
information about outstanding tool calls while keeping the cell
available to resume. On completion or termination, it performs the
appropriate callback cleanup before publishing the final result and
removing the cell from the session.
A small host interface connects the actor to session-owned facilities
such as tool dispatch, notifications, stored values, and final cell
removal, keeping those responsibilities outside the actor itself.
### Why
Previously, cell lifecycle state and coordination lived alongside
session management. The actor boundary makes each cell a self-contained
state machine with a single writer, while the service becomes a registry
and adapter around it.
This makes lifecycle behavior easier to reason about and test in
isolation. It also establishes a clean boundary for later changing where
cells run or how they communicate without recreating their lifecycle
rules.
* [codex] Test code-mode variable truncation (#28471)
## Summary
Code mode has two separate truncation points: the nested tool result
returned to JavaScript and the code-mode output later recorded for the
model. These tests now verify those behaviors independently.
- Report whether `result.output` was truncated before printing it.
- Verify omitted or sufficiently large nested limits produce `Variable
truncated: False`, while allowing the printed value to be truncated
downstream.
- Verify an explicit nested limit produces `Variable truncated: True`
when the command output exceeds it.
- Use a token-policy model fixture so downstream truncation is visible
as `…N tokens truncated…`.
- Align the explicit nested-truncation expectation with the warning
header.
This PR changes test coverage only; runtime truncation behavior is
unchanged.
## Validation
- `env -u CODEX_SANDBOX_NETWORK_DISABLED RUST_MIN_STACK=8388608 cargo
test -p codex-core --test all code_mode_exec -- --nocapture` (8 passed)
* Revert thread recencyAt for sidebar ordering (#28655)
## Why
Revert #27910 to remove the newly introduced thread `recencyAt`
persistence and API behavior from `main`.
## What changed
This reverts commit `fac3158c2a783095768076489815f361fa9b0db4`,
including the state migration, thread-store propagation, app-server API
surface, generated schemas, and related tests.
## Validation
Not run before opening; relying on CI for the initial fast signal.
* core: remove redundant TurnContext and Prompt fields (#28638)
## Why
`TurnContext` had accumulated dead fields and cached projections of
values already owned by its per-turn `Config` or `ModelInfo`. Keeping
both copies made ownership unclear and allowed artificial split-brain
states, such as a compatibility hash differing from the model metadata
it came from.
`Prompt` similarly carried a write-only personality after personality
selection had already been materialized into its base instructions.
This makes the canonical owner explicit: configuration-backed values
come from `config`, model-derived values come from `model_info`, and
prompts contain only data consumed by request construction.
## What changed
- Remove the unused `ghost_snapshot`, `codex_self_exe`, and
`thread_source` fields.
- Remove duplicate `comp_hash`, `truncation_policy`, `features`,
`shell_environment_policy`, `codex_linux_sandbox_exe`, `compact_prompt`,
and `tool_mode` fields.
- Read those values directly from `TurnContext::config` or
`TurnContext::model_info` at their consumers.
- Remove the write-only `Prompt::personality` field and its constructor
assignments.
- Preserve review-turn inheritance of the parent turn's shell policy,
Linux sandbox executable, and compact prompt through the review config.
## Testing
- `cargo check -p codex-core --tests`
* [codex] Persist built-in image results reported as generating (#28656)
## Why
#27920 stopped persisting image-generation items unless their status was
`completed`, preventing failed standalone extension items with empty
results from being saved. Built-in image generation can instead emit a
terminal `response.output_item.done` containing a complete base64 PNG
while the item status remains `generating`. In that case, app-server
emits no `savedPath`, so Codex Apps can render the inline image but
cannot expose a file artifact.
## What changed
- Persist image-generation items whenever `result` contains image data.
Failed terminal items still have empty results and remain unpersisted.
- Update the existing built-in image-generation integration test to
cover a terminal `generating` item and verify both `saved_path` and the
written PNG bytes.
## Validation
- Confirmed with a raw built-in websocket trace: the image progressed
through `in_progress`, `generating`, and `partial_image`, then emitted
one `response.output_item.done` with `status: "generating"` and a
complete PNG result.
- `just test -p codex-core builtin_image_generation_call_persisted` is
currently blocked before test execution by a pre-existing compile error
in `thread-store/src/thread_metadata_sync.rs:171`.
* Resume exec-server sessions after disconnect (#28512)
Supersedes #28288 (closed).
## Why
A short WebSocket interruption currently ends every client-side process
handle, even though exec-server keeps the server session and its
processes alive for a short time.
This is especially visible for executor-backed stdio MCP servers: a
temporary connection loss becomes a permanent `Transport closed` error.
The server already has the information needed to resume the session, but
the client opens a fresh session instead of using it.
This change reconnects below the process and MCP layers. Existing
process handles stay valid, missed output is recovered, and the same
server-side processes continue running.
## State machine
One logical `ExecServerClient` stays alive while its underlying RPC
connection changes generations.
```text
transport closes
+------------------------------------------------+
| v
+-------------+ +-------------+
| Connected | | Recovering |
+-------------+ +-------------+
^ |
| session resumed, processes caught up | retryable error
+------------------------------------------------+ loops until deadline
|
| deadline or permanent error
v
+-------------+
| Failed |
+-------------+
```
### `Connected`
- New RPC calls use the current connection.
- Process notifications are published in sequence order.
- A disconnect only starts recovery if it came from the current
connection generation. Late events from older generations cannot replace
the active connection.
### `Recovering`
- New calls wait instead of choosing a half-connected RPC client.
- Existing process handles, wake subscriptions, and event subscriptions
stay open.
- Streaming HTTP response bodies fail immediately because their byte
streams cannot be resumed safely.
- Recovery first waits for process starts that were already in flight. A
start whose result became ambiguous is cleaned up after reconnection
instead of being silently adopted.
- The client reconnects with the learned `session_id`. The server may
briefly report that the old connection is still attached, so that error
is retried until the detach finishes.
- The notification consumer starts before the resume handshake
completes. This prevents a busy process from filling the notification
queue and blocking the initialize response.
- Before installing the new connection, the client catches up every
recoverable process with `process/read`.
### `Failed`
- Recovery stops after 25 seconds or after a permanent error.
- Waiting calls are released with one stable disconnect error.
- Existing process sessions receive a terminal failure instead of
waiting forever.
## Recovering process events
Output, exit, and close events share one sequence. During normal
operation, the client buffers early events until every lower sequence
has been published.
After reconnection, the client reads each process starting after its
last published sequence:
1. Retained output chunks are inserted by sequence number.
2. Exit and close state are reconstructed in their sequence positions.
3. Events already received as live notifications are ignored as
duplicates.
4. Newly contiguous events are published in order.
5. If the server no longer retains enough output to fill a sequence gap,
only that process is terminated and failed. The recovered connection
remains usable for other processes.
The server reports its full next event sequence for unbounded reads,
including exit and close events. Closed processes remain readable for
the same 30-second window used to retain detached sessions.
## Other details
- Detached server sessions are retained for 30 seconds, leaving margin
around the client's 25-second recovery deadline.
- Session attach and detach update the active notification sender under
the same attachment lock, so an old connection cannot clear a newly
attached sender.
- A dedicated error code distinguishes the temporary "session is still
attached" race from permanent initialization errors.
- Process starts are identity-checked on both client and server. Cleanup
from an older start cannot remove a newer process that reused the same
ID.
- Mutating requests that were already in flight when the transport
closed are not replayed, because the client cannot know whether the
server applied them. Requests started after recovery is known wait for
the replacement connection.
- We assume the server/client version stays in sync (on the before/after
this PR)
## User impact
Long-running commands and stdio MCP servers can survive a temporary
exec-server WebSocket interruption without changing process IDs or
losing output produced during the outage.
* Back off registry retries during exec recovery (#28546)
## Why
PR #28512 retries a failed session recovery every 100 ms. Every Noise
recovery attempt first asks the environment registry for a fresh
connection bundle, even when the eventual failure comes from the
WebSocket or initialize handshake. During an outage, that could make
each disconnected client call the registry about 250 times during the
25-second recovery window.
## What changes
All retryable Noise recovery failures now use a separate backoff
schedule:
```text
base: 500 ms -> 1 s -> 2 s -> 4 s -> 5 s maximum
actual: 500-750 ms, 1-1.5 s, 2-3 s, 4-6 s, 5-7.5 s
```
The extra 0-50% is deterministic per-session jitter so disconnected
clients do not retry together. Direct WebSocket recovery keeps the
existing 100 ms retry because it does not re-enter the registry.
* Add join key for MAv2 inter-agent messages (#28561)
## Summary
This keeps inter-agent communication on the existing raw response item
path and adds a join key for MAv2 tool calls.
MAv2 `spawn_agent`, `send_message`, and `followup_task` now stamp the
originating tool call id into `ResponseItemMetadata.source_call_id` on
the raw `ResponseItem::AgentMessage`. App-server clients can join that
raw item back to the existing tool/activity event by call id, while
using the raw agent message's existing sender, receiver, and content
fields.
No new app-server `ThreadItem` or notification type is added.
## Tests
- `just fmt`
- `just write-app-server-schema`
- `just test -p codex-protocol`
- `just test -p codex-app-server-protocol`
- `just test -p codex-core
multi_agent_v2_spawn_returns_path_and_send_message_accepts_relative_path`
- `just test -p codex-core
multi_agent_v2_followup_task_completion_notifies_parent_on_every_turn`
- `just fix -p codex-protocol`
- `just fix -p codex-app-server-protocol`
- `just fix -p codex-core`
* app-server: keep the model cache warm (#28699)
## Why
The app server is long-lived, but its shared model cache otherwise
refreshes only when a caller needs it. Once the five-minute cache
expires, starting a thread or calling `model/list` can wait for
`/models` on the request path.
Refresh the cache in the background before it expires so foreground
callers normally use fresh local state.
## What changed
- Start an app-server worker that refreshes models immediately and then
every three minutes using the existing models-manager API.
- Hold only a weak reference to the models manager between refreshes, so
the worker does not extend its lifetime.
- Stop scheduling refreshes when the app-server lifecycle handle is shut
down or dropped. A refresh already in progress is allowed to finish.
- Adjust affected app-server test fixtures to distinguish the background
`/models` probe from the connection they are testing.
The existing models-manager cache, refresh strategies, auth handling,
ETag behavior, and concurrency semantics are unchanged.
## Testing
-
`models_refresh_worker::tests::refreshes_immediately_periodically_and_stops_when_dropped`
-
`suite::v2::remote_control::listen_off_honors_persisted_remote_control_enable`
-
`suite::v2::attestation::attestation_generate_round_trip_adds_header_to_responses_websocket_handshake`
* Replace SkillsManager with SkillsService (#28705)
## Why
Host skill discovery was still exposed as a manager even though it is a
process-owned service shared by sessions, the app-server catalog, and
file-watcher invalidation. The skills extension also consumed an ad hoc
loaded-skills wrapper instead of a named immutable snapshot.
## What changed
- replace `SkillsManager` with concrete `SkillsService`
- make the service cache and return immutable `HostSkillsSnapshot`
values
- migrate the skills extension host provider to the snapshot boundary
- migrate app-server catalog, watcher, and invalidation paths to the
service
This keeps the service limited to host discovery, caching, roots, and
invalidation. Catalog rendering and invocation remain extension
responsibilities for the next stacked change.
* [ez][codex-rs] Support apps._default.default_tools_approval_mode (#27965)
[from codex]
## Summary
- add `default_tools_approval_mode` to `[apps._default]` and expose it
through app-server v2 `config/read`
- apply it after managed, per-tool, and per-app approval settings,
before the built-in `auto` fallback
- document the precedence, regenerate config/app-server schemas, and add
unit plus end-to-end approval coverage
## Configuration
```toml
[apps._default]
default_tools_approval_mode = "prompt"
```
The effective precedence is managed requirements, tool-specific
`approval_mode`, app-specific `default_tools_approval_mode`,
`apps._default.default_tools_approval_mode`, then `auto`.
## Test plan
- `just write-config-schema`
- `just write-app-server-schema`
- `just write-app-server-schema --experimental`
- `just test -p codex-core app_tool_policy`
- `just test -p codex-core mcp_turn_metadata`
- `just test -p codex-config`
- `just test -p codex-app-server-protocol`
- `just test -p codex-app-server config_read_includes_apps`
- `just fix -p codex-config -p codex-core -p codex-app-server-protocol
-p codex-app-server`
- `just fmt`
* Run fs helper through Windows sandbox wrapper (#28359)
## Why
This is the final PR in the Windows fs-helper sandbox stack and contains
the actual bug fix.
The exec-server filesystem helper is a direct-spawn path: it asks
`SandboxManager` for a `SandboxExecRequest`, then launches the returned
argv itself. That works on macOS and Linux because the transformed argv
is already a self-contained sandbox wrapper. On Windows, the transformed
request carried `WindowsRestrictedToken` metadata, but the direct-spawn
fs-helper runner still launched the helper argv directly.
That means Windows filesystem built-ins backed by the fs-helper could
run with the parent Codex process permissions instead of the configured
Windows sandbox. This PR makes the direct-spawn transform produce a
self-contained Windows wrapper argv before fs-helper launches it.
## What Changed
- Added `SandboxManager::transform_for_direct_spawn()` for callers that
launch the returned argv themselves.
- Wrapped Windows restricted-token direct-spawn requests with `codex.exe
--run-as-windows-sandbox` and then marked the outer request as
unsandboxed, matching the macOS/Linux wrapper argv shape.
- Updated `exec-server/src/fs_sandbox.rs` to use the direct-spawn
transform for fs-helper launches.
- Materialized the inner `codex.exe --codex-run-as-fs-helper` executable
into `.sandbox-bin` so the sandboxed user can run it.
…
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
List tools from MCP servers concurrently while preserving configured server order, and skip optional servers that are still starting when they have no cached tool snapshot.
Required servers retain the existing blocking contract. Optional pending servers with a snapshot continue to expose cached tools, and completed servers continue to use live tools with the existing cache fallback on failure.
Why
The first turn currently waits for MCP servers serially, so independent waits accumulate and one slow optional server can prevent the first model request from being sent. Concurrent fan-in removes the serial sum, while the required/optional policy keeps unavailable optional servers off the first-turn critical path.
Validation
just test -p codex-mcp list_all_tools— 10 tests passed, including concurrent polling and the required/optional/cache behavior matrix.--otel-traces-endpointprofiling prerequisite is not present here; the draft remains pending release-trace confirmation.Addresses #19556 and #21318.