## Why
Some models need to select their code-execution behavior through model
catalog metadata. Models without that metadata must continue to follow
the existing `CodeMode` and `CodeModeOnly` feature flags, including when
a newer server sends an enum value this client does not recognize.
## What changed
- add optional `ModelInfo.tool_mode` metadata with `direct`,
`code_mode`, and `code_mode_only`
- treat omitted and unknown wire values as `None`
- resolve `None` from the existing feature flags
- carry the resolved `ToolMode` directly on `TurnContext`, outside
`Config`
- use the resolved value for turn creation, model switches, review
turns, tool planning, and code execution
## Coverage
- add protocol coverage for omitted, known, and unknown enum values
- add focused coverage for flag fallback and explicit metadata
overriding feature flags
- add core integration coverage that fetches remote model metadata
through `/v1/models` and verifies the outbound `/responses` tools for
explicit `direct` and `code_mode_only` selectors
## Stack
- followed by #25032
## Summary
- Clarify default, omission, and bounded behavior across built-in tool
schemas, including unified exec, classic shell, Code Mode exec/wait,
multi-agent, agent job, MCP resource, image, goal, plan, tool_search,
and test-sync fields.
- Convert update_plan status to an enum and add short field descriptions
where the schema previously relied on surrounding context.
- Remove the dedicated permission-approval schema test and keep only
updates to existing expected-spec tests.
## Validation
- Ran `just fmt`.
- Ran `git diff --check`.
- Did not run clippy or tests, per request.
Regression has been eval
[here](https://openai.slack.com/archives/C09GDSP1J9X/p1779905065496949)
and we proved there are no regressions
## Why
This PR is stacked on #24918, which moves goal steering onto
source-labeled internal model context fragments. Active-turn goal
steering should use the same running-turn injection path as other
runtime steering, so those fragments enter the pending input queue as
`ResponseItem`s through the existing
[`Session::inject_if_running`](8d6f6cdf69/codex-rs/core/src/session/inject.rs (L12-L27))
behavior instead of through a goal-specific conversion wrapper.
## What Changed
- Exposes a narrow `CodexThread::inject_if_running` bridge for callers
that only hold a thread handle.
- Changes `ext/goal` active-turn steering to pass `ResponseItem`s
directly.
- Builds goal steering prompts as contextual internal model context
`ResponseItem`s before injecting them into the running turn.
## Testing
Not run locally; PR metadata update only.
## Why
Goal steering is one form of runtime-owned model context, but the old
`<goal_context>` wrapper made the contextual-fragment hiding path
goal-specific. Using a source-labeled internal context fragment gives
core and extensions a shared shape for hidden model steering while
keeping those prompts out of visible turn history.
The change also keeps legacy `<goal_context>` messages recognized as
hidden contextual input so existing stored history does not start
rendering old goal-steering prompts as user-visible turn items.
## What Changed
- Replaces `GoalContext` with `InternalModelContextFragment` plus a
validated `InternalContextSource`.
- Renders goal steering as `<codex_internal_context
source="goal">...</codex_internal_context>`.
- Updates core goal steering and `ext/goal` steering to inject the new
internal-context fragment.
- Updates contextual-fragment, event-mapping, goal, and session tests
for the new wrapper.
## Test Coverage
- Adds coverage for detecting the new internal model context fragment.
- Preserves coverage for hiding legacy `<goal_context>` fragments.
- Verifies invalid internal context sources are rejected and arbitrary
context tags are not hidden.
- Updates goal steering/session assertions to expect the new
`source="goal"` wrapper.
## Why
Permission profiles can mark filesystem entries as unreadable with
`deny` rules, including glob patterns. Several shell execution paths
treated known-safe commands or execpolicy `allow` rules as sufficient to
run outside the filesystem sandbox. That is not valid for read-capable
commands: for example, `cat` or `ls` may be reasonable to allow
generally, but dropping the sandbox would also drop deny-read
constraints such as `**/*.env`.
## What changed
- Added a shared check that treats active deny-read restrictions as
incompatible with unsandboxed execution.
- Kept first-attempt execution sandboxed for explicit escalation and
execpolicy allow bypasses when deny-read entries are present.
- Prevented no-sandbox retry after a sandbox denial when the active
filesystem policy contains deny-read entries.
- Updated the zsh-fork execve path so prefix-rule `allow` decisions
continue inside the current sandbox when deny-read restrictions are
active.
## Verification
- `cargo test -p codex-core tools::sandboxing::tests`
- `cargo test -p codex-core
tools::runtimes::shell::unix_escalation::tests`
- `cargo test -p codex-core
shell_command_enforces_glob_deny_read_policy`
## Why
#23813 switches the Windows sandbox runner path to `PermissionProfile`,
but it still left one runtime anchor for resolving symbolic
`:workspace_roots` entries. That is not enough once a turn has multiple
effective workspace roots: exact entries and deny globs under
`:workspace_roots` need to be materialized for every runtime root before
the command runner chooses token mode or builds ACL plans.
## What Changed
- Replaces the Windows runner/setup `permission_profile_cwd` plumbing
with `workspace_roots: Vec<AbsolutePathBuf>`.
- Resolves Windows-local `PermissionProfile` data with
`materialize_project_roots_with_workspace_roots(...)` instead of the
single-cwd helper.
- Threads `Config::effective_workspace_roots()` through core execution,
unified exec, TUI setup/read-grant flows, app-server setup, app-server
`command/exec`, and `debug sandbox` on Windows.
- Preserves those workspace roots through the zsh-fork escalation
executor instead of rebuilding them from `sandbox_policy_cwd`.
- Makes `ExecRequest::new(...)` and the remaining
`build_exec_request(...)` helper path take
`windows_sandbox_workspace_roots` explicitly so new call sites cannot
silently fall back to `vec![cwd]`.
- Clarifies the `debug sandbox` non-Windows comment: remaining
cwd-dependent resolution still uses `sandbox_policy_cwd`, while
`:workspace_roots` entries are already materialized from config roots.
- Updates elevated runner IPC `SpawnRequest` to send `workspace_roots`
and bumps the framed IPC protocol version to `3` for the payload shape
change.
- Adds Windows-local resolver coverage for expanding exact and glob
`:workspace_roots` entries across multiple roots, plus core helper
coverage proving explicit roots are preserved.
## Verification
- `cargo check -p codex-windows-sandbox -p codex-core -p codex-tui -p
codex-cli -p codex-app-server`
- `cargo test -p codex-windows-sandbox`
- `cargo test -p codex-core windows_sandbox`
- `cargo test -p codex-core unix_escalation`
- `cargo test -p codex-app-server windows_sandbox`
- `cargo test -p codex-tui windows_sandbox`
- `cargo test -p codex-cli debug_sandbox`
- `just test -p codex-core unified_exec`
- `just test -p codex-core
build_exec_request_preserves_windows_workspace_roots`
- `env -u CODEX_NETWORK_PROXY_ACTIVE -u
CODEX_NETWORK_ALLOW_LOCAL_BINDING just test -p codex-app-server --lib
command_exec`
- `just test -p codex-windows-sandbox`
- `just test -p codex-exec sandbox`
- `just fix -p codex-core -p codex-app-server -p codex-windows-sandbox`
A local macOS cross-check with `cargo check --target
x86_64-pc-windows-msvc ...` did not reach crate Rust code because native
dependencies require Windows SDK headers (`windows.h` / `assert.h`) in
this environment; Windows CI remains the real target validation.
Two local targeted filters compile but do not run assertions on macOS:
`env -u CODEX_NETWORK_PROXY_ACTIVE -u CODEX_NETWORK_ALLOW_LOCAL_BINDING
just test -p codex-app-server --lib command_exec_processor` matched zero
tests, and `just test -p codex-linux-sandbox landlock` matched zero
tests because the landlock suite is Linux-only.
## Summary
Some permission profiles can encode filesystem reads that should remain
unavailable to the agent. Before this change, the model-visible context
and automatic approval review prompt summarized the effective
permissions as a legacy sandbox mode, which can omit permission-profile
filesystem entries from escalation decisions.
For example, a profile can grant workspace access while denying a
private subtree across every workspace root:
```toml
default_permissions = "restricted-workspace"
[permissions.restricted-workspace.workspace_roots]
"/Users/alice/project" = true
"/Users/alice/other-project" = true
[permissions.restricted-workspace.filesystem]
":minimal" = "read"
[permissions.restricted-workspace.filesystem.":workspace_roots"]
"." = "write"
"private" = "deny"
"private/**" = "deny"
```
The context window now describes the workspace roots and effective
filesystem side of the `PermissionProfile` directly, with deny entries
marked as non-escalatable:
```xml
<environment_context>
<cwd>/Users/alice/project</cwd>
<shell>zsh</shell>
<filesystem><workspace_roots><root>/Users/alice/project</root><root>/Users/alice/other-project</root></workspace_roots><permission_profile type="managed"><file_system type="restricted"><entry access="read"><special>:minimal</special></entry><entry access="write"><path>/Users/alice/project</path></entry><entry access="write"><path>/Users/alice/other-project</path></entry><entry access="deny" escalatable="false"><path>/Users/alice/project/private</path></entry><entry access="deny" escalatable="false"><path>/Users/alice/other-project/private</path></entry><entry access="deny" escalatable="false"><glob>/Users/alice/project/private/**</glob></entry><entry access="deny" escalatable="false"><glob>/Users/alice/other-project/private/**</glob></entry></file_system></permission_profile></filesystem>
</environment_context>
```
Managed requirements can impose the same kind of deny-read restriction:
```toml
[permissions.filesystem]
deny_read = [
"/Users/alice/project/private",
"/Users/alice/project/private/**",
]
```
The automatic approval review prompt also receives the parent turn's
denied-read context, so review decisions can account for the active
permission profile.
## What Changed
- Render the effective filesystem profile in `<environment_context>`,
including profile type, filesystem entries, workspace roots, and
non-escalatable deny entries.
- Persist effective `workspace_roots` in `TurnContextItem` so
resumed/replayed context does not have to bind `:workspace_roots`
through legacy `cwd` fallback.
- Add explicit permission instructions that denied reads are policy
restrictions, not escalation targets.
- Pass the parent turn's denied-read context into automatic approval
reviews.
- Add targeted coverage for prompt rendering, workspace-root
materialization, replay context, and review prompt context.
- Keep the prompt-context test expectations platform-aware so the same
filesystem rendering assertions pass on Unix and Windows paths.
## Testing
- `just test -p codex-core
context::environment_context::tests::serialize_environment_context_with_full_filesystem_profile`
- `just test -p codex-core
context::environment_context::tests::turn_context_item_filesystem_uses_workspace_roots_instead_of_cwd`
- `just test -p codex-core
context::permissions_instructions::permissions_instructions_tests::builds_permissions_from_profile_with_denied_reads`
- `just fix -p codex-core`
I also attempted `just test -p codex-core`; the changed prompt-context
tests passed, but the full local run did not complete cleanly in this
sandboxed macOS environment due unrelated user-shell `CODEX_SANDBOX*`
expectations and integration-test timeouts.
## Summary
Adds an optional `clientId` field to app-server v2 `UserInput` and
carries it through the core `UserInput` model so clients can correlate
echoed user input items without relying on payload equality.
## Details
- Adds `client_id: Option<String>` to core `UserInput` variants.
- Exposes the v2 app-server field as `clientId` on the wire and in
generated TypeScript.
- Preserves the id when converting between app-server v2 and core
protocol types.
- Regenerates app-server schema fixtures.
## Validation
- `just fmt`
- `just write-app-server-schema`
- `cargo test -p codex-app-server-protocol`
- `cargo test -p codex-protocol`
- `just fix -p codex-app-server-protocol`
- `just fix -p codex-protocol`
- `git diff --check`
## Why
When Guardian or the sandbox network proxy detects and denies a network
attempt, core cancels the associated execution through `ExecExpiration`.
The Windows sandbox capture path was only forwarding the timeout
component of that expiration state. As a result, a sandboxed Windows
command whose network attempt had already been denied could keep running
until its timeout elapsed rather than terminating promptly in response
to the denial.
This change closes that cancellation-propagation gap for Windows sandbox
execution.
## What changed
- Added `WindowsSandboxCancellationToken` as the cancellation hook
exposed to Windows capture backends.
- Extracted the cancellation token from `ExecExpiration` in core and
passed it to both the direct and elevated Windows sandbox capture paths
alongside the existing timeout.
- Updated direct capture to poll for either process exit, timeout, or
cancellation and to terminate cancelled processes without reporting them
as timed out.
- Updated elevated capture to watch for cancellation and send the
existing `Terminate` IPC frame to the elevated runner. The watcher parks
for 50 ms between checks to bound response latency without a tight busy
wait.
- Added Windows regression coverage for a long-running PowerShell
command: cancellation ends capture before its timeout and does not set
`timed_out`.
- Added a visible skip diagnostic when that PowerShell-dependent
regression test cannot execute, and consolidated the duplicated
expiration-policy branch identified in review.
## Security
This improves enforcement after a denied network attempt has been
attributed to a Windows sandboxed execution: the command no longer
remains alive simply because Windows capture lost the cancellation
signal.
This PR does not claim to make Windows offline mode an airtight
no-network or no-exfiltration boundary. It does not introduce
AppContainer or change how network denial is detected; it makes an
already-detected denial promptly stop the affected sandboxed command.
## Validation
### Commands run
- `just fmt`
- `cargo test -p codex-windows-sandbox`
- `cargo test -p codex-core network_denial`
- `cargo clippy -p codex-core -p codex-windows-sandbox --tests --no-deps
-- -D warnings`
- `just argument-comment-lint -p codex-windows-sandbox -p codex-core`
The new capture regression is `cfg(target_os = "windows")`, so Windows
CI is the execution coverage for that test path. The local macOS test
runs validate the host-runnable crate and core network-denial behavior.
---------
Co-authored-by: Codex <noreply@openai.com>
## Why
#23756 makes packaged Codex builds include and default to the bundled
zsh fork. The important reason to put that fork's directory at the front
of `PATH` is to keep executable-level escalation working after a command
leaves the original shell and later re-enters zsh through `env`.
The expected chain is:
1. The zsh fork runs the top-level shell command.
2. That command launches another program, such as `python3`, while
inheriting the `EXEC_WRAPPER` environment and the escalation socket fd.
3. That program spawns a shell script whose shebang is `#!/usr/bin/env
zsh` rather than `#!/bin/zsh`, and it does not close the escalation fd.
4. `/usr/bin/env` resolves `zsh` through `PATH`, so it must find the
packaged zsh fork before the system zsh.
5. Commands inside that nested script are intercepted by the zsh fork
and can still request escalation from Codex.
If `PATH` resolves `zsh` to the system shell instead, the nested script
loses zsh-fork exec interception. Commands that should request
escalation can then run only in the original sandbox, or fail there,
without Codex ever receiving the approval request.
Shell snapshots make this slightly more subtle: a snapshot can restore
an older `PATH` after the child shell starts. This PR treats the zsh
fork `PATH` prepend as an explicit environment override so snapshot
wrapping preserves it.
## What Changed
- Added shared zsh-fork runtime helpers that prepend the configured zsh
executable parent directory to `PATH` without duplicate entries.
- Applied the zsh fork `PATH` prepend to both zsh-fork `shell_command`
launches and unified-exec zsh-fork launches before sandbox command
construction.
- Kept the shell-command zsh-fork backend API narrow: it derives the
configured zsh path from session services and rebuilds its sandbox
environment from `req.env`, rather than accepting a second, competing
environment map or a separately threaded bin dir.
- Kept Unix-only zsh-fork `PATH` mutation out of Windows clippy-visible
mutability.
- Added coverage for duplicate `PATH` entries, for preserving the zsh
fork prepend through shell snapshot wrapping, and for the nested
`python3` -> `#!/usr/bin/env zsh` escalation flow.
## Testing
- `just fmt`
- `just fix -p codex-core`
I left final test validation to CI after the latest review-comment
cleanup. Before that cleanup, `just test -p codex-core zsh_fork` passed
locally for the zsh-fork-focused tests.
## Why
Add a standalone image generation path that can be exercised
independently of hosted Responses image generation, while retaining the
hosted tool as fallback unless the extension is actually available to
the model.
## What changed
- Added the `codex-image-generation-extension` crate with standalone
generate/edit execution, prior-image selection for edits, model-visible
image output, and local generated-image persistence.
- Installed the extension in app-server behind the disabled-by-default
`imagegenext` feature and backend eligibility checks.
- Updated core tool planning so eligible `image_gen.imagegen` exposure
replaces hosted `image_generation`, while unavailable configurations
retain hosted fallback.
- Added coverage for extension behavior, edit history reuse, feature
gating, auth eligibility, and hosted-tool replacement.
- The extension is installed through app-server only in this PR; other
execution paths retain hosted image generation because hosted
replacement occurs only when the standalone executor is actually
registered and model-visible.
- The initial extension contract intentionally fixes the image model to
`gpt-image-2` and uses automatic image parameters.
- Native generated-image history/card parity and rollout persistence
cleanup are intentionally deferred follow-up work.
## Validation
- `just test -p codex-image-generation-extension`
- `just test -p codex-features`
- `just test -p codex-core
hosted_tools_follow_provider_auth_model_and_config_gates`
- `just test -p codex-app-server`
- `just fix -p codex-image-generation-extension -p codex-features -p
codex-core -p codex-app-server`
- `just fmt`
- `just bazel-lock-update`
- `just bazel-lock-check`
---------
Co-authored-by: jif-oai <jif@openai.com>
## Why
#24744 introduced the thread idle lifecycle hook so idle continuation
can be owned by lifecycle contributors instead of hard-coded goal
runtime plumbing. Task completion still called
`goal_runtime_apply(GoalRuntimeEvent::MaybeContinueIfIdle)` directly, so
the post-turn idle transition remained goal-specific and did not notify
generic thread lifecycle contributors.
## What Changed
- Add `Session::emit_thread_idle_lifecycle_if_idle()` to gate idle
emission on both no active turn and no queued trigger-turn mailbox work.
- Call that helper when a task clears the active turn, replacing the
direct `GoalRuntimeEvent::MaybeContinueIfIdle` path.
- Cover the behavior with `codex-core` session tests for emitting after
task completion and suppressing idle emission while trigger-turn mailbox
work is pending.
## Verification
- New tests in `core/src/session/tests.rs` exercise the idle lifecycle
emission and trigger-turn mailbox guard.
## Why
PR #24813 added extension `TurnItemEmitter` coverage and introduced a
test that records a conversation history item before asserting
extension-emitted turn item events.
`record_conversation_items()` also emits a `RawResponseItem` event to
observers. The test was reading from the same event receiver and
expected the next event to be `ItemStarted`, so the test failed reliably
once the setup history item was present.
## What Changed
Update
`passes_turn_fields_and_scoped_turn_item_emitter_to_extension_call` to
consume and assert the expected setup `RawResponseItem` before checking
the extension `ItemStarted`, `WebSearchBegin`, `ItemCompleted`, and
`WebSearchEnd` events.
This is test-only and does not change extension runtime behavior.
## Verification
- `cargo nextest run --no-fail-fast -p codex-core
tools::handlers::extension_tools::tests::passes_turn_fields_and_scoped_turn_item_emitter_to_extension_call`
## Summary
- Let `close_agent` clean up an agent that is still registered in
`AgentRegistry` even when its underlying thread is already missing.
- Preserve the explicit-close boundary: for known stale thread-spawn
agents, mark the persisted spawn edge `Closed`, then treat
`ThreadNotFound` / `InternalAgentDied` as a successful close so the
registry slot can be released.
- Add a regression for MultiAgentV2 task-name targets where
`close_agent("worker")` succeeds after the worker thread has already
disappeared.
## Motivation
A worker can disappear from `ThreadManager` while its metadata still
exists in the root `AgentRegistry`. Before this change, the close tool
failed while trying to subscribe to the missing thread status, so it
never reached the cleanup path that releases the registered agent slot.
With `agents.max_threads = 1`, an explicit close of that stale task-name
agent could fail and leave the session unable to spawn a replacement.
## Scope
This PR intentionally does not add automatic stale-agent reaping to
`spawn_agent`, `resume_agent`, or `list_agents`. A thread being missing
from `ThreadManager` is not the same as an explicit close: persisted
open spawn edges are still the durable source of truth for resume and
task-name ownership until `close_agent` is called.
## Validation
- `just test -p codex-core -E
'test(multi_agent_v2_close_agent_reaps_stale_task_name_target) |
test(resume_agent_from_rollout_reopens_open_descendants_after_manager_shutdown)'`
- `just fix -p codex-core`
## Why
Extension-contributed tools need to emit visible turn items through
Codex's normal event and persistence pipeline.
## What
- Add `TurnItemEmitter` to extension `ToolCall`s and route the core
implementation through `Session::emit_turn_item_*`.
- Hold weak session and turn references so retained tool calls cannot
keep host state alive.
- Provide a no-op emitter for extension test callers.
## Test Plan
- `just test -p codex-core -E
'test(passes_turn_fields_and_scoped_turn_item_emitter_to_extension_call)'`
---------
Co-authored-by: jif-oai <jif@openai.com>
Summary
- Add TurnErrorInput and TurnLifecycleContributor::on_turn_error to the
extension API.
- Emit the turn-error lifecycle from core turn error paths, including
usage limit failures.
- Add direct lifecycle coverage for the emitted error facts and stores.
Tests
- just fmt
- git diff --check
- Not run: full tests or clippy (per instructions)
Summary: add session source and persistent-state availability to
ThreadStartInput; populate them from session init; update existing goal
test harness constructors. Tests: just fmt; git diff --check. No full
tests or clippy run per request.
## Why
Guardian reviews already emit analytics events, but we do not expose
aggregate OpenTelemetry metrics for review volume, latency, token usage,
or terminal outcomes. That makes it harder to monitor Guardian behavior
during rollouts and to compare review outcomes by source, action type,
session kind, model, and failure mode.
## What Changed
- Added Guardian review metric names for count, total duration, time to
first token, and token usage in `codex-rs/otel`.
- Added `core/src/guardian/metrics.rs` to convert
`GuardianReviewAnalyticsResult` into sanitized metric tags covering
decision, terminal status, failure reason, approval request source,
reviewed action, session kind, risk/outcome, model, reasoning effort,
and context/truncation state.
- Emitted the new metrics from `track_guardian_review` for each terminal
Guardian review result.
## Testing
- Added
`guardian_review_metrics_record_counts_durations_and_token_usage`, which
verifies the emitted count, duration, TTFT, token usage histograms, and
tag set through the in-memory metrics exporter.
## Summary
- Add the required `/*parent_thread_id*/` argument comment at the
Guardian review session test callsite flagged by CI.
## Validation
- `just fmt`
- Not run: clippy/tests, per request; CI will cover them.
Split from the Guardian prompt cache key change. This PR only updates
codex-rs/core/src/session/session.rs. Validation was not run per
request; this branch is expected to rely on the companion split PRs.
Split from the Guardian prompt cache key change. This PR only updates
codex-rs/core/src/guardian/tests.rs. Validation was not run per request;
this branch is expected to rely on the companion split PRs.
