Commit Graph
6099 Commits
Author SHA1 Message Date
Alex Kira 2ca71f1683 Record additional context counts, sizes, and truncation metrics (#58144)
## What changed

Add `codex.additional_context` to count submitted entries, including unchanged entries, and `codex.additional_context.value_bytes` to measure UTF-8 value sizes before and after the existing truncation limit. Sizes exclude message wrappers and include the truncation marker after truncation.

Tag metrics with context kind, truncation status, and an explicitly supplied name when present; never use opaque source keys as tags. Record metrics through the current turn's telemetry.

## Testing

Add integration coverage for message and tool-output formats, repeated submissions, truncation boundaries, empty and multibyte values, optional name tags, and turn-specific model attribution.

GitOrigin-RevId: 463f257af27be58c372970fac853b85fde495dd5
2026-10-29 18:24:34 +00:00
radai-oai 875a1a78f0 Persist subagent relationships before creating child storage (#58129)
## Why

Creating child history before persisting its parent-child relationship can leave history that parent subtree deletion cannot discover, especially when fork startup copies history. A relationship write failure must reject the spawn before child storage is created.

## What changed

- Reserve the child thread ID and persist its spawn edge before starting the child. Propagate persistence failures to the caller.
- Keep pending relationship writes owned through caller cancellation, and wait for them to finish before closing the edge during cleanup, including when a write commits but reports an error.
- Skip edge cleanup when publication was skipped, and reject IDs belonging to running threads before publication.

## Testing

Add regression coverage for cancellation before commit or acknowledgement, errors after commit, skipped publication, and duplicate child IDs. Integration tests cover fresh children and full-history forks, verifying that failed publication leaves no child history, execution, or creation notification and preserves unrelated history during parent deletion.

GitOrigin-RevId: 6e2e3f32d9558091312651ab47aa845d7d13b4ca
2026-10-29 17:25:51 +00:00
Sean Huang 09db085f94 Support negotiated server-side truncation in Code Mode clients (#58125)
## Why

Host-side truncation can bound Code Mode output before it crosses the transport, including script errors that would otherwise exceed the frame limit.

## What changed

- Negotiate server-side output truncation for stdio and gRPC sessions, with client-side fallback for older hosts.
- Forward output budgets for execution, waiting, and termination when server-side truncation is selected. Avoid truncating host-formatted output or appending script errors a second time.
- Preserve the negotiated mode across reconnects and reject hosts that cannot maintain it.

## Testing

Add transport and integration coverage for output budgets, oversized script errors, reconnect compatibility, and preservation of output-limit notices with a zero-token budget.

GitOrigin-RevId: 2075101e4845eca64f45491c27a3b326788645c8
2026-10-29 16:57:12 +00:00
Alex Kira b05c1a457e Add an incremental exec tools feature flag (#58122)
Register `incremental_exec_tools` as an under-development feature, disabled
by default, and expose it in the configuration schema.

GitOrigin-RevId: 444767b7a46e3ada7efb3f6c4fb570feb299e0c7
2026-10-29 16:41:12 +00:00
sgarg-oai 53d3d7f64e Centralize provider request preparation auth checks (#58121)
## What changed

Move the model client's request-preparation loop into the shared `prepare_provider_request` helper and add `RequestAuthChange` to classify auth state changes. For managed auth, retry preparation when credentials change and reject account changes, including when the provider starts or stops exposing an auth manager during setup.

Capture the auth owner and credential generations together and pass them into the preparation callback so the resulting client setup records the state used for that attempt.

## Testing

Add a test verifying that an explicit provider bearer token survives a workspace switch during setup without retrying preparation.

GitOrigin-RevId: ed8710d1c344b48ebf6182e25e6269214dd2d3fe
2026-10-29 16:39:38 +00:00
jif 9def09afa7 Use serving assessments for adaptive Guardian review (#58075)
## What changed

- Match final serving assessments to their response and output item, then retain them with tool invocations and code cells across waits.
- Use matching assessments instead of asynchronous scoring for adaptive review: low-risk actions can skip synchronous review, while elevated-risk actions require it.
- Revalidate captured authorization and policy before approval. Require fresh review for stale or incomplete authorization, preserve explicit review requirements, and discard assessments when hooks rewrite tool input.
- Keep asynchronous scoring as the fallback when an assessment is missing or does not match.

## Testing

Add integration coverage for direct and nested tool calls, hook rewrites, explicit escalation, interrupted responses, catalog-driven policy, incomplete authorization, and assessment retention across code-cell yields and waits.

GitOrigin-RevId: 4346a227204137f018c03c34b302d6f1e9592c96
2026-10-29 12:54:39 +00:00
ioanbudea-OAI 6b3b3349f3 Share the Guardian context delivery token limit (#58071)
Expose `MAX_TEXT_PART_ESTIMATED_TOKENS` from `codex-guardian-context` and use it for native tool calls, transcript entries, and delivery-size assertions. Keep the existing 9,000-token limit while replacing duplicated limits in code and tests.

GitOrigin-RevId: 90cb98eea9bade34f18db0d1666ed7ed1e93febe
2026-10-29 12:37:59 +00:00
jif 79be2e201f Initialize native_call in agent host test fixtures (#58057)
Set `native_call` to `None` in the `ToolCall` and `ToolInvocation` fixtures used by the agent host invocation and spawn tests.

GitOrigin-RevId: 2a6f9ccb95d99297085a2861bb44e5ce7599715c
2026-10-29 12:06:40 +00:00
jif 9cc14e7988 Add managed command checkpoints requiring fresh human approval (#58045)
## Why

Managed command policies need to require a human decision independently of automatic review, hook allowances, and saved approvals.

## What changed

- Support `decision = "prompt-user"` in managed `requirements.toml` prefix rules, requiring a fresh, one-time human decision for matching commands.
- Add experimental `humanCommand` approval requests for clients advertising both `experimentalApi` and `requiredHumanApproval`. Accept only `accept`, `decline`, or `cancel`; reject commands when no capable client is available.
- Restrict responses to capable clients that received the request, and replay pending prompts to compatible clients on reconnect.
- Enable the capability in the TUI and update protocol schemas and client bindings.

## Testing

Add coverage for fresh decisions overriding saved allowances, hook allowances, and automatic review, plus child-thread approval routing and rejection of responses from incompatible or unrelated clients.

GitOrigin-RevId: 0b95ff6aab7f8a04b7298c8a4329d1fddf6a603e
2026-10-29 11:23:51 +00:00
ioanbudea-OAI 058e99a3aa Preserve native tool-call evidence through Guardian review (#58040)
## Why

Guardian review needs the original function-call metadata even when the call is no longer in live conversation history. Looking it up in history can lose the evidence needed to review encrypted multi-agent messages.

## What changed

Capture native-review policy in the tool registry and retain the original call through dispatch for multi-agent tools that require review. Use that retained call for Guardian review, applying hook-rewritten arguments while preserving encrypted-argument metadata and clearing stale replay content. Limit `ToolCall` debug output to the tool name and call ID.

## Testing

Add coverage for selective call retention across review settings and namespaces. Extend Guardian tests to cover calls absent from live history and hook rewrites that preserve call identity and encrypted-argument metadata across incremental reviews.

GitOrigin-RevId: 0e49833741c09c29228440bddd2e12695ae99583
2026-10-29 11:15:52 +00:00
jif 588a277b30 Add an option to disable successful SSE delta logs (#58034)
## What changed

Add `otel.log_sse_deltas`, enabled by default. Setting it to `false` suppresses successful SSE event logs whose kind ends in `.delta` while preserving errors, completion events, and metrics.

Apply the setting to session and memory telemetry, support managed requirements, and expose it in the app-server configuration requirements API and generated schemas and bindings.

## Testing

Add coverage for delta log suppression with errors, completion logs, event counts, and duration metrics preserved. Extend configuration and app-server tests to cover defaults, overrides, and managed requirements.

GitOrigin-RevId: 751c7f357c41f51848b025c838246a733eb92e2f
2026-10-29 10:34:17 +00:00
swastik-oai 3535c22b43 Add direct Guardian review of recorded exec calls (#58028)
## What changed

Add `CodexThread::guardian_review` to assess the final injected `exec_command` call against recorded history without executing the command. Reuse command preparation and approval policy, returning the call ID and an optional assessment when approval is required.

Reject active threads, unsupported calls, commands intercepted by `apply_patch`, and histories containing `request_permissions`. Honor cancellation and the configured review timeout.

## Testing

Add a scenario test and request snapshot verifying that direct review matches live Guardian input, leaves commands unexecuted, skips assessment for explicitly allowed commands, and rejects trailing tool output and recorded permission grants.

GitOrigin-RevId: 704f8dc77bbb991c231fce03cfda1c238c4e0b02
2026-10-29 09:48:55 +00:00
Anthony Tafoya 2e764743a6 Apply session plugin overrides during MCP config refresh (#57974)
## What changed

Allow `refresh_mcp_config` to replace explicit session `plugins` overrides while preserving other session overrides and user settings. Omitting session plugins retains the current overrides; supplying an empty map clears them and restores inheritance from lower configuration layers.

## Testing

Add coverage for replacing and clearing plugin overrides, preserving unrelated settings, rejecting stale refreshes, revoking and restoring plugin tools and hooks, and retaining refreshed plugin policy when a deferred executor becomes ready.

GitOrigin-RevId: 26913fac0f8ae9590b39bdc037a78c9be86abb67
2026-10-29 03:14:13 +00:00
Joe Savona 797bab4746 Add an experimental UI tree renderer for the status line (#57970)
## What changed

- Add the disabled-by-default `tui_extensions_statusline` feature to render configured status items and their picker preview through a shared UI tree renderer.
- Support text, HTTP(S) links, rows, spacers, dividers, and gaps with rendering budgets, text sanitization, and width-aware truncation that keeps the ellipsis outside hyperlinks.
- Use plain text for the extension status line and hide the picker's theme color toggle while preserving the saved color preference.

## Testing

Add coverage for tree serialization and URL validation, Unicode widths, hyperlink clipping, picker previews, color preference preservation, and token usage updates across renderer switches.

GitOrigin-RevId: a6b87c2b1096765d61296d73e142834fcd53ceee
2026-10-29 03:07:15 +00:00
pansul47-oai 4e2dfd2230 Retry transient environment startup failures on the next turn (#57866)
## Why

A transient environment startup failure persisted across turns when the selected environment and workspace were unchanged, preventing commands from recovering after the executor became available.

## What changed

Retry failed environment resolution at the start of a new turn when the error is retryable and the workspace remains selected. Reuse healthy or pending resolutions and preserve provisioning failures until the selection is updated.

## Testing

Add coverage for command execution recovering on an unchanged next turn after endpoint availability is restored. Extend provisioning failure tests to verify that subsequent turns receive the original error without restarting resolution.

GitOrigin-RevId: b80af60990f30d277804160c9c39c286024c7667
2026-10-29 00:42:21 +00:00
liame-oai f276b82185 Allow upstream interruption when canceling Responses Lite turns (#57864)
## What changed

- Add the disabled-by-default `interrupt_on_cancel` feature to interrupt upstream inference when a Responses Lite turn is canceled, with a one-second cleanup deadline.
- Detach canceled WebSocket responses so replacement requests can open a new connection without waiting for cleanup.
- Add an optional `interrupt_on_cancel` override for thread samples, inheriting the parent setting when omitted.

## Testing

Add coverage for cancellation during inference and steering, replacement turns proceeding before cleanup finishes, and prompt cancellation of model response interceptors.

GitOrigin-RevId: 501368746fc04fd3fdbd45c1f34c75ce57f4e955
2026-10-29 00:03:28 +00:00
alexsong-oai f91edce53b Add Google Cloud Vertex AI setup to TUI onboarding (#57857)
## What changed

- Add a sign-in wizard behind `features.gcp_setup_wizard`, disabled by default, to collect a project ID, optional location, and optional credential file. Users can leave the credential file blank to use Application Default Credentials.
- Delegate credential validation and configuration persistence to the app-server, showing progress and allowing retries after errors.
- Share editable single-line fields with Bedrock setup and compact the sign-in menu when terminal height is limited.

## Testing

Add onboarding tests and snapshots covering feature gating, field editing, navigation, validation failures, successful configuration persistence, and compact menu rendering.

GitOrigin-RevId: 96eb31e847aa1c2d503176be68ff2e2efe4722fa
2026-10-28 23:06:34 +00:00
aakash-oai d373d64967 Add optional process metrics sampling for code-mode hosts (#57841)
## What changed

- Add `GetProcessMetrics` and provider sampling APIs for process ID, start time, cumulative CPU time, and resident memory, exposed through `ThreadManager` without creating a thread or session.
- Distinguish unsupported collection, a locally absent host, and sampling failures. Process-owned providers never start or replace a host when sampling.
- Bound gRPC sampling to one second without retries or explicit connection resets. Run optional collectors on a detached worker with at most one collection in flight, so stalled collection cannot block the HTTP/2 driver or runtime shutdown.

## Testing

Add coverage for snapshots, collector errors and panics, deadlines, overload, and shutdown with a stuck collector. Extend lifecycle tests to check that sampling preserves session state and keepalive behavior without starting or restarting local hosts.

GitOrigin-RevId: eeafe275891d0464636f28a29345a87aed1a12dc
2026-10-28 21:34:25 +00:00
Abhinav 7ce5fa7bdc Validate thread providers before resume and input (#57840)
## Why

Existing chats retain their provider configuration when the selected provider changes. Reject resume and input until the configured provider matches the chat, while keeping history readable and allowing recovery after configuration is restored.

## What changed

- Validate saved providers before applying resume overrides, preserving the provider when only model or reasoning settings change.
- Check the current provider selection, including profiles and launch overrides, before accepting input. Reject input when configuration cannot be loaded.
- Return structured `ModelProviderError` data for provider mismatches, missing definitions, and policy conflicts, with generated JSON Schema, TypeScript, and Python types.

## Testing

Add coverage for override bypass prevention, provider restoration, readable history after rejection, and blocked requests sending no inference traffic. Verify schema fields and Python round trips match the error wire format.

GitOrigin-RevId: 051f97c3815c547d8411c71b770df30eae969f68
2026-10-28 21:33:02 +00:00
andrewgu-oai c62e9df39f Preserve reasoning token accounting across resume and fork (#57839)
## Why

Resuming or forking a thread loses the session flag indicating that reported token usage already includes retained reasoning. Counting that reasoning again can trigger automatic compaction too early.

## What changed

Persist `last_usage_includes_reasoning` with `TokenUsageInfo` and use it when estimating context usage. Record the flag with completed sampling and compaction usage, and mark local estimates as including reasoning. Older records without the field retain the existing accounting behavior.

## Testing

Add coverage for saved inclusive, exclusive, and legacy token counts. Extend the encrypted-reasoning compaction test to verify the compaction threshold for live, forked, and cold-resumed threads.

GitOrigin-RevId: 9f22262a693111ec0406dce11804eb4844ad8983
2026-10-28 21:22:06 +00:00
Krish Chainani 55503b23d4 Limit concurrent image availability checks to eight (#57835)
Replace `join_all` with a buffered stream in `replace_unresolvable_image_files` so at most eight distinct image file IDs are checked concurrently. Collect unavailable IDs as checks complete before replacing their image content with placeholders.

GitOrigin-RevId: 2a9ed8742fbb6157c99c69a15f212df3ed88810c
2026-10-28 21:08:10 +00:00
swastik-oai 3da7de59b6 Add isolated Guardian assessments for prepared actions (#57834)
## What changed

Add `ReviewInput` to bundle an approval request with its captured parent history and review context. Its `assess_isolated` method runs a fresh reviewer without publishing a live approval, honors deadlines and cancellation, and returns a `GuardianAssessment` or review error.

Share review parameter construction with the live review path and extract assessment parsing and error conversion into `GuardianReviewSessionOutcome::into_assessment`.

## Testing

Add a mock-backed test that verifies the isolated assessment is returned and the reviewer receives the captured history without instructions added after capture.

GitOrigin-RevId: 10dd5fddf5fc58bfab124f5962a6b7f45d435c83
2026-10-28 21:07:36 +00:00
Jiahao ef186f812a Detach realtime sideband observers without ending the session (#57817)
## Why

Stopping a sideband observer should close its WebSocket while preserving the frontend-owned realtime session. The existing frameless session close also sends `session.close`, so observer shutdown needs a transport-only close.

## What changed

- Add `close_transport()` to send a normal WebSocket close without `session.close`.
- Close the sideband transport on cancellation or when the call becomes inactive, including an existing connection awaiting adoption by the input task. Bound the close attempt to five seconds.
- Thread an optional sideband base URL through realtime start processing and allow nested thread ID fields in request serialization scopes.

## Testing

Add WebSocket and core conversation regression tests covering observer detachment for both realtime parsers and conversation versions. Verify that explicit frameless session close still sends `session.close`.

GitOrigin-RevId: 874a8980b0fcd45058e976b00c0d968f5b4b2404
2026-10-28 20:10:15 +00:00
ioanbudea-OAI c6fce1cace Centralize synchronous Guardian prompt preparation (#57790)
## What changed

- Separate evidence collection from prompt preparation and pass reviewer history, model limits, and image preparation policy through shared `guardian-context` APIs.
- Detect captured images before composition and measure reviewer history tokens only when images and a valid context window require it.
- Share instruction and native-message delivery filtering across budget checks, including restoration of retained evidence after history changes.

GitOrigin-RevId: 65ef53cf0bc22e1bde9abfb7a9ffce65cfab6f1c
2026-10-28 17:13:33 +00:00
jif 84bef6ea8c Refresh the model catalog once when Guardian's default reviewer is missing (#57789)
## Why

When `codex-auto-review` is missing from the model catalog, Guardian falls back to the parent model. Refresh the catalog before accepting that fallback so reviews and checkpoint compatibility checks can use the default reviewer when it becomes available.

## What changed

- Share one refresh attempt per parent session, preserving it across cancelled waiters and retaining the parent fallback if recovery fails.
- Respect explicit reviewer overrides and custom catalogs, and skip recovery for basic sessions.
- Resolve reviewers from the in-memory catalog so a failed disk cache write does not discard a recovered reviewer.
- Apply review deadlines and cancellation to configuration setup, and avoid reviewer resolution during history replay unless checkpoint compatibility requires it.

## Testing

Add coverage for recovery, failed fetches and cache writes, overrides and custom catalogs, checkpoint compatibility, cancelled prewarming, review deadlines, and interruption.

GitOrigin-RevId: 87ae285713afde518a12bc2daba006a043b13bc5
2026-10-28 17:04:31 +00:00
Charlie Marsh 575cc5351e Fix app-server thread listener startup and shutdown races (#57787)
## Why

A thread runtime can shut down while listener setup is awaiting I/O, and a cold resume can start before the previous listener forwards `ShutdownComplete`. Listener replacement must preserve those buffered terminal events.

## What changed

- Coordinate listener installation and connection subscriptions with pending unloads, and reject closing or replaced runtimes.
- Allow a newly created runtime that has already shut down to drain its queued events, while enforcing one event reader per runtime.
- Track shutdown draining with a shared watch channel so cold resume and runtime replacement wait for terminal events to be processed.
- Serialize idle unloading with goal edits and cold resume, and recheck thread state after acquiring the lock.

GitOrigin-RevId: d05e348b0a81bddb2defa963d0cfd3fde08dbd03
2026-10-28 16:53:55 +00:00
aakash-oai 5fdb50af95 Restrict code-mode host replacement to session admission (#57782)
## Why

Automatic gRPC reconnection could spawn a replacement host while dispatching an RPC for an existing session. Host loss must remain visible so recovery can report lost state without replaying execution or completed external effects.

## What changed

- Bind each local host channel to one child process and allow only `OpenSession` to replace a dead host.
- Add caller-owned transport support to `GrpcCodeModeSessionProvider` and map local transport failures to `Unavailable` so initial connection failures use the existing bounded admission retries.

## Testing

Add regression coverage for child exit during initial connection and between channel selection and dispatch, concurrent session recovery, and child reaping after teardown. An integration test kills the host after a command completes and verifies that recovery reports the reset, clears stored state, and leaves the command's external effect unchanged.

Increase the remote MCP stalled-handshake test's initialization timeout and mock response delay to let requests reach the server on loaded CI workers.

GitOrigin-RevId: 4ffd810e4cfcfb7a56e606816aa867bc206a09b6
2026-10-28 16:46:41 +00:00
ioanbudea-OAI 5c0e42161e Respect model catalog origins when selecting Guardian reviewers (#57770)
## Why

A bundled reviewer entry does not establish that a custom gateway offers that model. Guardian should fall back to the parent model when the reviewer is neither advertised by the provider nor explicitly configured.

## What changed

Track model metadata origins as `Bundled`, `Provider`, or `ExplicitConfig`, including across catalog merges and cache reloads. Exclude bundled entries from Guardian reviewer selection unless `auto_review_model_override` is set. Continue selecting reviewers from provider and explicitly configured catalogs.

## Testing

Add coverage for gateway reviewer selection with each catalog origin, plus origin tracking across cache reloads, empty provider responses, and authentication changes.

GitOrigin-RevId: f8bf865d406e56ee0ade90087b55f2e33b98181e
2026-10-28 16:00:30 +00:00
jif 7c11b8074d Include terminal launch commands in Guardian stdin reviews (#57758)
## Why

Guardian's `write_stdin` approval requests lacked the command used to launch the receiving terminal, leaving reviews without that context.

## What changed

- Carry `original_command` from the terminal's launch command into the approval request and Guardian action JSON. This describes the original launch, not the terminal's current state.
- Exclude `original_command` from the 8,000-byte stdin approval size check so launch context does not consume the budget for input, permission details, and the approval reason. Launch context remains subject to Guardian's normal request budget.

## Testing

Extend serialization coverage to check `original_command` and the strict stdin review integration test to verify that a launch command longer than 8,000 bytes reaches Guardian intact.

GitOrigin-RevId: cc765a6bf80d080baf94c959eafa6227b2889e1f
2026-10-28 15:10:52 +00:00
jif fc916a22b5 Expose Guardian assessments in response streams (#57757)
## What changed

Decode `response.metadata` Guardian assessments into typed `GuardianAssessment` events over SSE and WebSocket, carrying `response_id`, `item_id`, and `requires_review`. Ignore assessments with missing, empty, or incorrectly typed required fields while preserving independent delivery of other metadata.

Expose the assessment type to request-scoped extension interceptors. Keep assessments out of conversation history; they do not authorize tool execution.

## Testing

Add SSE coverage for assessment ordering, malformed metadata, and mixed metadata delivery. Extend model request tests to cover SSE and WebSocket interceptor delivery and exclusion from subsequent model input.

GitOrigin-RevId: 282b8817ee97346d76b40bcb412eef023df60ff1
2026-10-28 15:04:29 +00:00
jif ce2d47bdc4 Move Guardian MCP elicitation helpers into the reviewer crate (#57743)
## What changed

Extract MCP approval elicitation parsing and review-decision response conversion from `core/src/session/mcp.rs` into `codex-guardian-reviewer`. Export the helpers and `GuardianElicitationReview` for the session code to use, while keeping session lookup, provenance checks, and approval enforcement in the host.

## Testing

Move the existing elicitation tests into the reviewer crate, retaining coverage for request construction, opt-in requirements, unsupported forms, decision mapping, and model-specific timeout instructions.

GitOrigin-RevId: fb745fa0e49a29f015bbda16c14b5fc2b12278b9
2026-10-28 13:56:17 +00:00
jif 6c839b3694 Separate test-only Rust sources from Bazel library and binary inputs (#57741)
## What changed

Add `unit_test_srcs` to `codex_rust_crate` to exclude test-only modules from library and binary sources while making them available to unit tests through `compile_data`.

Configure `codex-core` and `codex-tui` to use this option for their test modules and support files.

GitOrigin-RevId: 437b5ddb75e66a96f3c5cb9b309cb0e2608ba169
2026-10-28 13:53:43 +00:00
ioanbudea-OAI 5fd36fc550 Remove free_guardian configuration from Guardian tests (#57729)
Remove `[features.guardianv2].free_guardian` setup and test parameters
from the Guardian review suite, along with redundant opt-in cases.
Retain coverage for authentication-dependent requests, HTTP fallback,
and session prewarming and reuse without explicitly setting the flag.

GitOrigin-RevId: ba1f985ae2f7f0e0b0c8b32cb47ac9993cbb43f2
2026-10-28 13:16:05 +00:00
jif 5d5e5c507d Move ReviewAction into the Guardian reviewer crate (#57720)
## What changed

Export `ReviewAction` and its conversion from `GuardianApprovalRequest` from
`codex-guardian-reviewer`. Keep host-specific approval preparation and environment
validation in core as standalone functions, and update callers accordingly.

## Testing

Move the `write_stdin` serialization and tool-call provenance test into the
reviewer crate, preserving coverage for input, Windows working directories, and
sandbox permission modes.

GitOrigin-RevId: 9a1caf9bbb67ba017e4950a2da9ef2543b013007
2026-10-28 12:31:24 +00:00
jif e742db9f7b Preserve parent routing affinity for Guardian reviews (#57710)
## Why

Shared routing cookies can change after a parent connection is established. Guardian reviews need the routing affinity captured with their parent response so unrelated traffic does not replace that hint.

## What changed

- Capture `__oailb` and `__cflb` cookies from successful ChatGPT WebSocket handshakes and pair them with the parent response ID.
- Retain the snapshot throughout a review and its retries, and apply it only to Guardian requests with the matching parent response. Reconnect cached reviewer sockets when the parent affinity changes.
- Preserve cookie expiry and restrict reuse to the original secure origin and endpoint path. Replace routing cookies while retaining unrelated cookies, and keep the captured affinity out of prompts, history, and serialized metadata.

## Testing

Add coverage for cookie scope, expiry, duplicate rejection, and merging; replaying a captured pin after shared cookies change or are deleted; and retaining the matching parent snapshot through compaction without exposing cookies in review prompts or metadata.

GitOrigin-RevId: d180679e6265b612a3617f20ca1f5dec89e73aa8
2026-10-28 11:48:00 +00:00
jif e6de6a2423 Move Guardian review evidence into the reviewer extension (#57702)
## What changed

Move review evidence storage, authorization snapshots, sender exchange selection,
and bounded Node REPL evidence into `codex-guardian-reviewer`. Update core to
call or re-export these APIs while retaining host-specific context rendering and
capture-mode selection.

Move the existing Node REPL evidence and text truncation tests alongside their
implementations in the reviewer crate.

GitOrigin-RevId: d851cd353cf9f31eb47ce13f43f6f935dc2f5d93
2026-10-28 11:25:13 +00:00
jif 13b0217567 Move Guardian context policy and prompts into the reviewer extension (#57698)
## What changed

Move `GuardianContextMode`, parent compaction checkpoint policy, and reviewer instruction assembly into `codex-guardian-reviewer`. Export the conversation-history prompt and follow-up reminder from the same crate, and update core to use these shared definitions.

Keep session-specific authorization version lookup in core while preserving existing context selection, checkpoint validation, and prompt behavior.

GitOrigin-RevId: a4cadb6c8ad95e756467316b7d8658e8cd527b24
2026-10-28 11:02:51 +00:00
jif d09a5cd8cf Move Guardian review state into the reviewer extension (#57697)
## What changed

Move review state, fork history, checkpoint recovery bookkeeping, and Node REPL evidence admission into `codex-guardian-reviewer`. Expose helpers for session initialization and snapshot publication, and update core to delegate to them while preserving the existing review logic.

GitOrigin-RevId: 56d6a0124301a3c70908614d4bc3bfe970ac1c85
2026-10-28 11:01:32 +00:00
ioanbudea-OAI f021168bde Wait for turn profiles before asserting sampling counts in trace tests (#57692)
## Why

`ShutdownComplete` can arrive before detached task cleanup records and exports `codex.turn_profile` events, causing the turn phase trace test to read incomplete profile counts.

## What changed

Poll exported spans until the expected number of profiles is available, with a 10-second timeout, before asserting sampling request and retry counts for reply and interrupt cases.

GitOrigin-RevId: 43cdcdde71bbcd16ed2e6c8b2accaa2a5de03001
2026-10-28 10:13:29 +00:00
Abhinav 154e8736a5 Support managed OpenTelemetry metadata and PEM client identities (#57658)
## Why

Managed OpenTelemetry requirements could not specify span attributes or tracestate metadata.

## What changed

- Accept `otel.span_attributes` and `otel.tracestate` in requirements, merge them by attribute or member field, and expose them through app-server configuration and requirements reads.
- Protect required entries and their ancestors from configuration writes while preserving unmanaged sibling values and their origins. For example, requiring `otel.tracestate.vendor.tenant` leaves `vendor.region` writable. Empty metadata maps impose no requirements.
- Select Rustls when configuring an OTLP HTTP client certificate and private key so PEM client identities are accepted.

## Testing

Add coverage for metadata precedence, effective configuration, field-level write protection, and preservation of unmanaged values. Add blocking and asynchronous HTTP client construction tests using generated PEM certificates and keys.

GitOrigin-RevId: ada6000cc5e575b9a33451ec4aa58b6686dc20f7
2026-10-28 06:24:24 +00:00
arnavpatidar-OAI bea47c06df Redact OTEL exporter headers in app-server config responses (#57654)
## Why

OTEL exporter headers can contain credentials. Configuration responses exposed these values in effective config, source layers, managed requirements, and write-override metadata.

## What changed

Replace header values with `[redacted]` across these responses for log, trace, and metrics exporters while preserving header names. Keep the original values available to runtime exporters, and redact them in debug output. Update protocol schemas and generated client documentation to describe the redacted values.

## Testing

Add coverage for effective config and source-layer redaction, managed requirements, and single and batch write-override metadata, including header names containing dots. Extend telemetry integration coverage to verify that runtime requests still carry the configured header value.

GitOrigin-RevId: ad24e8b91e615bd36c03c4c0ca205ac64bbd8cbf
2026-10-28 05:46:04 +00:00
Abhinav 727283cea5 Support managed OpenTelemetry configuration requirements (#57649)
## What changed

- Allow `requirements.toml` to manage OpenTelemetry logging flags, tool-result byte limits, environment, and log, trace, and metrics exporters.
- Apply higher-priority managed values over user configuration, replacing exporters atomically and warning when configured values conflict. Preserve unmanaged settings, including `span_attributes` and `tracestate`.
- Expose managed `otel` settings through `configRequirements/read`, apply them in `config/read` without ordinary field origins, and reject configuration writes that overlap managed fields.
- Distinguish omitted tool-result limits from explicit values while preserving the runtime default.

## Testing

Add coverage for requirement precedence, relative TLS paths, override warnings, managed write protection, effective configuration reads, and exporter serialization through the app-server API.

GitOrigin-RevId: 6a3343563bd3acb53a366ea6723684e3d0a4a01d
2026-10-28 05:11:45 +00:00
Brent Traut 04e365c698 Release sampling prompts after stream setup (#57618)
## Why

Legacy notifications only need user text, but retaining the full sampling input keeps image and audio payloads alive while the response streams.

## What changed

Extract notification text from the first request attempt only when `notify` is configured, and retain those strings across retries. Consume and drop the prompt after setting up the response stream, excluding attachments and media wrapper labels from notification text.

## Testing

Extend the notification integration test to verify that `input-messages` contains only user text while the model request still includes image and audio attachments.

GitOrigin-RevId: af4653c0e7ac2bf7006f5ae5a82a3e724c30c734
2026-10-28 02:00:30 +00:00
cassirer-openai 68b67f0ff4 Preserve rollout tracing across sub-agent reloads (#57615)
## Why

Reloading an evicted sub-agent disabled its rollout tracing, leaving later turns out of the trace. Replacement Code Mode sessions could also reuse cell IDs from earlier sessions.

## What changed

- Reattach reloaded children to the root rollout trace and emit `ThreadResumed` without duplicating `ThreadStarted`.
- Reopen existing threads during replay, preserving their identity and history while clearing stale runtime cell and terminal process mappings.
- Allocate Code Mode cell generations across sessions for both gRPC and stdio hosts to prevent ID reuse after reload.
- Record an aborted trace when session startup fails before the session loop begins.

## Testing

Add regression coverage for child eviction and reload, including a failed reload, continued inference tracing, distinct cell IDs, and preserved mail and follow-up context. Extend replay tests for history preservation and reused runtime handles, and host tests for cell IDs across session replacement and reconnection.

GitOrigin-RevId: 6c146c04b5cda9295afb7d973c0d5474b363a822
2026-10-28 01:29:34 +00:00
cooper-oai 9cce575134 Share runtime authentication with model providers and catalogs (#57607)
## Why

Model catalog refreshes need to use the same registered actor and credential owner as model requests when runtime authentication is enabled.

## What changed

- Add `ProviderAuthContext` to carry credentials and optional runtime authentication into model providers and catalog clients.
- Select runtime authentication independently for inference and catalog destinations. Reject unsupported OpenAI origins while preserving provider authentication for other destinations.
- Include API organization and project selectors in runtime registration. Reuse registered runtimes across projects within an organization and replace them when the organization changes.
- Preserve transport restrictions and account bindings, and strip sensitive headers on cross-origin redirects.
- Apply the catalog timeout to authentication and fetching while allowing credential refresh and registration to finish safely.

## Testing

Add coverage for runtime-authenticated catalogs, provider isolation, resource selection, gateway authentication, cross-origin redirects, and timeout recovery that preserves registration and refreshed tokens.

GitOrigin-RevId: c14f1a02159031ea0af1f9b26ea890f8dca7d6ad
2026-10-28 00:51:29 +00:00
goodman-oai e8e6a922c0 Support catalog headings for saved command-prefix notices (#57599)
## What changed

- Add `permissions.approved_command_prefix_saved` to model messages and use valid plain-text headings in saved-prefix notices, with or without full permissions instructions. Reject empty, oversized, control-character, and markup-like headings.
- Keep `Approved command prefix saved:` as the leading marker for older clients and recognize saved-prefix content kinds when trimming rolled-back context.
- Shut down analytics test exporters asynchronously and serialize workspace-switching tests that share authentication environment state.

## Testing

Extend coverage for heading validation, sparse wire values, single-notice emission, rollback with and without metadata, and saved-prefix behavior across model switches.

GitOrigin-RevId: a70175deca64a6339abac2fa69af970f470e3f4f
2026-10-27 23:53:34 +00:00
rhan-oai 37f199dab1 Add host-reserved V2 child creation APIs (#57586)
## What changed

- Add `CodexThread::create_child` to create an idle child from a live tool invocation using a host-reserved thread ID and controller.
- Add `prepare_child` and `ThreadManager::create_prepared_child` to capture inherited settings and fresh or forked history, then start the child on another manager even after the parent shuts down. Reject detached preparation when live bindings cannot be transferred.
- Add opt-in `HostToolInvocationLookup` to track live calls and reject duplicate active call IDs within a turn. Revoke invocation access when dispatch ends and clean up provisional children when startup fails or is cancelled.

## Testing

Add tests for invocation lifetime and call ID reservation, inherited policy and live bindings, detached startup with fresh and full history, and cancellation cleanup.

GitOrigin-RevId: 71f17ecb4cd7636517e675df59289a8e8c7105b8
2026-10-27 23:02:27 +00:00
Krish Chainani cb66dbecf2 Replace unavailable image references in model requests (#57585)
## Why

Image files can become unavailable between turns. Model requests should reflect their current availability while preserving the original references in saved history.

## What changed

- Add opt-in availability checks through `AttachmentStore::is_resolvable`, retaining references by default and on backend or access errors.
- Check distinct file IDs concurrently for each prepared request and replace confirmed-missing images with `[image unavailable]` in messages and tool outputs.
- Apply the same handling to compaction requests, with background image checks running outside the session lock without delaying foreground turn completion.

## Testing

Add unit tests and multi-turn scenarios covering placeholders, preserved history and annotations, concurrent deduplicated checks, backend errors, and nonblocking background compaction.

GitOrigin-RevId: ec98c7014c6e6b5063d87b943d2cd66a9ce7c1e3
2026-10-27 22:59:22 +00:00
Sean Huang 5078420b01 Preserve output boundaries through token truncation (#57584)
## What changed

Add `OutputBoundary` and token-budget formatting APIs that map item indices and UTF-8 text offsets to their positions after truncation. Normalize input positions, remap boundaries within omitted text, and keep end boundaries before generated omission notices.

Unify code mode output formatting: merge text-only output for truncation with a warning, and budget mixed text and audio in input order while preserving images and encrypted content. Accept owned output items to reuse allocations and move retained media without cloning.

## Testing

Add coverage for boundary normalization and remapping, UTF-8 offsets, zero and nonzero budgets, owned and borrowed output equivalence, and allocation reuse.

GitOrigin-RevId: 598407caa88b5a1f84130c4bc2273b85ab02f7f2
2026-10-27 22:56:23 +00:00
goodman-oai 133c11db43 Allow model catalogs to customize approval instruction headings (#57581)
## What changed

Add optional `request_permissions` and `auto_review_suffix` fields to catalog approval messages. Use them to customize the permission tool heading and introduce automatic review guidance while preserving the bundled instruction bodies and existing policy gates. Accept `request_permissions_tool` as a deserialization alias.

Limit headings to 500 UTF-8 bytes on a single line and fall back to bundled text for missing or invalid values. Complete catalog approval messages retain precedence.

## Testing

Add coverage for heading validation, serialization aliases, policy gates, fallback behavior, and complete-message precedence. Integration tests verify that model switches append the selected model's headings without rewriting history or repeating unchanged instructions.

GitOrigin-RevId: 3311591a464cc086783fe25271a352dd5a229790
2026-10-27 22:26:12 +00:00