3972 Commits

This Branch

This Branch

All Branches

• Send warmup request (#11258)

  Send a request with `generate: falls` but a full set of tools and
  instructions to pre-warm inference.
  
  ---------
  
  Co-authored-by: Codex <noreply@openai.com>

  pakrym-oai · 2026-02-24 08:15:47 -08:00
  97d0068658
• fix: replay after /agent (#12663)

  Filter the events after a`/agent` replay to prevent replaying decision
  events

  jif-oai · 2026-02-24 12:08:38 +00:00
  0679e70bfc
• memories: tighten memory lookup guidance and citation requirements (#12635)

  ## Summary
  - tighten the memory-use decision boundary so agents skip memory only
  for clearly self-contained asks
  - make the quick memory pass more explicit and bounded (including a
  lightweight search budget)
  - add structured `<memory_citation>` requirements and examples for final
  replies
  - clarify memory update guidance and end-state wording for memory lookup
  
  ## Why
  The previous template was directionally correct, but still left room for
  inconsistent memory lookup behavior and citation formatting. This change
  makes the default behavior, quick-pass scope, and citation output
  contract much more explicit.
  
  ## Testing
  - not run (prompt/template text change only)
  
  Co-authored-by: jif-oai <jif@openai.com>

  zuxin-oai · 2026-02-24 11:46:28 +00:00
  3fe365ad8a
• feat: mutli agents persist config overrides (#12667)

  Fix propagation of runtime config changes and `--yolo`

  jif-oai · 2026-02-24 11:33:00 +00:00
  8758db5d5b
• memories: tighten consolidation prompt schema and indexing guidance (#12653)

  ## Summary
  - tighten the Phase 2 consolidation prompt for task-oriented `MEMORY.md`
  generation
  - address Phase 2 under-coverage / "laziness" with stronger workflow +
  final-pass checks
  - improve recency/ordering behavior for `MEMORY.md` and
  `memory_summary.md`
  - rewrite `## What's in Memory` as a clearer routing index with explicit
  recent-3-day structure
  
  ## Key Changes
  - `MEMORY.md` schema cleanup:
  - align on `## Task <n>` task sections (remove stale `task:`
  rule/example references)
    - include `thread_id` in rollout provenance examples
    - compact comma-separated `### keywords` format
  - Phase 2 completeness guardrails:
    - chunked INIT coverage pass over `raw_memories.md`
    - incremental net-new indexing / routing steps
  - stronger final checks (day ordering, topic coverage, keyword
  searchability, accidental duplication)
  - Recency / ordering rules:
  - clearer scan-order guidance for raw memories (newest-first bias in
  incremental mode)
  - utility+recency ordering guidance for `MEMORY.md` task groups and
  summary topics
    - rebuild recent active window from current `updated_at` coverage
  - `## What's in Memory` rewrite:
    - index/routing-layer framing (not a mini-handbook)
    - explicit recent 3 distinct memory-day layout
    - richer recent-topic entries + compact lower-priority routing entries
  - clearer `desc` / `learnings` expectations and separation from `##
  General Tips`
  - Explicitly allow rollout-summary reuse across multiple tasks/blocks
  when it supports distinct task angles (with distinct task-local value)
  
  ## Notes
  - Prompt-template only:
  `codex-rs/core/templates/memories/consolidation.md`
  - No runtime/code changes
  
  ## Validation
  - Manual diff review only

  zuxin-oai · 2026-02-24 09:41:20 +00:00
  15f6cfb047
• Simplify skill tracking (#12652)

  Remove a few layers of structs and store SkillMetadata.
  
  ---------
  
  Co-authored-by: alexsong-oai <alexsong@openai.com>

  pakrym-oai · 2026-02-23 22:47:39 -08:00
  68a7d98363
• chore: rm hardcoded PRESETS list (#12650)

  rm `PRESETS` list harcoded in `model_presets` as we now have bundled
  `models.json` with equivalent info.
  
  update logic to rely on bundled models instead, update tests.

  sayan-oai · 2026-02-23 22:35:51 -08:00
  7e46e5b9c2
• Add skill approval event/response (#12633)

  Set the stage for skill-level permission approval in addition to
  command-level.
  
  Behind a feature flag.

  pakrym-oai · 2026-02-23 22:28:58 -08:00
  58763afa0f
• Avoid AbsolutePathBuf::parent() panic under EMFILE by skipping re-absolutization (#12647)

  Fixes #12216
  
  Fixes a panic in `AbsolutePathBuf::parent()` when the process hits file
  descriptor exhaustion (`EMFILE` / "Too many open files").
  
  ### Root cause
  
  `AbsolutePathBuf::parent()` was re-validating the parent path via
  `from_absolute_path(...).expect(...)`.
  
  `from_absolute_path()` calls `path_absolutize::absolutize()`, which can
  depend on `std::env::current_dir()`. Under `EMFILE`, that can fail,
  causing `parent()` to panic even though the parent of an absolute path
  is already known.
  
  ### Change
  
  - Stop re-absolutizing the result of `self.0.parent()`
  - Construct `AbsolutePathBuf` directly from the known parent path
  - Keep an invariant check with `debug_assert!(p.is_absolute())`
  
  ### Why this is safe
  
  `self` is already an `AbsolutePathBuf`, so `self.0` is
  absolute/normalized. The parent of an absolute path is expected to be
  absolute, so re-running fallible normalization here is unnecessary and
  can introduce unrelated panics.

  Eric Traut · 2026-02-23 21:59:33 -08:00
  a4076ab4b1
• Support implicit skill invocation analytics events (#12049)

  - use `skills_for_cwd` lookup to scope allowed skills and build
  invocation context for downstream processing
  - add detection in `stream_events_utils` to classify tool calls as
  implicit skill invocations per the proposal (script runners, extensions,
  `scripts` dirs, and SKILL.md reads)
  - deduplicate invocations per turn and emit analytics/OTEL events on the
  same background queue as explicit invokes

  alexsong-oai · 2026-02-23 21:55:49 -08:00
  09a82f364f
• fix(exec) Patch resume test race condition (#12648)

  ## 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.

  Dylan Hurd · 2026-02-23 21:54:25 -08:00
  fbeda61cc3
• feat(core): persist network approvals in execpolicy (#12357)

  ## Summary
  Persist network approval allow/deny decisions as `network_rule(...)`
  entries in execpolicy (not proxy config)
  
  It adds `network_rule` parsing + append support in `codex-execpolicy`,
  including `decision="prompt"` (parse-only; not compiled into proxy
  allow/deny lists)
  - compile execpolicy network rules into proxy allow/deny lists and
  update the live proxy state on approval
  - preserve requirements execpolicy `network_rule(...)` entries when
  merging with file-based execpolicy
  - reject broad wildcard hosts (for example `*`) for persisted
  `network_rule(...)`

  viyatb-oai · 2026-02-23 21:37:46 -08:00
  c3048ff90a
• refactor: decouple shell-escalation from codex-core (#12638)

  ## Why
  
  After removing `exec-server`, the next step is to wire a new shell tool
  to `codex-rs/shell-escalation` directly.
  
  That is blocked while `codex-shell-escalation` depends on `codex-core`,
  because the new integration would require `codex-core` to depend on
  `codex-shell-escalation` and create a dependency cycle.
  
  This change ports the reusable pieces from the earlier prep work, but
  drops the old compatibility shim because `exec-server`/MCP support is
  already gone.
  
  ## What Changed
  
  ### Decouple `shell-escalation` from `codex-core`
  
  - Introduce a crate-local `SandboxState` in `shell-escalation`
  - Introduce a `ShellCommandExecutor` trait so callers provide process
  execution/sandbox integration
  - Update `EscalateServer::exec(...)` and `run_escalate_server(...)` to
  use the injected executor
  - Remove the direct `codex_core::exec::process_exec_tool_call(...)` call
  from `shell-escalation`
  - Remove the `codex-core` dependency from `codex-shell-escalation`
  
  ### Restore reusable policy adapter exports
  
  - Re-enable `unix::core_shell_escalation`
  - Export `ShellActionProvider` and `ShellPolicyFactory` from
  `shell-escalation`
  - Keep the crate root API simple (no `legacy_api` compatibility layer)
  
  ### Port socket fixes from the earlier prep commit
  
  - Use `socket2::Socket::pair_raw(...)` for AF_UNIX socketpairs and
  restore `CLOEXEC` explicitly on both endpoints
  - Keep `CLOEXEC` cleared only on the single datagram client FD that is
  intentionally passed across `exec`
  - Clean up `tokio::AsyncFd::try_io(...)` error handling in the socket
  helpers
  
  ## Verification
  
  - `cargo shear`
  - `cargo clippy -p codex-shell-escalation --tests`
  - `cargo test -p codex-shell-escalation`

  Michael Bolin · 2026-02-23 20:58:24 -08:00
  af215eb390
• refactor: delete exec-server and move execve wrapper into shell-escalation (#12632)

  ## Why
  
  We already plan to remove the shell-tool MCP path, and doing that
  cleanup first makes the follow-on `shell-escalation` work much simpler.
  
  This change removes the last remaining reason to keep
  `codex-rs/exec-server` around by moving the `codex-execve-wrapper`
  binary and shared shell test fixtures to the crates/tests that now own
  that functionality.
  
  ## What Changed
  
  ### Delete `codex-rs/exec-server`
  
  - Remove the `exec-server` crate, including the MCP server binary,
  MCP-specific modules, and its test support/test suite
  - Remove `exec-server` from the `codex-rs` workspace and update
  `Cargo.lock`
  
  ### Move `codex-execve-wrapper` into `codex-rs/shell-escalation`
  
  - Move the wrapper implementation into `shell-escalation`
  (`src/unix/execve_wrapper.rs`)
  - Add the `codex-execve-wrapper` binary entrypoint under
  `shell-escalation/src/bin/`
  - Update `shell-escalation` exports/module layout so the wrapper
  entrypoint is hosted there
  - Move the wrapper README content from `exec-server` to
  `shell-escalation/README.md`
  
  ### Move shared shell test fixtures to `app-server`
  
  - Move the DotSlash `bash`/`zsh` test fixtures from
  `exec-server/tests/suite/` to `app-server/tests/suite/`
  - Update `app-server` zsh-fork tests to reference the new fixture paths
  
  ### Keep `shell-tool-mcp` as a shell-assets package
  
  - Update `.github/workflows/shell-tool-mcp.yml` packaging so the npm
  artifact contains only patched Bash/Zsh payloads (no Rust binaries)
  - Update `shell-tool-mcp/package.json`, `shell-tool-mcp/src/index.ts`,
  and docs to reflect the shell-assets-only package shape
  - `shell-tool-mcp-ci.yml` does not need changes because it is already
  JS-only
  
  ## Verification
  
  - `cargo shear`
  - `cargo clippy -p codex-shell-escalation --tests`
  - `just clippy`

  Michael Bolin · 2026-02-23 20:10:22 -08:00
  38f84b6b29
• app-server: fix connecting via websockets with Sec-WebSocket-Extensions: permessage-deflate (#12629)

  # External (non-OpenAI) Pull Request Requirements
  
  Before opening this Pull Request, please read the dedicated
  "Contributing" markdown file or your PR may be closed:
  https://github.com/openai/codex/blob/main/docs/contributing.md
  
  If your PR conforms to our contribution guidelines, replace this text
  with a detailed and high quality description of your changes.
  
  Include a link to a bug report or enhancement request.

  Javi · 2026-02-24 02:41:03 +00:00
  5a3bdcb27b
• Update models.json (#11408)

  Automated update of models.json.
  
  ---------
  
  Co-authored-by: sayan-oai <244841968+sayan-oai@users.noreply.github.com>
  Co-authored-by: sayan-oai <sayan@openai.com>

  github-actions[bot] · 2026-02-23 18:37:31 -08:00
  d580995957
• Handle realtime spawn_transcript delegation (#12619)
  Ahmed Ibrahim · 2026-02-23 14:39:07 -08:00
  10a3adad8e
• voice transcription (#3381)

  Adds voice transcription on press-and-hold of spacebar.
  
  
  https://github.com/user-attachments/assets/85039314-26f3-46d1-a83b-8c4a4a1ecc21
  
  ---------
  
  Co-authored-by: Codex <199175422+chatgpt-codex-connector[bot]@users.noreply.github.com>
  Co-authored-by: David Zbarsky <zbarsky@openai.com>

  Jeremy Rose · 2026-02-23 22:15:18 +00:00
  855e275591
• fix: show command running in background terminal in details under status indicator (#12549)

  #### What
  Display in-progress background terminal command in `status.details`
  (right under header) rather than inline, as it gets cut off currently.
  
  ###### Before
  <img width="993" height="395" alt="image"
  src="https://github.com/user-attachments/assets/6792b666-8184-40f7-bf29-409bb06c21d5"
  />
  
  ###### After
  <img width="469" height="137" alt="image"
  src="https://github.com/user-attachments/assets/4d6a2481-bd19-4333-8c1a-92f521b09b3d"
  />
  
  #### Tests
  Added/updated tests

  sayan-oai · 2026-02-23 21:04:24 +00:00
  50953ea39a
• chore(deps): bump owo-colors from 4.2.3 to 4.3.0 in /codex-rs (#12530)

  Bumps [owo-colors](https://github.com/owo-colors/owo-colors) from 4.2.3
  to 4.3.0.
  <details>
  <summary>Release notes</summary>
  <p><em>Sourced from <a
  href="https://github.com/owo-colors/owo-colors/releases">owo-colors's
  releases</a>.</em></p>
  <blockquote>
  <h2>owo-colors 4.3.0</h2>
  <h3>Fixed</h3>
  <ul>
  <li>Scripts in the <code>scripts/</code> directory are no longer
  published in the crate package. Thanks <a
  href="https://redirect.github.com/owo-colors/owo-colors/pull/152">weiznich</a>
  for your first contribution!</li>
  </ul>
  <h3>Changed</h3>
  <ul>
  <li>
  <p>Mark methods with
  <code>#[rust_analyzer::completions(ignore_flyimport)]</code> and the
  <code>OwoColorize</code> trait with
  <code>#[rust_analyzer::completions(ignore_flyimport_methods)]</code>.
  This prevents owo-colors methods from being completed with rust-analyzer
  unless the <code>OwoColorize</code> trait is included.</p>
  <p>Unfortunately, this also breaks explicit autocomplete commands such
  as Ctrl-Space in many editors. (The language server protocol doesn't
  appear to have a way to differentiate between implicit and explicit
  autocomplete commands.) On balance we believe this is the right
  approach, but please do provide feedback on [PR <a
  href="https://redirect.github.com/owo-colors/owo-colors/issues/141">#141</a>](<a
  href="https://redirect.github.com/owo-colors/owo-colors/pull/141">owo-colors/owo-colors#141</a>)
  if it negatively affects you.</p>
  </li>
  <li>
  <p>Updated MSRV to Rust 1.81.</p>
  </li>
  </ul>
  </blockquote>
  </details>
  <details>
  <summary>Changelog</summary>
  <p><em>Sourced from <a
  href="https://github.com/owo-colors/owo-colors/blob/main/CHANGELOG.md">owo-colors's
  changelog</a>.</em></p>
  <blockquote>
  <h2>[4.3.0] - 2026-02-22</h2>
  <h3>Fixed</h3>
  <ul>
  <li>Scripts in the <code>scripts/</code> directory are no longer
  published in the crate package. Thanks <a
  href="https://redirect.github.com/owo-colors/owo-colors/pull/152">weiznich</a>
  for your first contribution!</li>
  </ul>
  <h3>Changed</h3>
  <ul>
  <li>
  <p>Mark methods with
  <code>#[rust_analyzer::completions(ignore_flyimport)]</code> and the
  <code>OwoColorize</code> trait with
  <code>#[rust_analyzer::completions(ignore_flyimport_methods)]</code>.
  This prevents owo-colors methods from being completed with rust-analyzer
  unless the <code>OwoColorize</code> trait is included.</p>
  <p>Unfortunately, this also breaks explicit autocomplete commands such
  as Ctrl-Space in many editors. (The language server protocol doesn't
  appear to have a way to differentiate between implicit and explicit
  autocomplete commands.) On balance we believe this is the right
  approach, but please do provide feedback on [PR <a
  href="https://redirect.github.com/owo-colors/owo-colors/issues/141">#141</a>](<a
  href="https://redirect.github.com/owo-colors/owo-colors/pull/141">owo-colors/owo-colors#141</a>)
  if it negatively affects you.</p>
  </li>
  <li>
  <p>Updated MSRV to Rust 1.81.</p>
  </li>
  </ul>
  </blockquote>
  </details>
  <details>
  <summary>Commits</summary>
  <ul>
  <li><a
  href="https://github.com/owo-colors/owo-colors/commit/baf10f9a74007501379afdc002394a15a787dc59"><code>baf10f9</code></a>
  [owo-colors] version 4.3.0</li>
  <li><a
  href="https://github.com/owo-colors/owo-colors/commit/6abe2026c56ef6eefda30d417462f28f9ea8854d"><code>6abe202</code></a>
  [meta] prepare changelog</li>
  <li><a
  href="https://github.com/owo-colors/owo-colors/commit/ca814470410c819568b381b9b2b3f7a184e7de28"><code>ca81447</code></a>
  [RFC] add ignore_flyimport and ignore_flyimport_methods (<a
  href="https://redirect.github.com/owo-colors/owo-colors/issues/141">#141</a>)</li>
  <li><a
  href="https://github.com/owo-colors/owo-colors/commit/61de72e7f9487a945e8ea59afb9aa1b2046a2bd7"><code>61de72e</code></a>
  Exclude development script from published package (<a
  href="https://redirect.github.com/owo-colors/owo-colors/issues/152">#152</a>)</li>
  <li><a
  href="https://github.com/owo-colors/owo-colors/commit/b2ad6bcd4156207bcbc4c756c0c926e02667a872"><code>b2ad6bc</code></a>
  update MSRV to Rust 1.81 (<a
  href="https://redirect.github.com/owo-colors/owo-colors/issues/156">#156</a>)</li>
  <li>See full diff in <a
  href="https://github.com/owo-colors/owo-colors/compare/v4.2.3...v4.3.0">compare
  view</a></li>
  </ul>
  </details>
  <br />
  
  
  [![Dependabot compatibility
  score](https://dependabot-badges.githubapp.com/badges/compatibility_score?dependency-name=owo-colors&package-manager=cargo&previous-version=4.2.3&new-version=4.3.0)](https://docs.github.com/en/github/managing-security-vulnerabilities/about-dependabot-security-updates#about-compatibility-scores)
  
  Dependabot will resolve any conflicts with this PR as long as you don't
  alter it yourself. You can also trigger a rebase manually by commenting
  `@dependabot rebase`.
  
  [//]: # (dependabot-automerge-start)
  [//]: # (dependabot-automerge-end)
  
  ---
  
  <details>
  <summary>Dependabot commands and options</summary>
  <br />
  
  You can trigger Dependabot actions by commenting on this PR:
  - `@dependabot rebase` will rebase this PR
  - `@dependabot recreate` will recreate this PR, overwriting any edits
  that have been made to it
  - `@dependabot show <dependency name> ignore conditions` will show all
  of the ignore conditions of the specified dependency
  - `@dependabot ignore this major version` will close this PR and stop
  Dependabot creating any more for this major version (unless you reopen
  the PR or upgrade to it yourself)
  - `@dependabot ignore this minor version` will close this PR and stop
  Dependabot creating any more for this minor version (unless you reopen
  the PR or upgrade to it yourself)
  - `@dependabot ignore this dependency` will close this PR and stop
  Dependabot creating any more for this dependency (unless you reopen the
  PR or upgrade to it yourself)
  
  
  </details>
  
  Signed-off-by: dependabot[bot] <support@github.com>
  Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

  dependabot[bot] · 2026-02-23 13:01:15 -08:00
  cd5acf6af7
• fix(tui): queue steer Enter while final answer is still streaming to prevent dead state (#12569)

  ## Summary
  This fixes a TUI race (https://github.com/openai/codex/issues/11008)
  where pressing Enter with Steer enabled while the assistant is still
  streaming the final answer could put Codex into a non-recoverable
  “running” state (no further prompts handled until exiting and resuming).
  
  ## Root Cause
  In steer mode, `InputResult::Submitted` could submit immediately even
  while a final-answer stream was active. That immediate submission races
  with turn completion and can strand turn state.
  
  ## Fix
  When handling `InputResult::Submitted`, we now queue instead of
  immediate-submit if a final-answer stream is active
  (`stream_controller.is_some()`).
  
  This keeps behavior deterministic:
  - Prompt is preserved in the queue.
  - `on_task_complete()` drains queued input through
  `maybe_send_next_queued_input()`.
  - Follow-up prompts continue in FIFO order after completion.
  
  ## Why this resolves the “dead mode”
  The problematic timing window is now converted into queueing, so prompts
  entered during final streaming are not lost and are processed after the
  current output ends. The model continues handling prompts normally
  without requiring `/quit` + `resume`.
  
  ## Tests
  Added regression coverage in `tui/src/chatwidget/tests.rs`:
  
  - `steer_enter_queues_while_final_answer_stream_is_active`
  - `steer_enter_during_final_stream_preserves_follow_up_prompts_in_order`
  
  Both fail on old behavior and pass with this fix.

  Beehive Innovations · 2026-02-23 12:58:40 -08:00
  be4203023d
• fix(tui): recover on owned wrap mapping mismatch (#12609)

  ## Summary
  
  - Replace the `panic!` in `map_owned_wrapped_line_to_range` with a
  recoverable flow that skips synthetic leading characters, logs a warning
  on mid-line mismatch, and returns the mapped prefix range instead of
  crashing
  - Fixes a crash when `textwrap` produces owned lines with synthetic
  indent prefixes (e.g. non-space indents via
  `initial_indent`/`subsequent_indent`)
  
  ## Test plan
  
  - [x] Added unit test for direct mismatch recovery
  (`map_owned_wrapped_line_to_range_recovers_on_non_prefix_mismatch`)
  - [x] Added end-to-end `wrap_ranges` test with non-space indents that
  forces owned wrapped lines and validates full source reconstruction
  - [x] Verify no regressions in existing `wrapping.rs` tests (`cargo test
  -p codex-tui`)

  Felipe Coury · 2026-02-23 20:14:50 +00:00
  48e08a1561
• fix: add ellipsis for truncated status indicator (#12540)

  #### What
  
  - Add ellipsis truncation of the status indicator, similar to equivalent
  truncation done in the footer.
  - Extract truncation helpers into separate file
  
  
  
  https://github.com/user-attachments/assets/a2d5f22f-8adc-456e-8059-97359194c25c
  
  
  #### Tests
  Updated relevant snapshot tests

  sayan-oai · 2026-02-23 11:45:46 -08:00
  bfe622f495
• Use Arc-based ToolCtx in tool runtimes (#12583)

  ## Why
  Tool handlers and runtimes needed to pass the same turn/session context
  for shell and non-shell workflows without duplicative ownership churn.
  Using shared pointers avoids temporary lifetimes and keeps existing
  behavior unchanged while simplifying call sites.
  
  ## What changed
  - Converted `ToolCtx` to store shared context handles (`Arc`-based),
  including updates across shell, apply-patch, and unified-exec paths.
  - Updated orchestrator/runtime call sites to consume the shared context
  consistently and remove brittle move/borrow patterns.
  - Kept behavior unchanged while preparing the type surface for the new
  shell escalation integration in the next stack commit.
  
  ## Verification
  - Validated this commit stack point with `just clippy` and confirmed
  workspace compiles cleanly in this stack state.
  
  [//]: # (BEGIN SAPLING FOOTER)
  Stack created with [Sapling](https://sapling-scm.com). Best reviewed
  with [ReviewStack](https://reviewstack.dev/openai/codex/pull/12583).
  * #12584
  * __->__ #12583
  * #12556

  Michael Bolin · 2026-02-23 18:29:26 +00:00
  7f75e74201
• chore(deps): bump syn from 2.0.114 to 2.0.117 in /codex-rs (#12529)

  Bumps [syn](https://github.com/dtolnay/syn) from 2.0.114 to 2.0.117.
  <details>
  <summary>Release notes</summary>
  <p><em>Sourced from <a
  href="https://github.com/dtolnay/syn/releases">syn's
  releases</a>.</em></p>
  <blockquote>
  <h2>2.0.117</h2>
  <ul>
  <li>Fix parsing of <code>self::</code> pattern in first function
  argument (<a
  href="https://redirect.github.com/dtolnay/syn/issues/1970">#1970</a>)</li>
  </ul>
  <h2>2.0.116</h2>
  <ul>
  <li>Optimize parse_fn_arg_or_variadic for less lookahead on erroneous
  receiver (<a
  href="https://redirect.github.com/dtolnay/syn/issues/1968">#1968</a>)</li>
  </ul>
  <h2>2.0.115</h2>
  <ul>
  <li>Enable GenericArgument::Constraint parsing in non-full mode (<a
  href="https://redirect.github.com/dtolnay/syn/issues/1966">#1966</a>)</li>
  </ul>
  </blockquote>
  </details>
  <details>
  <summary>Commits</summary>
  <ul>
  <li><a
  href="https://github.com/dtolnay/syn/commit/7bcb37cdb3399977658c8b52d2441d37e42e48f2"><code>7bcb37c</code></a>
  Release 2.0.117</li>
  <li><a
  href="https://github.com/dtolnay/syn/commit/9c6e7d3b8df7b30909d60395f88a6ca07688e1c1"><code>9c6e7d3</code></a>
  Merge pull request <a
  href="https://redirect.github.com/dtolnay/syn/issues/1970">#1970</a>
  from dtolnay/receiver</li>
  <li><a
  href="https://github.com/dtolnay/syn/commit/019a84847eded0cdb1f7856e0752ba618155cfc9"><code>019a848</code></a>
  Fix self:: pattern in first function argument</li>
  <li><a
  href="https://github.com/dtolnay/syn/commit/23f54f3cf61ddedd5daea4f347eca2d4b84c8abb"><code>23f54f3</code></a>
  Update test suite to nightly-2026-02-18</li>
  <li><a
  href="https://github.com/dtolnay/syn/commit/b99b9a627c46580343398472e7b08a131357a994"><code>b99b9a6</code></a>
  Unpin CI miri toolchain</li>
  <li><a
  href="https://github.com/dtolnay/syn/commit/a62e54a48b3b05add5df0e80fe93906509ad72ae"><code>a62e54a</code></a>
  Release 2.0.116</li>
  <li><a
  href="https://github.com/dtolnay/syn/commit/5a8ed9f32e572f35a952c05f25beb3bd976300a4"><code>5a8ed9f</code></a>
  Merge pull request <a
  href="https://redirect.github.com/dtolnay/syn/issues/1968">#1968</a>
  from dtolnay/receiver</li>
  <li><a
  href="https://github.com/dtolnay/syn/commit/813afcc7733b02a8ad0a829eef431e593a906379"><code>813afcc</code></a>
  Optimize parse_fn_arg_or_variadic for less lookahead on erroneous
  receiver</li>
  <li><a
  href="https://github.com/dtolnay/syn/commit/c17215011363b8e936b98a9053726abfbc2bdcc4"><code>c172150</code></a>
  Add regression test for issue 1718</li>
  <li><a
  href="https://github.com/dtolnay/syn/commit/0071ab367ca6c42f94209f8187de3e540231427f"><code>0071ab3</code></a>
  Ignore type_complexity clippy lint</li>
  <li>Additional commits viewable in <a
  href="https://github.com/dtolnay/syn/compare/2.0.114...2.0.117">compare
  view</a></li>
  </ul>
  </details>
  <br />
  
  
  [![Dependabot compatibility
  score](https://dependabot-badges.githubapp.com/badges/compatibility_score?dependency-name=syn&package-manager=cargo&previous-version=2.0.114&new-version=2.0.117)](https://docs.github.com/en/github/managing-security-vulnerabilities/about-dependabot-security-updates#about-compatibility-scores)
  
  Dependabot will resolve any conflicts with this PR as long as you don't
  alter it yourself. You can also trigger a rebase manually by commenting
  `@dependabot rebase`.
  
  [//]: # (dependabot-automerge-start)
  [//]: # (dependabot-automerge-end)
  
  ---
  
  <details>
  <summary>Dependabot commands and options</summary>
  <br />
  
  You can trigger Dependabot actions by commenting on this PR:
  - `@dependabot rebase` will rebase this PR
  - `@dependabot recreate` will recreate this PR, overwriting any edits
  that have been made to it
  - `@dependabot show <dependency name> ignore conditions` will show all
  of the ignore conditions of the specified dependency
  - `@dependabot ignore this major version` will close this PR and stop
  Dependabot creating any more for this major version (unless you reopen
  the PR or upgrade to it yourself)
  - `@dependabot ignore this minor version` will close this PR and stop
  Dependabot creating any more for this minor version (unless you reopen
  the PR or upgrade to it yourself)
  - `@dependabot ignore this dependency` will close this PR and stop
  Dependabot creating any more for this dependency (unless you reopen the
  PR or upgrade to it yourself)
  
  
  </details>
  
  Signed-off-by: dependabot[bot] <support@github.com>
  Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

  dependabot[bot] · 2026-02-23 10:25:05 -08:00
  fec517cd38
• chore(deps): bump libc from 0.2.180 to 0.2.182 in /codex-rs (#12528)

  Bumps [libc](https://github.com/rust-lang/libc) from 0.2.180 to 0.2.182.
  <details>
  <summary>Release notes</summary>
  <p><em>Sourced from <a
  href="https://github.com/rust-lang/libc/releases">libc's
  releases</a>.</em></p>
  <blockquote>
  <h2>0.2.182</h2>
  <h3>Added</h3>
  <ul>
  <li>Android, Linux: Add <code>tgkill</code> (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4970">#4970</a>)</li>
  <li>Redox: Add <code>RENAME_NOREPLACE</code> (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4968">#4968</a>)</li>
  <li>Redox: Add <code>renameat2</code> (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4968">#4968</a>)</li>
  </ul>
  <h2>0.2.181</h2>
  <h3>Added</h3>
  <ul>
  <li>Apple: Add <code>MADV_ZERO</code> (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4924">#4924</a>)</li>
  <li>Redox: Add <code>makedev</code>, <code>major</code>, and
  <code>minor</code> (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4928">#4928</a>)</li>
  <li>GLibc: Add <code>PTRACE_SET_SYSCALL_INFO</code> (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4933">#4933</a>)</li>
  <li>OpenBSD: Add more kqueue related constants for (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4945">#4945</a>)</li>
  <li>Linux: add CAN error types (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4944">#4944</a>)</li>
  <li>OpenBSD: Add siginfo_t::si_status (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4946">#4946</a>)</li>
  <li>QNX NTO: Add <code>max_align_t</code> (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4927">#4927</a>)</li>
  <li>Illumos: Add <code>_CS_PATH</code> (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4956">#4956</a>)</li>
  <li>OpenBSD: add <code>ppoll</code> (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4957">#4957</a>)</li>
  </ul>
  <h3>Fixed</h3>
  <ul>
  <li><strong>Breaking</strong>: Redox: Fix the type of <code>dev_t</code>
  (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4928">#4928</a>)</li>
  <li>AIX: Change 'tv_nsec' of 'struct timespec' to type 'c_long' (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4931">#4931</a>)</li>
  <li>AIX: Use 'struct st_timespec' in 'struct stat{,64}' (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4931">#4931</a>)</li>
  <li>Glibc: Link old version of <code>tc{g,s}etattr</code> (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4938">#4938</a>)</li>
  <li>Glibc: Link the correct version of <code>cf{g,s}et{i,o}speed</code>
  on mips{32,64}r6 (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4938">#4938</a>)</li>
  <li>OpenBSD: Fix constness of tm.tm_zone (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4948">#4948</a>)</li>
  <li>OpenBSD: Fix the definition of <code>ptrace_thread_state</code> (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4947">#4947</a>)</li>
  <li>QuRT: Fix type visibility and defs (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4932">#4932</a>)</li>
  <li>Redox: Fix values for <code>PTHREAD_MUTEX_{NORMAL, RECURSIVE}</code>
  (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4943">#4943</a>)</li>
  <li>Various: Mark additional fields as private padding (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4922">#4922</a>)</li>
  </ul>
  <h3>Changed</h3>
  <ul>
  <li>Fuchsia: Update <code>SO_*</code> constants (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4937">#4937</a>)</li>
  <li>Revert &quot;musl: convert inline timespecs to timespec&quot;
  (resolves build issues on targets only supported by Musl 1.2.3+ ) (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4958">#4958</a>)</li>
  </ul>
  </blockquote>
  </details>
  <details>
  <summary>Changelog</summary>
  <p><em>Sourced from <a
  href="https://github.com/rust-lang/libc/blob/0.2.182/CHANGELOG.md">libc's
  changelog</a>.</em></p>
  <blockquote>
  <h2><a
  href="https://github.com/rust-lang/libc/compare/0.2.181...0.2.182">0.2.182</a>
  - 2026-02-13</h2>
  <h3>Added</h3>
  <ul>
  <li>Android, Linux: Add <code>tgkill</code> (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4970">#4970</a>)</li>
  <li>Redox: Add <code>RENAME_NOREPLACE</code> (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4968">#4968</a>)</li>
  <li>Redox: Add <code>renameat2</code> (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4968">#4968</a>)</li>
  </ul>
  <h2><a
  href="https://github.com/rust-lang/libc/compare/0.2.180...0.2.181">0.2.181</a>
  - 2026-02-09</h2>
  <h3>Added</h3>
  <ul>
  <li>Apple: Add <code>MADV_ZERO</code> (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4924">#4924</a>)</li>
  <li>Redox: Add <code>makedev</code>, <code>major</code>, and
  <code>minor</code> (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4928">#4928</a>)</li>
  <li>GLibc: Add <code>PTRACE_SET_SYSCALL_INFO</code> (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4933">#4933</a>)</li>
  <li>OpenBSD: Add more kqueue related constants for (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4945">#4945</a>)</li>
  <li>Linux: add CAN error types (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4944">#4944</a>)</li>
  <li>OpenBSD: Add siginfo_t::si_status (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4946">#4946</a>)</li>
  <li>QNX NTO: Add <code>max_align_t</code> (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4927">#4927</a>)</li>
  <li>Illumos: Add <code>_CS_PATH</code> (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4956">#4956</a>)</li>
  <li>OpenBSD: add <code>ppoll</code> (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4957">#4957</a>)</li>
  </ul>
  <h3>Fixed</h3>
  <ul>
  <li><strong>breaking</strong>: Redox: Fix the type of dev_t (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4928">#4928</a>)</li>
  <li>AIX: Change 'tv_nsec' of 'struct timespec' to type 'c_long' (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4931">#4931</a>)</li>
  <li>AIX: Use 'struct st_timespec' in 'struct stat{,64}' (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4931">#4931</a>)</li>
  <li>Glibc: Link old version of <code>tc{g,s}etattr</code> (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4938">#4938</a>)</li>
  <li>Glibc: Link the correct version of <code>cf{g,s}et{i,o}speed</code>
  on mips{32,64}r6 (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4938">#4938</a>)</li>
  <li>OpenBSD: Fix constness of tm.tm_zone (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4948">#4948</a>)</li>
  <li>OpenBSD: Fix the definition of <code>ptrace_thread_state</code> (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4947">#4947</a>)</li>
  <li>QuRT: Fix type visibility and defs (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4932">#4932</a>)</li>
  <li>Redox: Fix values for <code>PTHREAD_MUTEX_{NORMAL, RECURSIVE}</code>
  (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4943">#4943</a>)</li>
  <li>Various: Mark additional fields as private padding (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4922">#4922</a>)</li>
  </ul>
  <h3>Changed</h3>
  <ul>
  <li>Fuchsia: Update <code>SO_*</code> constants (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4937">#4937</a>)</li>
  <li>Revert &quot;musl: convert inline timespecs to timespec&quot;
  (resolves build issues on targets only supported by Musl 1.2.3+ ) (<a
  href="https://redirect.github.com/rust-lang/libc/pull/4958">#4958</a>)</li>
  </ul>
  </blockquote>
  </details>
  <details>
  <summary>Commits</summary>
  <ul>
  <li><a
  href="https://github.com/rust-lang/libc/commit/e879ee90b6cd8f79b352d4d4d1f8ca05f94f2f53"><code>e879ee9</code></a>
  chore: Release libc 0.2.182</li>
  <li><a
  href="https://github.com/rust-lang/libc/commit/2efe72f4dae6feebacaf5ec8a4ec5fdc79569e7b"><code>2efe72f</code></a>
  remove copyright year in LICENSE-MIT</li>
  <li><a
  href="https://github.com/rust-lang/libc/commit/634bc4e66e944d54ebc3d1610175c8c6d390bd29"><code>634bc4e</code></a>
  ci: Update the list of tested and documented targets</li>
  <li><a
  href="https://github.com/rust-lang/libc/commit/d7aa109ab5074dbbd35fb52cc72620e29961e76d"><code>d7aa109</code></a>
  Revert &quot;Disable hexagon-unknown-linux-musl testing for
  now&quot;</li>
  <li><a
  href="https://github.com/rust-lang/libc/commit/14e2f5641e2d4356953b0c95959ccfc86af5dcc3"><code>14e2f56</code></a>
  Revert &quot;ci: Skip hexagon-unknown-linux-musl&quot;</li>
  <li><a
  href="https://github.com/rust-lang/libc/commit/b7807c369b468c369661e81ea6f9f649f3b3ddf3"><code>b7807c3</code></a>
  Revert &quot;aix: Temporarily skip checking powerpc64-ibm-aix
  builds&quot;</li>
  <li><a
  href="https://github.com/rust-lang/libc/commit/abe93a0bfedfe6159252d43e5c4273d0b0833ca4"><code>abe93a0</code></a>
  feat(linux): add <code>tgkill</code> for Linux and Android</li>
  <li><a
  href="https://github.com/rust-lang/libc/commit/25f7dde943988c81871d95aaea1afd49cf11425d"><code>25f7dde</code></a>
  feat(redox): add <code>RENAME_NOREPLACE</code></li>
  <li><a
  href="https://github.com/rust-lang/libc/commit/4b4ce4f2205d22121c5e913b118f8fc776d39897"><code>4b4ce4f</code></a>
  feat(redox): add <code>renameat2</code></li>
  <li><a
  href="https://github.com/rust-lang/libc/commit/ab8c36c49327eeee2b5c3818d6706b499dd890a4"><code>ab8c36c</code></a>
  build(deps): bump vmactions/solaris-vm from 1.2.8 to 1.3.0</li>
  <li>Additional commits viewable in <a
  href="https://github.com/rust-lang/libc/compare/0.2.180...0.2.182">compare
  view</a></li>
  </ul>
  </details>
  <br />
  
  
  [![Dependabot compatibility
  score](https://dependabot-badges.githubapp.com/badges/compatibility_score?dependency-name=libc&package-manager=cargo&previous-version=0.2.180&new-version=0.2.182)](https://docs.github.com/en/github/managing-security-vulnerabilities/about-dependabot-security-updates#about-compatibility-scores)
  
  Dependabot will resolve any conflicts with this PR as long as you don't
  alter it yourself. You can also trigger a rebase manually by commenting
  `@dependabot rebase`.
  
  [//]: # (dependabot-automerge-start)
  [//]: # (dependabot-automerge-end)
  
  ---
  
  <details>
  <summary>Dependabot commands and options</summary>
  <br />
  
  You can trigger Dependabot actions by commenting on this PR:
  - `@dependabot rebase` will rebase this PR
  - `@dependabot recreate` will recreate this PR, overwriting any edits
  that have been made to it
  - `@dependabot show <dependency name> ignore conditions` will show all
  of the ignore conditions of the specified dependency
  - `@dependabot ignore this major version` will close this PR and stop
  Dependabot creating any more for this major version (unless you reopen
  the PR or upgrade to it yourself)
  - `@dependabot ignore this minor version` will close this PR and stop
  Dependabot creating any more for this minor version (unless you reopen
  the PR or upgrade to it yourself)
  - `@dependabot ignore this dependency` will close this PR and stop
  Dependabot creating any more for this dependency (unless you reopen the
  PR or upgrade to it yourself)
  
  
  </details>
  
  Signed-off-by: dependabot[bot] <support@github.com>
  Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

  dependabot[bot] · 2026-02-23 10:24:35 -08:00
  5c52ef8e60
• app-server: box request dispatch future to reduce stack pressure (#12421)
  Charley Cunningham · 2026-02-23 10:05:41 -08:00
  3cea3e665e
• refactor: normalize unix module layout for exec-server and shell-escalation (#12556)

  ## Why
  Shell execution refactoring in `exec-server` had become split between
  duplicated code paths, which blocked a clean introduction of the new
  reusable shell escalation flow. This commit creates a dedicated
  foundation crate so later shell tooling changes can share one
  implementation.
  
  ## What changed
  - Added the `codex-shell-escalation` crate and moved the core escalation
  pieces (`mcp` protocol/socket/session flow, policy glue) that were
  previously in `exec-server` into it.
  - Normalized `exec-server` Unix structure under a dedicated `unix`
  module layout and kept non-Unix builds narrow.
  - Wired crate/build metadata so `shell-escalation` is a first-class
  workspace dependency for follow-on integration work.
  
  ## Verification
  - Built and linted the stack at this commit point with `just clippy`.
  
  [//]: # (BEGIN SAPLING FOOTER)
  Stack created with [Sapling](https://sapling-scm.com). Best reviewed
  with [ReviewStack](https://reviewstack.dev/openai/codex/pull/12556).
  * #12584
  * #12583
  * __->__ #12556

  Michael Bolin · 2026-02-23 09:28:17 -08:00
  5221575f23
• tweaked /clear to support clear + new chat, also fix minor bug for macos terminal (#12520)

  # /clear feature! 
  
  Use /clear to start a new chat with Codex on a clean terminal!

  Won Park · 2026-02-23 09:11:05 -08:00
  a606e85859
• remove feature flag collaboration modes (#12028)

  All code should go in the direction that steer is enabled
  
  ---------
  
  Co-authored-by: Codex <noreply@openai.com>

  Ahmed Ibrahim · 2026-02-23 09:06:08 -08:00
  6e60f724bc
• chore: better bazel test logs (#12576)

  ## Summary
  
  Improve Bazel CI failure diagnostics by printing the tail of each failed
  target’s test.log directly in the GitHub Actions output.
  
  Today, when a large Bazel test target fails (for example tests of
  `codex-core`), the workflow often only shows a target-level Exit 101
  plus a path to Bazel’s test.log. That makes it hard to see the actual
  failing Rust test and panic without digging into artifacts or
  reproducing locally.
  
  This change makes the workflow automatically surface that information
  inline.
  
    ## What Changed
  
  In .github/workflows/bazel.yml:
  
    - Capture Bazel console output via tee
    - Preserve the Bazel exit code when piping (PIPESTATUS[0])
    - On failure:
        - Parse failed Bazel test targets from FAIL: //... lines
        - Resolve Bazel test log directory via bazel info bazel-testlogs
        - Print tail -n 200 for each failed target’s test.log
        - Group each target’s output in GitHub Actions logs (::group::)
  
  ## Bonus
  Disable `experimental_remote_repo_contents_cache` to prevent "Permission
  Denied"

  jif-oai · 2026-02-23 08:13:29 -08:00
  3b6c50d925
• feat: land sqlite (#12141)
  jif-oai · 2026-02-23 16:12:23 +00:00
  eace7c6610
• feat: role metrics multi-agent (#12579)

  add metrics for agent role

  jif-oai · 2026-02-23 15:55:48 +00:00
  2119532a81
• Allow exec resume to parse output-last-message flag after command (#12541)

  Summary
  - mark `output-last-message` as a global exec flag so it can follow
  subcommands like `resume`
  - add regression tests in both `cli` and `exec` crates verifying the
  flag order works when invoking `resume`
  
  Fixes #12538

  Eric Traut · 2026-02-23 07:55:37 -08:00
  862a5b3eb3
• chore: rename memory feature flag (#12580)

  `memory_tool` -> `memories`

  jif-oai · 2026-02-23 15:37:12 +00:00
  e8709bc11a
• feat: add uuid helper (#12500)
  jif-oai · 2026-02-23 14:14:36 +00:00
  764ac9449f
• feat: agent nick names to model (#12575)
  jif-oai · 2026-02-23 13:44:37 +00:00
  cf0210bf22
• feat: keep dead agents in the agent picker (#12570)
  jif-oai · 2026-02-23 12:58:55 +00:00
  829d1080f6
• fix: TUI constraint (#12571)
  jif-oai · 2026-02-23 12:49:54 +00:00
  9d826a20c6
• chore: phase 2 name (#12568)
  jif-oai · 2026-02-23 11:04:55 +00:00
  6fbf19ef5f
• chore: add doc to memories (#12565)

  ]

  jif-oai · 2026-02-23 10:52:58 +00:00
  2b9d0c385f
• chore: awaiter (#12562)
  jif-oai · 2026-02-23 10:28:24 +00:00
  cfcbff4c48
• chore: nit name (#12559)
  jif-oai · 2026-02-23 08:49:41 +00:00
  8e9312958d
• refactor: decouple MCP policy construction from escalate server (#12555)

  ## Why
  The current escalate path in `codex-rs/exec-server` still had policy
  creation coupled to MCP details, which makes it hard to reuse the shell
  execution flow outside the MCP server. This change is part of a broader
  goal to split MCP-specific behavior from shared escalation execution so
  other handlers (for example a future `ShellCommandHandler`) can reuse it
  without depending on MCP request context types.
  
  ## What changed
  - Added a new `EscalationPolicyFactory` abstraction in `mcp.rs`:
    - `crate`-relative path: `codex-rs/exec-server/src/posix/mcp.rs`
  -
  https://github.com/openai/codex/blob/main/codex-rs/exec-server/src/posix/mcp.rs#L87-L107
  - Made `run_escalate_server` in `mcp.rs` accept a policy factory instead
  of constructing `McpEscalationPolicy` directly.
  -
  https://github.com/openai/codex/blob/main/codex-rs/exec-server/src/posix/mcp.rs#L178-L201
  - Introduced `McpEscalationPolicyFactory` that stores MCP-only state
  (`RequestContext`, `preserve_program_paths`) and implements the new
  trait.
  -
  https://github.com/openai/codex/blob/main/codex-rs/exec-server/src/posix/mcp.rs#L100-L117
  - Updated `shell()` to pass a `McpEscalationPolicyFactory` instance into
  `run_escalate_server`, so the server remains the MCP-specific wiring
  layer.
  -
  https://github.com/openai/codex/blob/main/codex-rs/exec-server/src/posix/mcp.rs#L163-L170
  
  ## Verification
  - Build and test execution was not re-run in this pass; changes are
  limited to `mcp.rs` and preserve the existing escalation flow semantics
  by only extracting policy construction behind a factory.
  
  
  
  
  ---
  [//]: # (BEGIN SAPLING FOOTER)
  Stack created with [Sapling](https://sapling-scm.com). Best reviewed
  with [ReviewStack](https://reviewstack.dev/openai/codex/pull/12555).
  * #12556
  * __->__ #12555

  Michael Bolin · 2026-02-23 00:31:29 -08:00
  956f2f439e
• Return image content from view_image (#12553)

  Responses API supports image content

  pakrym-oai · 2026-02-22 23:00:08 -08:00
  335a4e1cbc
• test: vendor zsh fork via DotSlash and stabilize zsh-fork tests (#12518)

  ## Why
  
  The zsh integration tests were still brittle in two ways:
  
  - they relied on `CODEX_TEST_ZSH_PATH` / environment-specific setup, so
  they often did not exercise the patched zsh fork that `shell-tool-mcp`
  ships
  - once the tests consistently used the vendored zsh fork, they exposed
  real Linux-specific zsh-fork issues in CI
  
  In particular, the Linux failures were not just test noise:
  
  - the zsh-fork launch path was dropping `ExecRequest.arg0`, so Linux
  `codex-linux-sandbox` arg0 dispatch did not run and zsh wrapper-mode
  could receive malformed arguments
  - the
  `turn_start_shell_zsh_fork_subcommand_decline_marks_parent_declined_v2`
  test uses the zsh exec bridge (which talks to the parent over a Unix
  socket), but Linux restricted sandbox seccomp denies `connect(2)`,
  causing timeouts on `ubuntu-24.04` x86/arm
  
  This PR makes the zsh tests consistently run against the intended
  vendored zsh fork and fixes/hardens the zsh-fork path so the Linux CI
  signal is meaningful.
  
  ## What Changed
  
  - Added a single shared test-only DotSlash file for the patched zsh fork
  at `codex-rs/exec-server/tests/suite/zsh` (analogous to the existing
  `bash` test resource).
  - Updated both app-server and exec-server zsh tests to use that shared
  DotSlash zsh (no duplicate zsh DotSlash file, no `CODEX_TEST_ZSH_PATH`
  dependency).
  - Updated the app-server zsh-fork test helper to resolve the shared
  DotSlash zsh and avoid silently falling back to host zsh.
  - Kept the app-server zsh-fork tests configured via `config.toml`, using
  a test wrapper path where needed to force `zsh -df` (and rewrite `-lc`
  to `-c`) for the subcommand-decline test.
  - Hardened the app-server subcommand-decline zsh-fork test for CI
  variability:
    - tolerate an extra `/responses` POST with a no-op mock response
  - tolerate non-target approval ordering while remaining strict on the
  two `/usr/bin/true` approvals and decline behavior
  - use `DangerFullAccess` on Linux for this one test because it validates
  zsh approval flow, not Linux sandbox socket restrictions
  - Fixed zsh-fork process launching on Linux by preserving `req.arg0` in
  `ZshExecBridge::execute_shell_request(...)` so `codex-linux-sandbox`
  arg0 dispatch continues to work.
  - Moved `maybe_run_zsh_exec_wrapper_mode()` under
  `arg0_dispatch_or_else(...)` in `app-server` and `cli` so wrapper-mode
  handling coexists correctly with arg0-dispatched helper modes.
  - Consolidated duplicated `dotslash -- fetch` resolution logic into
  shared test support (`core/tests/common/lib.rs`).
  - Updated `codex-rs/exec-server/tests/suite/accept_elicitation.rs` to
  use DotSlash zsh and hardened the zsh elicitation test for Bazel/zsh
  differences by:
    - resolving an absolute `git` path
    - running `git init --quiet .`
  - asserting success / `.git` creation instead of relying on banner text
  
  ## Verification
  
  - `cargo test -p codex-app-server turn_start_zsh_fork -- --nocapture`
  - `cargo test -p codex-exec-server accept_elicitation -- --nocapture`
  - `bazel test //codex-rs/exec-server:exec-server-all-test
  --test_output=streamed --test_arg=--nocapture
  --test_arg=accept_elicitation_for_prompt_rule_with_zsh`
  - CI (`rust-ci`) on the final cleaned commit: `Tests — ubuntu-24.04 -
  x86_64-unknown-linux-gnu` and `Tests — ubuntu-24.04-arm -
  aarch64-unknown-linux-gnu` passed in [run
  22291424358](https://github.com/openai/codex/actions/runs/22291424358)

  Michael Bolin · 2026-02-22 19:39:56 -08:00
  e8949f4507
• Add PR babysitting skill for this repo (#12513)

  ## PR Notes
  
  This PR adds a project-scoped `babysit-pr` skill for ongoing PR
  monitoring (CI, reviews, mergeability).
  
  Simply invoke this skill after creating a PR, and codex will do its best
  to get it to a mergeable state:
  
  ### What the skill does
  * Fixes CI failures related to the PR
  * Retries CI failures due to flaky tests
  * Addresses code review comments if it agrees with them
  * Addresses merge conflicts on main branch
  
  ### How the skill works
  - Polls PR status on a loop (CI checks, workflow runs, review activity,
  mergeability, and review decision).
  - Detects new review feedback (including inline comments and automated
  Codex review comments) and prompts/handles follow-up work.
  - Distinguishes pending vs failed vs passed CI and identifies likely
  flaky failures.
  - Can retry failed checks/workflows when appropriate.
  - Prioritizes actionable code review feedback over flaky CI retries (to
  avoid rerunning CI on a SHA that is about to be replaced).
  - Continues monitoring after fixes are applied and pushed, rather than
  stopping after a progress update.
  - Uses a slower backoff polling cadence once CI is green, while still
  watching for new review feedback or state changes.
  - Treats required review/approval as a blocking condition and keeps
  watching until the PR is actually merge-ready (or merged/closed, or
  human intervention is needed).
  
  ### Intended outcome
  
  Keep the PR moving with minimal manual babysitting by continuously
  watching for CI failures, reviewer feedback, and merge blockers, and
  responding in the right order until the PR is ready to merge.

  Eric Traut · 2026-02-22 15:36:28 -08:00
  7e569f1162
• Add C# syntax option to highlight selections (#12511)

  Summary
  - map csharp/c-sharp aliases to the existing C# syntax in the highlight
  matcher
  - ensure the extension list and tests include .cs and the new aliases so
  coverage stays accurate
  
  Testing
  
  <img width="543" height="266" alt="image"
  src="https://github.com/user-attachments/assets/e6c8a42f-649c-4c30-b574-421b4287534c"
  />

  Eric Traut · 2026-02-22 12:15:20 -08:00
  d5fef5c190
• Sort themes case-insensitively in picker (#12509)

  ## Summary
  - order bundled and custom themes together by name while keeping entries
  stable across platforms
  - update the theme fixture names and tests to assert case-insensitive
  ordering

  Eric Traut · 2026-02-22 12:12:36 -08:00
  5684c82e45
• Revert "Revert "Route inbound realtime text into turn start or steer"" (#12480)

  With working tests this time
  
  ---------
  
  Co-authored-by: Codex <noreply@openai.com>

  Ahmed Ibrahim · 2026-02-22 11:54:16 -08:00
  e00fa19328