## Summary
Removes Plugin descriptions from the dev message, since descriptions of
skills and MCPs cover the capabilities offered by the plugin.
## Testing
- [x] Updates unit tests
## Summary
Clear inherited legacy `notify` from Guardian review session config,
since we should not be passing auto review threads into `notify`
targets. Keeps legacy notify payload and hook runtime behavior unchanged
for normal user turns.
## Testing
- [x] add a Guardian config regression and dedicated Guardian
integration test so review sessions cannot inherit parent notify hooks
## Summary
- add the missing response.created event to the mocked empty follow-up
response in the compact rollback test
- keep the fix scoped to the flaky mocked stream shape, without
increasing timeouts
## Recent flakes on main
- `snapshot_rollback_followup_turn_trims_context_updates` failed in
`rust-ci-full` on `main` in the Ubuntu remote test job on 2026-05-14:
https://github.com/openai/codex/actions/runs/25891434395/job/76095284830
- The same `compact_resume_fork` suite also failed recently on `main`
with `snapshot_rollback_past_compaction_replays_append_only_history`,
which has the same mocked Responses stream shape sensitivity this PR is
tightening:
https://github.com/openai/codex/actions/runs/25892437363/job/76098329098
## Verification
- env -u CODEX_SANDBOX_NETWORK_DISABLED cargo test -p codex-core --test
all snapshot_rollback_followup_turn_trims_context_updates -- --nocapture
- repeated the same focused test 3 consecutive times locally
- UV_CACHE_DIR=/private/tmp/uv-cache-codex-fmt just fmt
## Summary
Remove the deprecated `experimental_instructions_file` config setting
from the typed config surface and the remaining deprecation-notice
plumbing. `model_instructions_file` remains the supported setting and
its loading path is unchanged.
The setting was deprecated when it was renamed to
`model_instructions_file` on January 20, 2026 in
https://github.com/openai/codex/pull/9555.
## Changes
- Remove `experimental_instructions_file` from `ConfigToml` and
`ConfigProfile`.
- Delete the custom config-layer scan and session deprecation notice for
the removed setting.
- Stop clearing the removed field from generated session config locks.
- Remove the obsolete deprecation-notice test case while keeping
`model_instructions_file` coverage intact.
## Validation
- `just write-config-schema`
- `just fmt`
- `cargo test -p codex-config`
- `cargo test -p codex-core model_instructions_file`
- `just fix -p codex-core`
- `git diff --check`
Co-authored-by: Codex <noreply@openai.com>
## Summary
- move removed feature enum variants under the existing Removed section
- keep active feature variants grouped away from no-op compatibility
flags
## Test plan
- just fmt
- cargo test -p codex-features
Co-authored-by: Codex <noreply@openai.com>
## Summary
Removes the feature since this is effectively on by default in all cases
where we should use it, or can be configured via models.json.
## Testing
- [x] unit tests pass
## Summary
Get rid of the `experimental_use_freeform_apply_patch` config option,
since it is now encoded in model config. No deprecation message since it
has been experimental this entire time.
## Testing
- [x] Updated unit tests
---------
Co-authored-by: Codex <noreply@openai.com>
## Summary
It appears this config flag has been broken/a noop for quite some time:
since https://github.com/openai/codex/pull/8850. Let's simplify and get
rid of this.
## Testing
- [x] Updated unit tests
## Summary
Adds include_collaboration_mode_instructions, which is a config
equivalent to include_permissions_instructions for collaboration modes.
Desired for situations where we want to disable this instruction from
entering the context
## Testing
- [x] Added unit test
## Summary
Restricts behavior of `is_known_safe_command` only to modes where it is
explicitly part of the documented behavior:
- when `environment_lacks_sandbox_protections`
- in `AskForApproval::UnlessTrusted`
Notably, as a result of this, escalations for commands that pass
`is_known_safe_commands` are no longer auto-approved in
AskForApproval::OnRequest or AskForApproval::Granular.
## Testing
- [x] Updated unit tests
- [x] Updated approvals scenario tests.
---------
Co-authored-by: Codex <noreply@openai.com>
## Summary
Fixes a regression introduced in #10941 so that heredocs do not permit
file redirects to be approved by rules, and adds scenario tests to cover
this behavior.
Previously, heredoc command parsing would allow redirects and
environment variables:
```bash
# commands_for_exec_policy() would parse this via parse_shell_lc_single_command_prefix
PATH=/tmp/bad:$PATH cat <<'EOF' > /tmp/bad/hello.txt
hello
EOF
```
This conflicts with the Codex Rules documentation; heredoc parsing logic
should abide by the same strictness of parsing.
## Tests
- [x] Updated unit tests accordingly
- [x] Added scenario tests for these cases
---------
Co-authored-by: Codex <noreply@openai.com>
## Summary
Starts the process of getting rid of `--full-auto`, with some
concessions:
1. Fully removes the command from the tui, since it just resolves to the
default permissions there, and encourages users to use the one-time
trust flow if they're not in a trusted repo.
2. Marks the command as deprecated in `codex exec`, in case users are
actively relying on this. We'll remove in an upcoming n+X release.
3. Cleans up some of the `codex sandbox` cli logic, to keep supporting
legacy sandbox policies for now.
This isn't the cleanest setup, but I think it is worthwhile to warn
users for one release before hard-removing it.
## Testing
- [x] Updated unit tests
## Why
The log DB writer batches tracing events before inserting them into
SQLite, but `tokio::time::interval` produces an immediate first tick.
That meant the inserter could flush the first accepted log entry before
`batch_size` was reached, making
`configured_batch_size_flushes_without_explicit_flush` timing-sensitive
in CI.
## What Changed
- Consume the interval's startup tick before entering the inserter loop,
so interval flushing starts after the configured delay.
- Remove the test's startup sleep, which was masking the race instead of
proving the batch-size behavior.
## Validation
- `cargo test -p codex-state`
- `cargo test -p codex-state
configured_batch_size_flushes_without_explicit_flush` passed 3
consecutive focused runs
- PR checks passed across `rust-ci`, Bazel, `ci`, `sdk`, `cargo-deny`,
Codespell, blob-size policy, and CLA
## Why
Recent `main` CI had repeated flakes in the plugin fixture tests:
- `codex-core::all
suite::plugins::explicit_plugin_mentions_inject_plugin_guidance` failed
in runs
[24909500958](https://github.com/openai/codex/actions/runs/24909500958),
[24908076251](https://github.com/openai/codex/actions/runs/24908076251),
[24906197645](https://github.com/openai/codex/actions/runs/24906197645),
and
[24898949647](https://github.com/openai/codex/actions/runs/24898949647).
- `codex-core::all suite::plugins::plugin_mcp_tools_are_listed` failed
in runs
[24909500958](https://github.com/openai/codex/actions/runs/24909500958),
[24908076251](https://github.com/openai/codex/actions/runs/24908076251),
and
[24898949647](https://github.com/openai/codex/actions/runs/24898949647).
The failures were in the same plugin/MCP fixture family: assertions
expected sample plugin guidance or tool inventory, but the test could
observe the session before the sample MCP server had finished startup.
## Root Cause
`explicit_plugin_mentions_inject_plugin_guidance` submitted the user
turn immediately after constructing the session. MCP startup is
asynchronous, so on a slower or busier CI runner the prompt could be
built before the sample plugin MCP server had reported its tools. That
made the test depend on scheduler timing rather than the fixture being
ready.
`plugin_mcp_tools_are_listed` already needed the same readiness
condition, but its wait logic was local to that test.
## What Changed
- Added a shared `wait_for_sample_mcp_ready` helper for the plugin
fixture tests.
- Wait for `McpStartupComplete` before submitting the explicit plugin
mention turn.
- Reuse the same readiness helper in the MCP tool-listing test.
## Why This Should Be Reliable
The tests now wait for the explicit readiness signal from the sample MCP
server before asserting guidance or tools derived from that server. This
removes the startup race while still exercising the real fixture path,
so the assertions should only run after the plugin inventory is
deterministic.
## Verification
- `cargo test -p codex-core --test all plugins::`
- GitHub CI for this PR is passing.
## Why
Recent `main` CI repeatedly timed out in:
- `codex-core::all suite::approvals::approval_matrix_covers_all_modes`
It failed in runs
[24909500958](https://github.com/openai/codex/actions/runs/24909500958),
[24908076251](https://github.com/openai/codex/actions/runs/24908076251),
[24906197645](https://github.com/openai/codex/actions/runs/24906197645),
[24905823212](https://github.com/openai/codex/actions/runs/24905823212),
[24903439629](https://github.com/openai/codex/actions/runs/24903439629),
[24903336028](https://github.com/openai/codex/actions/runs/24903336028),
and
[24898949647](https://github.com/openai/codex/actions/runs/24898949647).
The failure pattern was a 60s Linux remote timeout. Logs showed many
approval scenarios completing before the single matrix test timed out.
## Root Cause
`approval_matrix_covers_all_modes` packed every approval/sandbox/tool
scenario into one test case. That made the test vulnerable to normal CI
variance: one slow scenario or a slow process startup could push the
whole monolithic case past the 60s per-test timeout. It also hid which
part of the matrix was slow because the runner only reported the one
large matrix test.
## What Changed
- Keep the shared `scenarios()` table as the single source of approval
matrix coverage.
- Use one `#[test_case]` per `ScenarioGroup` to generate five async
Tokio tests: danger/full-access, read-only, workspace-write,
apply-patch, and unified-exec.
- Keep the group runner small and add per-scenario error context so a
failure still reports the specific scenario name.
## Why This Should Be Reliable
Each scenario group now has its own test harness timeout instead of
sharing one timeout window with the full matrix. That removes the long
sequential loop from a single test while keeping the implementation
compact and easy to scan.
The tests still run through the same scenario definitions and runner, so
this preserves coverage. `test-case` already composes with
`#[tokio::test]` in this crate and is already available for test code.
## Verification
- `cargo test -p codex-core --test all approval_matrix_ -- --list`
- `cargo test -p codex-core --test all approval_matrix_`
## Summary
Allow the user to approve a request_permissions_tool request with the
condition that all commands in the rest of the turn are reviewed by
guardian, regardless of sandbox status.
## Testing
- [x] Added unit tests
- [x] Ran locally
## Summary
Short circuit the convo if auto-review hits too many denials
## Testing
- [x] Added unit tests
---------
Co-authored-by: Codex <noreply@openai.com>
## Summary
- Keep the guardian policy installed as guardian base instructions.
- Clear inherited parent `developer_instructions` for guardian review
sessions.
- Update guardian config tests to assert developer instructions are
cleared and policy text is sourced from base instructions.
## Why
Guardian review sessions are intended to run under an isolated guardian
policy. Because the guardian config is cloned from the parent config,
inherited custom or managed developer instructions could otherwise
remain active and conflict with guardian review behavior.
## Validation
- `just fmt`
- `cargo test -p codex-core guardian_review_session_config`
Co-authored-by: Codex <noreply@openai.com>
## Summary
When auto-review is enabled, it should handle request_permissions tool.
We'll need to clean up the UX but I'm planning to do that in a separate
pass
## Testing
- [x] Ran locally
<img width="893" height="396" alt="Screenshot 2026-04-17 at 1 16 13 PM"
src="https://github.com/user-attachments/assets/4c045c5f-1138-4c6c-ac6e-2cb6be4514d8"
/>
---------
Co-authored-by: Codex <noreply@openai.com>
## Summary
Remove the skills message from the guardian dev message
## Test Plan
- [x] Ran locally
- [x] Added unit test
---------
Co-authored-by: Codex <noreply@openai.com>
## Summary
Stop counting elicitation time towards mcp tool call time. There are
some tradeoffs here, but in general I don't think time spent waiting for
elicitations should count towards tool call time, or at least not
directly towards timeouts.
Elicitations are not exactly like exec_command escalation requests, but
I would argue it's ~roughly equivalent.
## Testing
- [x] Added unit tests
- [x] Tested locally
## Summary
https://github.com/openai/codex/pull/13860 changed the serialized output
format of Unified Exec. This PR reverts those changes and some related
test changes
## Testing
- [x] Update tests
---------
Co-authored-by: Codex <noreply@openai.com>
## Summary
In #11871 we started consolidating on ExecRequest.sandbox_policy instead
of passing in a separate policy object that theoretically could differ
(but did not). This finishes the some parameter cleanup.
This should be a simple noop, since all 3 callsites of this function
already used a cloned object from the ExecRequest value.
## Testing
- [x] Existing tests pass
## Summary
Adds support for approvals_reviewer to `Op::UserTurn` so we can migrate
`[CodexMessageProcessor::turn_start]` to use Op::UserTurn
## Testing
- [x] Adds quick test for the new field
Co-authored-by: Codex <noreply@openai.com>
## Summary
If we are in a mode that is already explicitly un-sandboxed, then
`ApprovalPolicy::Never` should not block dangerous commands.
## Testing
- [x] Existing unit test covers old behavior
- [x] Added a unit test for this new case
## Summary
If a subagent requests approval, and the user persists that approval to
the execpolicy, it should (by default) propagate. We'll need to rethink
this a bit in light of coming Permissions changes, though I think this
is closer to the end state that we'd want, which is that execpolicy
changes to one permissions profile should be synced across threads.
## Testing
- [x] Added integration test
---------
Co-authored-by: Codex <noreply@openai.com>
## Summary
We need to support allowing request_permissions calls when using
`Reject` policy
<img width="1133" height="588" alt="Screenshot 2026-03-09 at 12 06
40 PM"
src="https://github.com/user-attachments/assets/a8df987f-c225-4866-b8ab-5590960daec5"
/>
Note that this is a backwards-incompatible change for Reject policy. I'm
not sure if we need to add a default based on our current use/setup
## Testing
- [x] Added tests
- [x] Tested locally
## Summary
The apply_patch tool should also respect AdditionalPermissions
## Testing
- [x] Added unit tests
---------
Co-authored-by: Codex <noreply@openai.com>
## Summary
request_permissions flows should support persisting results for the
session.
Open Question: Still deciding if we need within-turn approvals - this
adds complexity but I could see it being useful
## Testing
- [x] Updated unit tests
---------
Co-authored-by: Codex <noreply@openai.com>
## Summary
Alternative to #14061 - we need to use a child process on windows to
correctly validate Powershell behavior.
## Testing
- [x] These are tests
## Summary
This test was missing the turn completion event in the responses stream,
so it was hanging. This PR fixes the issue
## Testing
- [x] This does update the test
## Summary
#13910 was merged with some unused imports, let's fix this
## Testing
- [x] Let's make sure CI is green
---------
Co-authored-by: Charles Cunningham <ccunningham@openai.com>
Co-authored-by: Codex <noreply@openai.com>
## Summary
This test is good, but flakey and we have to figure out some bazel build
issues. Let's get CI back go green and then land a stable version!
## Test Summary
- [x] CI Passes
## Summary
Simplify the trusted directory flow. This logic was originally designed
several months ago, to determine if codex should start in read-only or
workspace-write mode. However, that's no longer the purpose of directory
trust - and therefore we should get rid of this logic.
## Testing
- [x] Unit tests pass
## Summary
Codex discovered this one - shell_snapshot tests were breaking on my
machine because I had a multiline env var. We should handle these!
## Testing
- [x] existing tests pass
- [x] Updated unit tests
## Summary
Introduces the initial implementation of Feature::RequestPermissions.
RequestPermissions allows the model to request that a command be run
inside the sandbox, with additional permissions, like writing to a
specific folder. Eventually this will include other rules as well, and
the ability to persist these permissions, but this PR is already quite
large - let's get the core flow working and go from there!
<img width="1279" height="541" alt="Screenshot 2026-02-15 at 2 26 22 PM"
src="https://github.com/user-attachments/assets/0ee3ec0f-02ec-4509-91a2-809ac80be368"
/>
## Testing
- [x] Added tests
- [x] Tested locally
- [x] Feature
## Summary
The test exec_resume_last_respects_cwd_filter_and_all_flag makes one
session “newest” by resuming it, but rollout updated_at is stored/sorted
at second precision. On fast CI (especially Windows), the touch could
land in the same second as initial session creation, making ordering
nondeterministic.
This change adds a short sleep before the recency-touch step so the
resumed session is guaranteed to have a later updated_at, preserving the
intended assertion without changing product behavior.
## Summary
`gpt-5.3-codex` really likes to write complicated shell scripts, and
suggest a partial prefix_rule that wouldn't actually approve the
command. We should only show the `prefix_rule` suggestion from the model
if it would actually fully approve the command the user is seeing.
This will technically cause more instances of overly-specific
suggestions when we fallback, but I think the UX is clearer,
particularly when the model doesn't necessarily understand the current
limitations of execpolicy parsing.
## Testing
- [x] Add unit tests
- [x] Add integration tests
## Summary
We're seeing failures for shell-tool-mcp.yml during git checkouts. This
is a quick attempt to unblock releases - we should revisit this build
pipeline since we've hit a number of errors.
## Summary
Fixes a few things in our exec_policy handling of prefix_rules:
1. Correctly match redirects specifically for exec_policy parsing. i.e.
if you have `prefix_rule(["echo"], decision="allow")` then `echo hello >
output.txt` should match - this should fix#10321
2. If there already exists any rule that would match our prefix rule
(not just a prompt), then drop it, since it won't do anything.
## Testing
- [x] Updated unit tests, added approvals ScenarioSpecs
## Summary
This feature is now reasonably stable, let's remove it so we can
simplify our upcoming iterations here.
## Testing
- [x] Existing tests pass
## Summary
Based on our most recent [release
attempt](https://github.com/openai/codex/actions/runs/21980518940/job/63501739210)
we are not building the shell-tool-mcp job correctly. This one is
outside my expertise, but seems mostly reasonable.
## Testing
- [x] We really need dry runs of these
## Summary
CI is broken on main because our CI toolchain is trying to run 1.93.1
while our rust toolchain is locked at 1.93.0. I'm sure it's likely safe
to upgrade, but let's keep things stable for now.
## Testing
- [x] CI should hopefully pass
## Summary
If the model suggests a bad rule, don't show it to the user. This does
not impact the parsing of existing rules, just the ones we show.
## Testing
- [x] Added unit tests
- [x] Ran locally
## Summary
In an effort to start simplifying our sandbox setup, we're announcing
this approval_policy as deprecated. In general, it performs worse than
`on-request`, and we're focusing on making fewer sandbox configurations
perform much better.
## Testing
- [x] Tested locally
- [x] Existing tests pass
## Summary
Preserve the specified model slug when we get a prefix-based match
## Testing
- [x] added unit test
---------
Co-authored-by: Ahmed Ibrahim <aibrahim@openai.com>
## Summary
Consolidate `/status` Permissions lines into a simpler view. It should
only show "Default," "Full Access," or "Custom" (with specifics)
## Testing
- [x] many snapshots updated
## Summary
This should rarely, if ever, happen in practice. But regardless, we
should never provide an empty list of `commands` to ExecPolicy. This PR
is almost entirely adding test around these cases.
## Testing
- [x] Adds a bunch of unit tests for this
## Summary
- keep wiremock MockServer handles alive through async assertions in
remote model suite tests
- assert /models request count in remote_models_hide_picker_only_models
- use a slightly higher parallel timing threshold on aarch64 while
keeping existing x86 threshold
## Validation
- just fmt
- targeted tests:
- cargo test -p codex-core --test all
suite::remote_models::remote_models_merge_replaces_overlapping_model --
--exact
- cargo test -p codex-core --test all
suite::remote_models::remote_models_hide_picker_only_models -- --exact
- cargo test -p codex-core --test all
suite::tool_parallelism::shell_tools_run_in_parallel -- --exact
- soak loop: 40 iterations of all three targeted tests
## Notes
- cargo test -p codex-core has one unrelated local-env failure in
shell_snapshot::tests::try_new_creates_and_deletes_snapshot_file from
exported certificate env content in this workspace.
- local bazel test //codex-rs/core:core-all-test failed to build due
missing rust-objcopy in this host toolchain.
Summary
- add platform-aware defaults for shell command timeouts so Windows
tests get longer waits
- keep medium timeout longer on Windows to ensure flakiness is reduced
Testing
- Not run (not requested)
## Summary
When resuming with a different model, we should also append a developer
message with the model instructions
## Testing
- [x] Added unit tests
## Summary
When switching models, we should append the instructions of the new
model to the conversation as a developer message.
## Test
- [x] Adds a unit test
## Summary
We ideally shouldn't make it to this point in the first place, but if we
do try to append a rule that already exists, we shouldn't append the
same rule twice.
## Testing
- [x] Added unit test for this case
## Summary
We should probably warn users that they have a million rules, and help
clean them up. But for now, we should handle this unbounded case.
Limit rules listed in conversations, with shortest / broadest rules
first.
## Testing
- [x] Updated unit tests
Dylan Hurd