Commit Graph
11588 Commits
Author SHA1 Message Date
ioanbudea-OAI 603a7baff0 Centralize synchronous Guardian context preparation (#57494)
## What changed

- Retain captured root conversation sources until synchronous preparation applies request limits. Keep asynchronous snapshots limited to the bounded messages.
- Move approval-reason truncation and tool-description escaping and truncation from composition into preparation.
- Extract synchronous image preparation and admission into `guardian-context`, preserving the prompt token cap and model context-window checks.

GitOrigin-RevId: d3a864c77d8f16bd3e77df436d80c4d2738c43cc
2026-10-27 15:53:50 +00:00
jif 6b56106ed9 Move Guardian action JSON formatting into the reviewer crate (#57493)
## What changed

Move approval request serialization and model-visible action formatting from `codex-core` into `codex-guardian-reviewer`, expose both helpers publicly, and re-export them from core. Preserve the existing JSON shapes and formatting.

## Testing

Move the MCP tool-call and network-access trigger serialization tests into the reviewer crate, retaining coverage for omitted optional fields and Windows path rendering.

GitOrigin-RevId: c0093c023b2f4c1be16bac481f1173a5da6ac25c
2026-10-27 15:49:32 +00:00
jif 3b600d0566 Move guardian prompt composition into the reviewer crate (#57488)
## What changed

Export `GuardianPromptItems` and `compose_guardian_prompt` from `codex-guardian-reviewer`, and delegate prompt composition from core to the new helper. Core continues to collect context sections and choose the history version.

Preserve full and delta transcript selection, empty-transcript placeholders, cursor accounting, and Node REPL evidence sequencing.

GitOrigin-RevId: e3e2b2a87a42709ddb5441f47cbb7461b0130579
2026-10-27 15:39:10 +00:00
Charlie Marsh dddc3c0a79 Skip unrelated indexed rollouts during selected thread migration (#57483)
## Why

Migrating selected threads reads rollout headers even for unrelated threads. Use the state database to skip files whose indexed owner and canonical filename agree that they belong to an unselected thread.

## What changed

- Read thread IDs and rollout paths without creating or migrating the state database, and fall back to header inspection if the index cannot be read immediately.
- Keep inspecting selected, unindexed, misnamed, and conflicting paths so rollout headers still determine ownership.
- Recognize stable thread IDs in reverted rollout filenames and match plain and compressed paths when filtering.

## Testing

Add a regression test covering pruning of unrelated indexed rollouts, including archived and compressed files, while retaining header inspection for selected and ambiguously named files.

GitOrigin-RevId: d76827174c00f5bc9e38ad9b69dadeee91f4415f
2026-10-27 15:19:58 +00:00
jif 75667b341a Move Guardian approval request types into the reviewer crate (#57478)
## What changed

Export `GuardianApprovalRequest`, `GuardianNetworkAccessTrigger`, and
`GuardianMcpAnnotations` from `codex-guardian-reviewer` and re-export them in
core. Move environment lookup and scope classification with the request type,
preserving existing classification and credential-safe debug output.

GitOrigin-RevId: 5253a5124d5e370a1230b46aa7dc48b24b1c902e
2026-10-27 15:01:35 +00:00
jif e344f4b6fc Move Guardian session helpers into the reviewer crate (#57476)
## What changed

Move Guardian session-source detection, the reviewer name constant, and parent-thread prompt cache key generation into `codex-guardian-reviewer`. Re-export them from core, preserving existing behavior.

## Testing

Move the existing prompt cache key test into the reviewer crate, retaining coverage for parent-thread scoping, key length, and missing or inapplicable inputs.

GitOrigin-RevId: c14b3c030ba93676a0799aba56b23a676dbbc56d
2026-10-27 14:56:16 +00:00
jif 6b3b67d71f Test encrypted native call history across Guardian reviews (#57474)
## What changed

Extend the incremental Guardian review scenario with historical encrypted `send_message` calls in the default and custom tool namespaces. Add snapshots and assertions that both reviews retain each call and parent reply exactly once, in order, preserving call IDs, namespaces, and encryption metadata with a paired evidence-only output.

Verify that successive reviews reuse the same thread and append to the previous input. The scenarios use mocked ciphertext and decisions to check transport behavior, not backend decryption or authorization outcomes.

GitOrigin-RevId: a0927531b6cdae496fefa2a7f35032f7413807b6
2026-10-27 14:48:48 +00:00
jif b89952d582 Fix guardian handoff mocks for native history tests (#57468)
## What changed

Add mock guardian approvals for setup handoffs in `handoff_scenario` when native multi-agent history is enabled. Match only the trailing `function_call_output` so historical handoffs cannot consume the final command's separate denial response.

Require five approval requests with native history enabled and none with it disabled.

GitOrigin-RevId: 7b14c97aa85adc63c4d3bf95014c44831eae1c85
2026-10-27 14:28:15 +00:00
swastik-oai 2ebf42f269 Always put Code Mode tool descriptions before namespace descriptions (#57462)
## What changed

Make tool descriptions precede namespace descriptions in Code Mode execution and tool search. Mark `code_mode_tool_description_first` as removed and simplify cached definitions to depend only on the input schema budget.

## Testing

Update Code Mode integration assertions and catalog snapshots to expect tool descriptions before namespace descriptions.

GitOrigin-RevId: 1344d4c264d1d2b18ee0388a064b6813ca81f770
2026-10-27 14:14:16 +00:00
jif 50983c7a93 Move Guardian token usage calculation into the reviewer crate (#57458)
## What changed

Export `token_usage_delta` from `codex-guardian-reviewer` and use it in the core review session. Preserve the existing calculation, including clamping negative deltas to zero.

## Testing

Move the existing nonnegative token usage delta test alongside the helper in the reviewer crate.

GitOrigin-RevId: fcfce24b8d50a63dcd1122a4204939bc6750feda
2026-10-27 14:01:44 +00:00
jif 90de7b92b6 Centralize Guardian classification of multi-agent tool calls (#57455)
## What changed

Add `GuardianScope::for_multi_agent_tool` and use it for both tool review requirements and asynchronous scoring. Classify `spawn_agent`, `send_message`, `followup_task`, and `interrupt_agent` as native tools when they match the configured namespace, accepting default namespace spellings when none is configured. Keep review enablement checks at each call site.

GitOrigin-RevId: 9180ccdc629532f6c171d899dc5881e15b447141
2026-10-27 13:55:51 +00:00
Tom Leach (@toml) 6a6e2fb1f0 Distinguish policy-restricted capacity failures from server overloads (#57452)
## What changed

- Recognize `policy_capacity_unavailable` in HTTP and streaming API errors and expose it as `policyCapacityUnavailable` through the app-server protocol and generated bindings.
- Display “Policy-restricted capacity is unavailable. Please try again later.” while preserving overload retry behavior based on `Retry-After`.
- Handle the new error in TUI overload rendering, reviewer retries, and error telemetry.

## Testing

Add coverage for HTTP and streaming error classification, retry advice, WebSocket terminal error attribution, protocol serialization, and TUI rendering with and without an upstream message.

GitOrigin-RevId: 1b9536fba1e11cc891dcf689aee35410d82f5412
2026-10-27 13:31:38 +00:00
jif 824b7666fb Route state-changing multi-agent calls through Guardian review (#57451)
## What changed

When `GuardianNativeMultiAgentHistory` is enabled, require Guardian review for wrapped Multi-Agent V2 calls to `spawn_agent`, `send_message`, `followup_task`, and `interrupt_agent`.

Include these calls in Guardian's asynchronous native-tool scoring scope, honoring the configured multi-agent namespace and preserving that configuration across overrides.

## Testing

Extend the incremental-review scenario with an encrypted, namespaced `send_message` call. Verify that the encrypted parent reply remains present exactly once in both Guardian reviews and the subsequent agent request, and update the snapshot to cover native tool-call review evidence.

GitOrigin-RevId: 2af9561b0d69af4a8f878fa32365ce845368e02d
2026-10-27 13:21:20 +00:00
jif eec8bde1e9 Preserve native multi-agent calls in Guardian review history (#57448)
## Why

Native multi-agent calls must stay intact for backend decryption during Guardian review.

## What changed

When `guardian_native_multi_agent_history` is enabled for multi-agent v2, preserve `spawn_agent`, `send_message`, `followup_task`, and `interrupt_agent` function calls in the configured tool namespace as native transcript items. Use `functions` when no namespace is configured.

Include the paired evidence-only output in token accounting, and omit native evidence whole when limits require it.

## Testing

Extend transcript tests to cover native call ordering, paired-output budgeting, and whole-item omission. Add a native-history snapshot alongside the existing text-history delegation handoff scenario.

GitOrigin-RevId: e7d9b4f5ea772890bf14b79fdb53b0f9d0a842e4
2026-10-27 13:02:57 +00:00
jif a5ef02e561 Add Guardian native multi-agent history feature flag (#57441)
## What changed

Register `guardian_native_multi_agent_history` as an under-development feature,
disabled by default, and expose it in the configuration schema.

GitOrigin-RevId: d434750c1aa0cc78b36d1d661a4df0def5426f02
2026-10-27 12:28:07 +00:00
jif e1dd7ed1ba Preserve Guardian request metadata with custom OpenAI base URLs (#57439)
## Why

Custom OpenAI base URLs caused Codex backend authentication to lose Responses headers and Guardian metadata when the URL did not match a recognized backend route.

## What changed

Apply model-scoped Responses headers, Guardian reviewer and classifier markers, and eligible ChatGPT parent credit requests based on the OpenAI provider and authentication configuration, regardless of the base URL. Keep API-key requests and other providers excluded from this behavior.

## Testing

Extend coverage for custom `/v1` URLs and direct `/api/codex` endpoints with ChatGPT, header-based, and API-key authentication, including HTTP fallback and WebSocket requests.

GitOrigin-RevId: cc4bb5b4a1a6596904ecf8965237c7a4f09e112b
2026-10-27 12:17:14 +00:00
swastik-oai 5557e08853 Carry prepared commands through unified execution (#57438)
## What changed

Pass `PreparedExecCommand` through the process manager and runtime instead of copying command, shell, environment, and permission fields into separate request structs. Move approval requirement evaluation and approval action construction onto the prepared command, and use it for shell snapshots and launch failure events.

GitOrigin-RevId: 3e9f87e0974218fea6fffc0f7740616d696b9800
2026-10-27 12:12:55 +00:00
swastik-oai aeb6b22c8b Extract unified exec command preparation into a reusable module (#57435)
## What changed

Move argument parsing, working-directory and shell resolution, permission validation, and command lifetime setup into `unified_exec::preparation` with crate-visible parsed and prepared command types.

Update `ExecCommandHandler` to use the extracted preparation flow, preserving implicit skill event emission between parsing and preparation.

GitOrigin-RevId: 79e9be5d87c1b0ab905afdf1e343ac141fb7d4c6
2026-10-27 11:42:38 +00:00
jif ed1b92fa45 Add opt-in synchronous Guardian review for native tool calls (#57433)
## What changed

- Add `CoreToolRuntime::requires_guardian_review` to review opted-in function calls before dispatch and stop execution when review rejects the call.
- Deliver proposed calls as native review evidence, preserving encrypted argument metadata. When hooks rewrite arguments, review the updated arguments and discard the original encrypted replay content.
- Deduplicate native call evidence against reviewer history and include paired evidence outputs in input budgets.
- Keep review accounting and metrics for native calls without emitting action-specific client assessment events, and omit approval payloads from debug output.

## Testing

Extend regression coverage for rewritten native calls across review retries, denied calls that never execute their handlers, oversized native evidence, and review metrics without client events.

GitOrigin-RevId: c3b22c0220344581c2b38f233001abaed49cba34
2026-10-27 11:37:24 +00:00
jif b6ad8ea1fc Test goal provenance in existing workers' Guardian reviews (#57430)
## What changed

Add a parameterized app-server test that sets a paused goal after a worker has spawned and completed its first Guardian review, then asks the worker to check again.

Verify that a user-origin goal appears in the subsequent review's root authorization context as user-set, while an automatic-origin goal is excluded. Also assert that the initial review does not contain the later goal.

GitOrigin-RevId: 1d67c4a5481a717eac706b64ee714e7da2fc5511
2026-10-27 11:05:48 +00:00
ioanbudea-OAI 5c6ef3d1a6 Centralize Guardian root context preparation (#57425)
## Why

Verified answers omitted because of request-size limits must be marked as missing. Legacy checkpoint messages also need to match retained excerpts after truncation so workers receive only one copy.

## What changed

Move root message preparation into `codex-guardian-context::prepare_root_context`, including handoff selection, shared message limits, legacy matching, and incomplete-context markers. Pass original legacy text into preparation and mark verified answers incomplete when oversized records are excluded.

## Testing

Add a handoff regression scenario for an ID-less checkpoint with a retained excerpt and an oversized verified answer. Assert that the worker receives one capped user message, both incomplete-evidence markers, and an incomplete authorization version.

GitOrigin-RevId: bd3f5fe164ae2fddba440e9269fd63b52e8b12f8
2026-10-27 10:29:14 +00:00
Keyan Zhang 124e4e5bfa Make thread item timestamps optional in TypeScript exports (#57420)
## Why

Older servers can omit thread item timestamps. The TypeScript contract should accept those responses while still allowing explicit `null` values.

## What changed

Mark `ThreadItem.startedAtMs` and `ThreadItem.completedAtMs` as optional nullable fields. Update the generated TypeScript and precomputed stable and experimental exports, and allow these fields in the optional-nullable export check.

GitOrigin-RevId: 186be7aa45a3e7c65f133b1e320809a9912ff1f3
2026-10-27 10:05:37 +00:00
alexsong-oai 6fe77fc123 Add experimental GCP account setup to the app server (#57394)
## What changed

- Add `account/gcp/setup` to configure the built-in Vertex AI provider with a project, optional location (defaulting to `global`), and optional credential file. Omitting the file clears the active user override and uses Application Default Credentials.
- Validate configuration and obtain an access token before saving, return structured credential failure reasons, and reload the model catalog on the next `model/list` request.
- Share configuration previews and version-checked writes with Bedrock setup to reject conflicting settings and concurrent configuration changes.

## Testing

Add tests for credential selection and ADC fallback, catalog refresh, invalid credentials, policy and endpoint conflicts, cancellation when network permission is revoked, and configuration version conflicts.

GitOrigin-RevId: 6f411a85893f8da49581427451e0180529955b16
2026-10-27 08:06:10 +00:00
Eric Traut d0f33fab36 Render integral limits and parenthesized matrices in TUI math (#57393)
## What changed

- Stack integral limits above and below the integral in top-level display math, while keeping compact limits in lists and blockquotes.
- Render `pmatrix` environments with centered cells and scalable Unicode parentheses, aligning surrounding math to the matrix midpoint.
- Require spatial layouts, including single-row matrices, to fit the available width; preserve the source when they cannot fit or appear in nested Markdown.

## Testing

Add snapshots for integral limits and matrices, narrow-width fallback assertions, and rejection tests for ragged matrices and mismatched environment endings.

GitOrigin-RevId: 73b589c4422568b90d15f55bf7c9dddec0dccc97
2026-10-27 08:05:46 +00:00
Eric Traut d30fe145fc Drain pending terminal reports before returning to the shell (#57391)
## Why

Terminal reports still in flight when Codex exits can leak into the parent shell's input.

## What changed

Pause terminal event polling during shutdown and flush pending input before restoring normal terminal input mode. On Unix, use a primary device attributes (DA1) reply to wait for outstanding reports after disabling reporting, with a two-second timeout when startup detected DA1 support. During panic unwinding, flush queued input without locking the event reader.

## Testing

Extend PTY tests to cover mouse-report bursts with a split DA1 reply, clean shell input after exit, and queued input cleanup during panic unwinding.

GitOrigin-RevId: 304859b654fb2b2c8882e6cac4cb6522fe05475c
2026-10-27 08:04:25 +00:00
Anthony Tafoya 2cab11ca40 Pin the Rust toolchain to 1.95.0 (#57373)
## What changed

- Replace Rust `1.99.1` pins with `1.95.0` across Cargo, Bazel, CI, release workflows, and Windows setup, and update the Bazel lockfile.
- Adjust Rust APIs, borrowing, and lint-related code for the pinned toolchain, and box MCP tool dispatch to bound compiler async-layout query depth.
- Update the halfwidth sound mark snapshot and allow optional symbol hash suffixes in the code mode host package smoke test.

GitOrigin-RevId: 75460c3d0465dc601811e619bce373a4f7ff2912
2026-10-27 05:52:01 +00:00
Eric Traut 155dde7ff6 Show the selected ChatGPT workspace in /status (#57366)
## What changed

- Return the selected workspace's display name as `workspaceName` in ChatGPT account responses, caching it alongside workspace routing and updating generated schemas and bindings.
- Display a `ChatGPT workspace` row with the name and account ID, falling back to the ID when the name is unavailable or invalid. Refresh workspace details with account identity updates and hide the row for providers that do not use ChatGPT authentication.

## Testing

Update account-response tests to cover workspace names and null values, add workspace details to the status snapshot, and verify that switching to a custom API-key provider hides the workspace row.

GitOrigin-RevId: 4b30ab8a39e1fa7ad1ba7367c5295544be8f6001
2026-10-27 05:22:23 +00:00
Eric Traut 6c0db39b94 Offer safer rewrites after safety blocks in the TUI (#57365)
## What changed

- Show selectable safer follow-ups alongside safety-block explanations, with an option to enable Daybreak when disabled.
- Submit the selected rewrite in a verified fork before the blocked turn, preserving the original chat's draft and current model settings. Mark the new turn with `safety_recourse_follow_up`.
- Filter suggestions that could invoke commands, skills, apps, or plugins, and suppress recourse for unsupported servers and non-primary threads.

## Testing

Extend coverage for fork history, draft preservation, model selection, turn metadata, suggestion filtering, and the Daybreak picker.

GitOrigin-RevId: 49052df6301b6bfbba07262d40b007dcf97bb7c9
2026-10-27 05:17:43 +00:00
Eric Traut 118aae2a5f Explain safety policy blocks with turn-scoped recourse in the TUI (#57363)
## What changed

- Handle `TurnModerationMetadata` for the active turn, showing a temporary block notice while an explanation is pending.
- Validate safety recourse metadata and render topic-specific explanations with safer ways to continue for biological and cybersecurity policy blocks. Apply the explanation only when the terminal policy error matches its domain; otherwise retain the existing error handling.
- Queue input while recourse is pending and preserve prepared images when queued messages are submitted.

## Testing

Add an integration test covering biological and cybersecurity blocks, the loading notice, and fallback for a mismatched policy error, plus a snapshot of the cybersecurity explanation. Extend queued-image coverage to verify submission succeeds after the original image file is removed.

GitOrigin-RevId: dd9ca57cf3b2d842e3deb8aff5d7c3c73a92e14d
2026-10-27 05:15:31 +00:00
Dmitriy Kovalenko ff2346f33f Bound waits for the thread writer coordination lock (#57360)
## Why

Contention on the writer coordination lock can block writer cleanup indefinitely, preventing idle threads from unloading.

## What changed

Retry coordination lock acquisition every 50 ms and return `io::ErrorKind::TimedOut` after 30 seconds of contention. This lets cleanup release writer ownership and idle unloading finish even while another process holds coordination.

## Testing

Extend writer lock coverage to check the timeout error. Update the idle unload regression test to verify that the thread closes while coordination remains locked and can resume with its persisted turn after the lock is released.

GitOrigin-RevId: 513117fdaf3890283e7acdd149ce9abb66075b63
2026-10-27 04:52:09 +00:00
Dmitriy Kovalenko adfcac1a42 Serialize session index mutations across processes (#57358)
## Why

The process-local mutex cannot coordinate session name appends and removals from multiple Codex processes sharing the same home directory.

## What changed

- Guard index mutations with a persistent `session_index.jsonl.lock` file, retaining the lock until the blocking worker finishes even if its caller is cancelled.
- Return a timeout error after 30 seconds of lock contention without modifying the index.
- Rewrite removals through a unique temporary file while preserving index permissions, and skip removal when the index is absent.

## Testing

Update regression tests to check lock retention after cancellation, timeout without writing, successful writes after contention clears, and removal with a missing home directory. Retain coverage for cancelled queued updates preserving the next write.

GitOrigin-RevId: 5231c9ff85c6b34de5ee92931d119b6f46bd583b
2026-10-27 04:47:37 +00:00
Dmitriy Kovalenko f69b8d203d Report when NFS database migration finds no databases (#57357)
## Why

`codex migrate-nfs-db` reported a successful migration when the database directory existed but contained no databases.

## What changed

Count databases that were migrated or already use DELETE journal mode. If none are found, print `No databases found.` instead of the migration success message.

## Testing

Extend the CLI integration test to cover an empty directory, a failure on a corrupt logs database after an earlier conversion, and a successful rerun that recognizes the converted database as already in DELETE mode.

GitOrigin-RevId: 16b109645925164f3af5d35a909417872cfa004f
2026-10-27 04:46:24 +00:00
iceweasel-oai 3f74ff28fa Avoid object-identity preflight for MXC path grants (#57353)
## Why

MXC's `normalize_object_conflicts` can reject otherwise usable Windows policies when a network share does not support `FileIdInfo` or a protected deny path cannot be opened for metadata.

## What changed

Remove object-conflict normalization before and after volume-root expansion. Send explicit path grants directly to BaseContainer, which enforces path and symlink/junction alias precedence.

Document the resulting limitations:

- Hard-link names retain independent permissions, so an allowed name can access the same file through a less restrictive grant.
- Volume roots expand using their declared access without first reconciling aliases. A writable root can therefore generate more-specific writable grants that override a read-only root alias.

GitOrigin-RevId: de0b2d7b5b98eaa358a97156c02e9f7f85860335
2026-10-27 04:39:28 +00:00
Benjamin Carlsson b7f7e7f7e8 Separate TUI session resume preparation from activation (#57342)
## What changed

Split saved-thread resume into preparation and activation. Preparation resolves configuration and attaches to the destination thread before activation replaces the displayed conversation.

Keep stored-backend configuration and local settings separate from active settings during preparation, while retaining the existing eager settings refresh for active-backend resumes. Use the destination backend for working-directory, permission, and directory-trust checks.

Carry the backend identity in the prepared result and reject activation through a different backend before changing the active conversation or composer.

## Testing

Extend coverage for preserving active settings and composer drafts during preparation, rejecting activation through another backend, applying destination remote-permission checks, and retaining active-backend settings refresh behavior.

GitOrigin-RevId: 7171f1abcf4bdd62793c21232a8b895bb809cf70
2026-10-27 04:09:04 +00:00
goodman-oai c587d6e54c Add model catalog fields for granular approval templates (#57325)
## What changed

Add optional `approvals.granular` catalog fields for the policy introduction and prompted and rejected category lists. Preserve template strings verbatim in `GranularApprovalMessages`, leaving validation and rendering to consumers.

Document the 500-byte template limit, required `{{ policy }}` and `{{ categories }}` markers, and fallback rules.

## Testing

Extend serialization tests to cover null, empty, and populated granular templates, and verify their preservation through model responses and base instruction overrides.

GitOrigin-RevId: ce32e41a9b37e2e7c45ac9b079708a640c99326c
2026-10-27 01:50:46 +00:00
Alex Daley ce7adb6a31 Add opt-in concise hints for explicitly selected plugins (#57317)
## What changed

Add the `features.simplified_selection_hints` flag, disabled by default. When enabled, replace detailed plugin capability instructions with a concise developer message identifying the selected plugin and directing the model to use its available skills and tools.

Continue omitting hints for plugins without available capabilities and bounding hint length. Preserve the existing detailed instructions when the flag is disabled.

## Testing

Cover plugin refresh in an existing thread with the flag enabled and disabled. Verify that explicit skill prompts precede the simplified plugin hint and that the hint uses the developer role.

GitOrigin-RevId: 64fc3347bd49fffe675f343a4733ced1b4e5e88c
2026-10-27 00:52:31 +00:00
Benjamin Carlsson b01148de59 Keep failed subagents from marking tasks as needing attention (#57315)
## Why

A subagent failure ends its work without requiring user action, but the agent command center treated it as needing attention and promoted its parent task into the `Needs you` group.

## What changed

Classify failed subagents as inactive when computing task groups and selecting details. Recognize subagents by either a parent thread or a subagent session source, including background workers without a parent.

Keep approval and user-input requests from descendants actionable, even beneath a failed agent. Show these requests as `Needs input` rather than `Error` when the displayed task is a failed subagent. Errors in non-subagent tasks continue to require attention.

## Testing

Add and update regression tests and a snapshot covering stable ordering after subagent failures, idle parents staying ready, failed background workers, and pending requests beneath failed agents.

GitOrigin-RevId: 1a06d5d00c08bb77c9ed36aadffd5e0514f0b2c6
2026-10-27 00:29:59 +00:00
Joe Savona 95cafae96c Format all Rust sources and clean up repository checks (#57314)
## Why

Rust formatting should include source files that Cargo cannot discover, while keeping command lines within Windows limits.

## What changed

- Update `scripts/format.py` to discover tracked and non-ignored untracked Rust files with Git and invoke `rustfmt` in batches of 64 files. Use the workspace configuration and skip recursive child-module formatting.
- Pass `SystemDrive` and `LOCALAPPDATA` to Windows Bazel tests.
- Extend the spelling allowlist, remove a manifest feature exception, and clean up an unused import, an empty MCP scenario module, and its snapshot.

GitOrigin-RevId: c0d625fda9290ead6a63919325d65f7352abfa40
2026-10-27 00:23:39 +00:00
smurching-oai 9526f8d0d0 Default gateway OAuth delivery to Authorization bearer tokens (#57313)
## What changed

Allow omitting `gateway_oauth.delivery` to send the gateway token as `Authorization: Bearer …`. Custom headers and cookies remain available when a provider also needs its own credentials.

Reject conflicting provider authentication and duplicate delivery headers with actionable errors. For gateway bearer delivery, require `model_catalog_url` to share the scheme, host, and port of `base_url`. Consistently reject request header collisions across auth application methods.

## Testing

Add coverage for omitted delivery, provider credential conflicts, and model catalog origin validation. Update inference tests to verify gateway-only authentication while signed in and combined ChatGPT and gateway credentials in separate headers.

GitOrigin-RevId: 8564d3998511b3f6c55f3b65a01ed9056a90417e
2026-10-27 00:08:07 +00:00
ioanbudea-OAI 1f281290e3 Centralize Guardian root context preparation (#57312)
## What changed

Move root message preparation and handoff context selection into `codex-guardian-context`. Keep source collection and handoff boundary detection in core, while shared helpers handle text limits, whole-record verified answer admission, delivered assistant message lookup, and retained assistant fallback.

GitOrigin-RevId: a3eb35cddb3b0c5b93243b9d9c553397a7a61a14
2026-10-27 00:01:05 +00:00
cassirer-openai e977cb394d Preserve tool output versions and history order in rollout traces (#57311)
## Why

Tool calls can emit multiple outputs, and compaction can trim output content while retaining the same call and item IDs. The trace reducer rejected these versions as conflicting content. Late tool and code-cell events could also link outputs in map-key order instead of conversation order.

## What changed

- Preserve distinct tool output versions while retaining consistency checks for tool invocations.
- Link tool and code-cell outputs in thread history order, and resolve pending code cells after the entire event has been reduced.
- Prevent snapshot reconciliation from reusing the same conversation item twice when identical items move within history.

## Testing

Extend reducer regression tests for trimmed outputs, changed images and metadata, reordered identical messages, and late tool and code-cell events spanning the item-ID ordering boundary.

GitOrigin-RevId: eb4242d6a2d32d61c3caab9d3cc3cf6b374ce799
2026-10-26 23:55:28 +00:00
smurching-oai fa4c0607d4 Surface refresh-token rejection errors during workspace routing (#57304)
## Why

When workspace discovery receives a 401 and token refresh is permanently rejected, the generic discovery error hides the guidance to sign in again.

## What changed

Propagate cached permanent refresh-token failures for the current credentials through workspace routing and model-provider resolution as `CodexErr::RefreshTokenFailed`. Preserve account-change checks and only surface the cached rejection when external authentication is absent and the auth generation is unchanged.

## Testing

Add an app-server regression test that simulates `refresh_token_reused` during workspace discovery and verifies that the turn fails with `Unauthorized` and tells the user to sign in again.

GitOrigin-RevId: cf89c71fd6a2103e56d637838b0cb34d58b3cc32
2026-10-26 23:21:18 +00:00
ioanbudea-OAI 3e9264fb18 Defer synchronous Guardian evidence limits to preparation (#57295)
## What changed

Collect original retained instructions, assistant context, and verified answers before applying synchronous request limits in `guardian-context` preparation. Carry original record sizes through collection, preserving whole-record omission notices, source ordering, and host assistant-context framing.

Apply retained transcript source size checks during synchronous transcript preparation as well.

## Testing

Extend regression assertions for synchronous/asynchronous retained-context parity and equivalent verified-answer omissions when oversized records are filtered during preparation.

GitOrigin-RevId: 9042ee3fe60ebd455f6656b4008efb2acc44ac3d
2026-10-26 22:27:05 +00:00
Anthony Tafoya 62f7f3418f Extract plugin manager errors and box realtime handoff routing (#57294)
## What changed

- Move plugin install and uninstall errors and telemetry classification helpers into `codex-rs/core-plugins/src/manager_errors.rs`, preserving the existing public re-exports and error classifications.
- Box the realtime text routing future to bound the handoff future's size while retaining the admission permit across the await.

GitOrigin-RevId: ccf0c3bb93a8c69a0a618b96cd1525da87003176
2026-10-26 22:23:59 +00:00
BOZHANG-OAI 126a0ee92e Add GPT-6.1 Terra to model catalogs (#57293)
## What changed

- Add `gpt-6.1-terra` to the bundled catalog and GCP Vertex model picker.
- Add `openai.gpt-6.1-terra` to Amazon Bedrock catalogs, including Global and US cross-region runtime variants.
- Use `gpt-6.1-terra` as GCP's preferred memory consolidation model.

## Testing

Extend catalog, model picker, thread startup, and Vertex request tests to cover GPT-6.1 Terra.

GitOrigin-RevId: edb3be3c4286c191349cdee742005ed3c67029e8
2026-10-26 22:13:14 +00:00
Ian MacLeod b52a272cff Render async questions and replies with Markdown in the TUI (#57292)
## Why

Async question replies appeared as plain user text, leaving Markdown formatting literal and repeating the full question above each answer.

## What changed

- Accent question titles with a `?` prefix while preserving Markdown in titles and suggestions.
- Render replies with a `↪ Re:` question caption and a Markdown answer. Limit captions to two rows with an ellipsis on overflow, while keeping answers complete.
- Apply the same rendering to live messages and loaded transcripts, preserving inline styles, hyperlinks, and Markdown when copying selections.

## Testing

Add snapshots and assertions for matching live and loaded transcript rendering, plus tests for caption clipping, hyperlink bounds, Markdown copying, partial selections, and whitespace-only or nonprinting answers.

GitOrigin-RevId: 0290d8a2ebfb42bd5d7374c7037cde5ae8a13b5d
2026-10-26 22:12:07 +00:00
Ian MacLeod a014908c50 Represent user message presentation with an enum (#57291)
## What changed

Replace `UserHistoryCell`'s `spoken` boolean with `UserMessagePresentation`, defaulting to `Plain` and using `Spoken` for speech transcripts. Route rendering checks through `is_spoken()`, preserving the existing prompt styling and whitespace handling.

GitOrigin-RevId: d0683a2b536a7d71f68fd9a089e6839d0ab0cc65
2026-10-26 22:11:27 +00:00
Joe Savona 16effdfc8c Add a plugin process host with bounded IPC and lifecycle management (#57290)
## What changed

- Add `codex-plugin-host` to launch a helper with a cleared environment, verify its readiness handshake, and expose framed stdin/stdout pipes with a 256 KiB frame limit and write backpressure.
- Support startup cancellation and process-group cleanup, retaining exit status and the last 16 KiB of stderr while omitting raw diagnostics from `Debug` output.
- Extend the Bazel crate macro to support test-only binaries and fixture entry points outside `src/`.

## Testing

Add process integration tests for framed communication, startup cancellation, helper and descendant cleanup, repeated close calls, and bounded diagnostics from startup and runtime failures.

GitOrigin-RevId: ed539e7899af06ee83a0d8c2d0ea8cec10f19afb
2026-10-26 22:11:01 +00:00
Keyan Zhang b46a3801e4 Preserve parent tool declarations in prediction forks (#57287)
## What changed

Keep inherited incremental tool declarations in MCP-disabled prediction forks and `thread/sample` requests, and append a developer notice that MCP tools are unavailable. Skip tool declaration updates within the inherited context window; new windows still declare the current tools.

Add the default-enabled `prediction_preserve_parent_tools` feature flag and its configuration schema entries.

## Testing

Update core snapshots and app-server tests to cover inherited declarations, the MCP-unavailable notice, preservation across MCP refresh, and continued parent tool availability.

GitOrigin-RevId: 3adf96c7d0e0093986dbdbd540bd47674a2c105c
2026-10-26 22:05:23 +00:00
viyatb-oai 136d891a94 Support network_proxy in managed requirements (#57286)
## What changed

Accept `[network_proxy]` in `requirements.toml`, retaining `[experimental_network]` as a compatibility alias. Reject both names in a single layer and normalize them before merging across layers so precedence and source tracking are preserved.

Honor the new spelling during credential broker project discovery, display `network_proxy` in debug configuration, and update diagnostics and documentation. Document the distinction between managed requirements, proxy feature configuration, and permission profiles. Policy publishers must keep using `[experimental_network]` until all clients and policy validation layers support the new spelling.

## Testing

Add coverage for alias equivalence, duplicate rejection, cross-layer precedence and sources, and broker project discovery. Exercise both spellings in proxy configuration, network approval routing, app-server requirements responses, and Windows machine policy validation.

GitOrigin-RevId: 5c5bb82bd4b6b1431aa97f27248775820eae6302
2026-10-26 22:05:00 +00:00