## Why
Automatic goal continuations can repeatedly return empty final answers without
making progress. Stop this loop by marking the goal as `blocked` after three
consecutive empty turns with no other activity.
## What changed
- Observe completed turn items through a new `on_item_completed` lifecycle hook.
- Track empty final answers only for automatically admitted goal turns, resetting
the streak on activity, user turns, or goal changes.
- Preserve normal turn completion and streamed message deltas when blocking a goal.
## Testing
Add accounting coverage for the three-turn threshold and streak resets, plus
app-server tests for empty continuations and recovery through final-answer text,
commentary, or tool activity.
GitOrigin-RevId: 4b9d2cb2e306b0adc316972429cc35000115b88b
## Why
Hosted Codex Apps shared the `mcp_2026_07_28` protocol setting with other MCP servers, preventing independent protocol selection.
## What changed
- Add `features.codex_apps_mcp_2026_07_28`, disabled by default, for the host-owned HTTP `codex_apps` server. Enabling it attempts 2026-07-28 discovery with Legacy fallback.
- Keep other servers, including third-party registrations with Apps-like names or URLs and local stdio servers, governed by `mcp_2026_07_28`.
- Support the new flag through `experimentalFeature/enablement/set` and refresh protocol settings in existing sessions. Track protocol mode per server when deciding whether to reuse connections.
## Testing
Add coverage for independent feature and protocol selection, hosted event-stream protocol selection, standalone Legacy defaults and discovery fallback, and runtime enablement refreshing an existing app-server thread.
GitOrigin-RevId: 640e8e251f0c45e5f4829cc2b73c3082cf42f715
## Why
Saved threads need to resume after a managed daemon restart so active goals can continue without waiting for a client to reconnect.
## What changed
- Consume the recovery snapshot at startup and restore threads in the background through the shared cold-resume path.
- Trigger idle lifecycle work without attaching a client subscription, and leave already-loaded threads untouched.
- Continue recovery after individual thread failures and abort background recovery during shutdown.
## Testing
Extend the daemon restart integration test to verify goal continuation before client reconnection, restored conversation history and tools, idle-thread restoration, and recovery snapshot consumption. Include malformed snapshots and an invalid thread ID.
GitOrigin-RevId: 90e7b8f74ddc0ae327bacde1b2108b9749c65f2a
## Why
Remote-control requests could bypass a server's `Retry-After` deadline through pairing, auth changes, or reconnects. Proactive token refresh could also continue using a valid token despite an explicit server delay.
## What changed
- Share retry deadlines across enrollment, token refresh, pairing, pairing status, and WebSocket handshakes, preserving them across enrollment replacement and disable/enable transitions.
- Honor `Retry-After` on HTTP `429` and `503` pairing and handshake responses, including incomplete pairing response bodies.
- Add up to 30 seconds of jitter once per response without shortening the server's deadline, including for `Retry-After: 0`.
- Defer new requests while retaining valid tokens, and prevent auth changes from shortening the server delay while keeping shutdown responsive.
## Testing
Add regression coverage for shared deadlines, concurrent pairing and handshake failures, auth recovery, disable/enable transitions, retries after expiry, prompt shutdown, and jitter bounds.
GitOrigin-RevId: 69a5a46559031709a4558190c3142eb65e2ae634
## What changed
- Add a tag-release workflow gated by `CODEX_PROVISIONED_MACOS_CANDIDATE` for Apple Silicon and Intel macOS candidate artifacts.
- Package the CLI in `CodexCLI.app` with an embedded provisioning profile and a relocatable `bin/codex` launcher. Validate independently supplied profile, certificate, and team expectations before signing.
- Sign and notarize candidates, then verify signatures, entitlements, architecture, stapling, and Gatekeeper acceptance on macOS. Retain verified artifacts after package smoke tests pass.
- Filter code-mode smoke-test requests to `/v1/responses` so analytics requests are excluded from response parsing.
## Testing
Add tests for profile validation, launcher relocation and symlink handling, standard and provisioned signing flows, and rejection of signing, notarization, identity, and entitlement failures. The candidate workflow runs package smoke tests, including sandboxed code mode.
GitOrigin-RevId: f214f6a23dd10df62cd72a2c63c20cb0864fbc9c
## What changed
- Atomically save successfully persisted, loaded root thread IDs to `app-server-daemon/loaded-threads.json` during graceful managed shutdown, excluding ephemeral threads and threads pending unload.
- Clear stale recovery state at startup and around daemon lifecycle operations, including explicit stops and fresh starts.
- Probe support for `--managed-daemon` before launching managed app servers, and clear recovery state best-effort when launching older binaries.
- Keep forced shutdown responsive while saving the snapshot and log snapshot failures without preventing shutdown.
## Testing
Add coverage for snapshots after active turns finish, forced shutdown without a snapshot, excluded threads, snapshot write failures, and recovery-state cleanup for stops, fresh starts, and legacy launches.
GitOrigin-RevId: 422dc32b85355510f7e4c57b7540597b47d9f0b2
## What changed
Populate `turn_trigger` in request metadata with `guardian_review` for guardian reviews, `guardian_classifier` for classifier requests, and `memory_consolidation` for both detached memory requests and consolidation agent turns.
## Testing
Extend request metadata assertions to cover each trigger across guardian reviews, classifier requests, and both memory startup phases.
GitOrigin-RevId: 2dc3c173ede6563b942b8099ad9901c32967b969
Run `hook_drains_output_and_times_out_while_stdin_is_blocked` with
`cmd.exe /D /C` on Windows and `/bin/sh -c` elsewhere to keep user shell
startup files out of the test.
GitOrigin-RevId: 296e65cf76b74543e91c9ed9024ba2442e7e33af
## Why
Async classifications need to account for the complete request, including parent compaction checkpoints and images, before sending it to the classifier.
## What changed
- Resolve the input allowance from the classifier's model metadata, independently of parent-model context-window overrides.
- Reject requests whose estimated input exceeds that allowance minus a 256-token reserve. Record `input_too_large` and defer to synchronous review without sending the oversized request or dropping evidence to make it fit.
## Testing
Add integration coverage for checkpoint and image budgets, verifying that oversized inputs defer to synchronous review and inputs that fit retain their evidence. Update Guardian context-budget tests to exercise V2 remote compaction.
GitOrigin-RevId: 765060e08d2d5f3028ef5207508d256d4bf8d856
## Why
`update_goal` only accepted `complete` and `blocked`, preventing the agent from pausing a goal in response to an explicit user request.
## What changed
- Accept `paused` and account for final goal progress when pausing, with budget limits taking precedence.
- Update tool instructions and goal prompts to allow pauses only at the user's explicit request, report the returned status, and stop goal work. A later resume revokes the pause request.
## Testing
Extend coverage for pause accounting, budget-limit precedence, rejection of resume and system-limit statuses, and preservation of a tool-paused goal when resuming a thread.
GitOrigin-RevId: c3c15a51f848ce6eb65854d51b1a1bc75bb456a7
## Why
MXC volume-root grants are nonrecursive, and filesystem deny globs were rejected before launch.
## What changed
- Expand volume-root grants to include existing immediate children, preserving read-only carveouts and denies. Children created directly under a volume root after policy construction are not granted access.
- Normalize filesystem object aliases before and after expansion so generated grants preserve tighter access restrictions.
- Skip unavailable volumes and uninspectable generated children.
- Resolve deny globs into concrete file and directory paths using the Windows sandbox resolver.
## Testing
Add regression tests for read-only children and aliases, normalized root access, uninspectable children, and deny glob expansion alongside explicit denies.
GitOrigin-RevId: 3242e9a30d16581d6fe5b7b6e39e0e0f71ca6ef4
## Why
Writing hook input before draining output can deadlock when pipe buffers fill. Stdin writes also ran outside the hook timeout, allowing hooks that never read input to hang indefinitely.
## What changed
Write stdin concurrently with draining stdout and stderr, and apply the configured timeout to both input delivery and command completion. Preserve handling of broken pipes and distinct stdin and wait errors.
## Testing
Add a regression test that sends a large input to a hook that fills stdout and stderr without reading stdin. Verify that output drains and the hook reports its configured two-second timeout.
GitOrigin-RevId: 48cf33a2acfebbc5122a4522660d7a6b3fb1e9b0
## Why
With network access enabled, WSL interop can launch Windows processes outside the Linux filesystem sandbox, including using `wsl.exe` to reenter the distribution as root.
## What changed
- Mask WSL interop sockets in bubblewrap when filesystem write access is restricted, and hide inherited procfs when running without a fresh proc mount.
- Deny `AF_VSOCK` sockets and `io_uring` under restricted filesystem policies even when ordinary network access is allowed, closing the VM socket route to host services.
- Reject legacy Landlock configurations that combine restricted filesystem access with unrestricted networking when the WSL interop directory exists.
## Testing
Add unit coverage for mount masking and legacy policy rejection, plus WSL integration tests for Windows process launch, bind aliases, direct `/init` invocation, root reentry, and `--no-proc` isolation. WSL tests skip when host prerequisites are unavailable.
GitOrigin-RevId: 1e8b97ac43675807ee541ca614534d9b1febf410
## Why
Initial history replay and rollback before the first turn could clear the reasoning-effort baseline established by startup prewarm, preventing the first request from preserving that baseline when the selected effort changes.
## What changed
Preserve the reasoning-effort pin during history reconstruction. Clear it on rollback only after startup prewarm has been consumed. With reasoning-effort overrides enabled, a prewarmed `medium` baseline remains on the request while a subsequent `high` selection is sent as a `configuration_update`.
## Testing
Add regression coverage for prewarm completing before initial replay on resume and fork. Expand WebSocket tests to cover resumed and forked threads with overrides enabled and disabled, plus rollback before the first turn.
GitOrigin-RevId: 2e2d0e8a06b24f01b717458ca0576d63107c7697
## What changed
Probe the default Windows system config namespace at session start and record `codex.windows_system_config.namespace_squatting_probe` once per process when metrics are available.
When `config.toml` or `requirements.toml` exists, check the containing directories for expected ownership and broad standard-user mutation allow ACEs, including permissions inherited by child files. Report coarse results for missing files, expected directories, unexpected types or owners, mutation permissions, and check errors.
The probe is observational: it does not change configuration loading or enforce a trust decision.
GitOrigin-RevId: e1c79dca5548e16e83c7b84b44ab0de4e2643d0d
## Why
Loaded threads, including idle threads whose rollout files are still deferred, need to survive a managed daemon restart. Shutdown must also remain forceable when rollout I/O is blocked.
## What changed
- Add a hidden `--managed-daemon` option for app-server Unix socket transports. After active turns and admitted requests drain, save loaded persistent root threads before exiting.
- Skip ephemeral threads, non-root agent threads, and threads pending unload. Log persistence failures and continue saving other threads.
- Apply shutdown admission checks to thread and turn settings updates, thread deletion, and archiving.
- Keep force signals and daemon shutdown requests responsive during persistence. Return `AppServerExit::Forced` so executables can exit without waiting for runtime teardown.
## Testing
Add integration coverage for resuming active and idle threads after restart, forcing shutdown during active work, and forcing shutdown with a blocked rollout writer. Extend shutdown rejection coverage to deletion and settings updates.
GitOrigin-RevId: 4344e97d39f9f80c5d84c41c17300fd6aa99b4cc
## Why
Review evidence can fit its own limits while exceeding the reviewer context window once history, tools, output format, and reminders are included. Tool continuations can also push an existing review over budget.
## What changed
- Select evidence after tools and turn context are resolved, reserving request overhead and a 256-token margin. Omit optional evidence while preserving required sections, and reject required evidence that cannot fit.
- Keep incoming evidence pending until budget selection succeeds so cancellation and compaction cannot record unselected input in reusable history.
- Check the assembled prompt before sampling, including retries and tool continuations. Retry budget failures with summary compaction once per model step, and retire reviewers whose budgets remain exhausted. Avoid history-resetting token-budget compaction for this recovery.
- Record section costs after budget selection.
## Testing
Add unit and integration coverage for required-evidence rejection, optional-evidence omission, reminder overhead, cancellation safety, incoming-review and tool-continuation compaction, and reviewer retirement after ineffective compaction.
GitOrigin-RevId: 2f7ef191b3f46032d295ad363eb7debf9d7a9684
## Why
Failed environments were omitted from environment context, and `wait_for_environment` returned a generic failure or treated an already-failed environment as unselected. The model needs the failure reason to explain problems such as an empty repository.
## What changed
- Preserve the selected environment and error when startup or configuration fails.
- Include failed status and XML-escaped error details in environment context, and return the reason from `wait_for_environment` for failures detected before or during a wait.
- Limit error details to 256 bytes per environment and 512 bytes total in context, preserving UTF-8 boundaries. Limit wait failure details to 256 bytes.
## Testing
Add coverage for failures before a turn and while waiting, verifying that the model receives the reason and the turn completes. Add unit tests for XML escaping, incremental context updates, clearing errors on recovery, and UTF-8-safe truncation.
GitOrigin-RevId: 093ab9b447c7b3243fbb53b665180db95479f2a7
## Why
Recovering a suspended turn can append a duplicate reasoning effort update when history already ends with a matching trusted update but the runtime has not established its effort pin.
## What changed
Reuse a matching harness-authored `ConfigurationUpdate` at the end of history even when the runtime effort pin is unset.
## Testing
Add a regression test that suspends and recovers a turn without WebSocket prewarming, then verifies the recovered request contains exactly one effort update and retains `high` reasoning effort.
GitOrigin-RevId: 2e49073eb5e8d4fae39a8dc1d9d2b07f77802adb
## Why
Remote compaction already uses the normal Responses API, leaving the legacy `/responses/compact` runner unused.
## What changed
- Remove the legacy runner, endpoint client, request types, analytics variant, and obsolete test helpers and snapshots.
- Move tool-output trimming and its metadata tests into `compact_remote_history` for reuse by remote compaction v2.
- Consolidate retained-history filtering in the v2 implementation and remove redundant `RemoteCompactionV2` settings from tests.
## Testing
Extend the retained-history unit test to explicitly check that hook prompts survive compaction.
GitOrigin-RevId: 999bb1391aa22c3c281d6a96f98817d6c9e46ed9
## What changed
Remove the TUI command introduced in https://github.com/openai/codex/pull/11512 for granting sandbox read access to additional directories, along with its event handlers and non-elevated read-root grant helper.
## Testing
Enable the existing voice command popup snapshot test on Windows as well as macOS.
GitOrigin-RevId: 8a3a06b76749420101e34aa6587a8beab8eb5fdd
## What changed
- Route manual and automatic remote compaction through the v2 implementation.
- Retire the `remote_compaction_v2` feature toggle while continuing to advertise it in the model client beta features header.
- Migrate compaction tests from `/responses/compact` to streamed `/responses` requests with `compaction_trigger` items and remove legacy parity coverage.
## Testing
Add streamed compaction coverage for tool search output trimming and session base instruction token estimates. Update app-server tests to verify started/completed notifications for both local and streamed remote compaction.
GitOrigin-RevId: cea392a6166a9b116ff83708fd9126d5546476e4
## What changed
Move the synchronous review loop, outcome mapping, deadline helpers, and session pool into `codex-guardian-reviewer`. Core supplies adapters for evidence capture, authorization checks, session creation, and event publication through the new host interfaces.
Have `guardian-v2` initialize the reviewer pool and host separately, with the pool managing prewarming, session reuse, concurrent forks, invalidation, and shutdown.
## Testing
Extend the app-server reviewer lifecycle test to cover interrupted concurrent reviews as well as completed reviews, including resuming a reviewer after parent shutdown. Adapt core reviewer tests to use the pool and factory interfaces.
GitOrigin-RevId: 2d69bd20c169b20534764b98cbdb63f38564a530
## Why
Saved model selections can outlive their bundled catalog entries and still need migration guidance.
## What changed
- Remove `gpt-5.2` and `gpt-5.4-mini` from the bundled model catalog.
- Preserve migration from saved `gpt-5.4-mini` selections to `gpt-5.6-luna` when the old model is absent, while respecting catalog-provided migration metadata.
- Update tests and snapshots to use current models or explicit test fixtures, and remove the obsolete prompt fixture.
## Testing
Extend migration tests to cover catalog metadata precedence, saved selections, replacement model and reasoning-effort persistence, acknowledged prompts, and missing or hidden migration targets.
GitOrigin-RevId: 82beea504cc4a15699677c28a61ec0939af3069c
## Why
Switching models can send images from conversation history with `detail: original` to a model that does not support it.
## What changed
Normalize image details on request copies using the receiving model's capabilities. Downgrade `original` to `high` when unsupported, covering user messages and function and custom tool outputs. Preserve stored image details so switching back to a supporting model retains `original`, and continue stripping details for Responses Lite.
## Testing
Expand unit coverage across image detail values and model capabilities. Extend integration coverage for model switches within and between turns, text-only models, and resumed and forked threads, including assertions that persisted tool outputs remain unchanged.
GitOrigin-RevId: 2561651ea74f045e00f708ae53771a41dcd49c94
## Why
Replaying tool outputs under a different model can expand or shrink the history shown to the model if truncation uses the new model's budget.
## What changed
Save the originating history truncation budget on function and custom tool outputs and reuse it during replay, preserving existing tool-specific overrides. Include the existing 20% serialization allowance once, before converting byte budgets to tokens.
Rename the metadata field to `history_truncation_token_limit` while retaining `fallback_token_limit_override` as its serialized name for compatibility.
## Testing
Add regression coverage for resume and fork with different model budgets, including custom tool outputs. Extend unit coverage for existing overrides and byte-budget conversion.
GitOrigin-RevId: a62c8fd3e43198e2252b150fba72f022939c1070
## Why
History recording used the model captured at turn start, so switching models within a turn could apply stale image preparation and tool-output truncation rules.
## What changed
Pass explicit model metadata through conversation recording. Record model responses and tool results using their originating step's settings, and capture the current model for pending input recorded during startup interruption or task completion.
## Testing
Add regression coverage for tool results across a model switch, explicit model selection for plain and annotated history, and image input retained during interruption and completion. Extend model-switching tests to verify that prepared media and item identity survive switching back to a multimodal model in live, resumed, and forked threads.
GitOrigin-RevId: 5bc750c30c334acb9b802cb80fdea01daffa84a1
## Why
After a model change within a turn, tools could still use the turn's original model capabilities, output limits, and telemetry attribution.
## What changed
- Build each step's tool plan from its captured model settings.
- Use those settings for image support and detail, command and MCP output truncation, extension metadata, and tool telemetry.
- Preserve the issuing step's model for delayed command completion events while allowing later `write_stdin` calls to use their own step's output limits.
## Testing
Add regression coverage for model changes within a turn, including tool replanning, Code Mode execution, image handling, MCP tool and resource output limits, extension settings, and delayed command completion.
GitOrigin-RevId: ef991598e89a846889baf770a0c9e12e56886d5d
## What changed
- Let higher-priority providers in `features.network_proxy.credentials` displace lower-priority providers with overlapping `env` sources, including in profiles. Preserve inherited settings when providers swap sources, and compare environment names case-insensitively on Windows.
- Preserve provider definitions across ordered batch remaps and persist displaced-provider removals without overwriting unrelated sibling updates.
- Validate remapped providers using the credential broker's compilation rules before persisting writes, while allowing incomplete drafts and explicit deletions.
- Omit displaced providers from config origins and report writes overridden by another provider's source ownership.
- Build trusted credential-broker configuration before merging managed layers for project discovery, preserving remapped providers' environment protections.
## Testing
Add unit and config RPC regression tests for source precedence, source swaps, ordered edits, concurrent sibling updates, atomic rejection of invalid remaps, persistence, and override reporting. Add a macOS managed-config regression test for remapped provider bindings during project discovery.
GitOrigin-RevId: b7d402727acf4e5b0f25946db1cf18811f5e2679
## Why
Form and URL elicitations could keep waiting for a response and leave tool timeouts paused after cancellation. Remembered cancellations could also affect reused request IDs on a new connection.
## What changed
- Honor request cancellation and server cancellation notifications for all elicitations, without requiring user-verification support. Return `cancel` and release pending response routes and timeout pauses.
- Create a fresh elicitation service for each connection attempt, including initialization retries and session recovery, so cancellation state stays scoped to its connection.
## Testing
Add regression tests for form and URL elicitation cancellation, cleanup on service shutdown, and acceptance of previously cancelled request IDs after connection recovery.
GitOrigin-RevId: 388a8d72e55e89ad3d44d4739d8434485588aa1a
## What changed
Move assessment parsing and schema, model selection, review outcomes, retry handling, and rejection circuit breakers into `codex-guardian-reviewer`. Update core to use the extracted APIs while retaining session execution and decision enforcement in the host.
## Testing
Move existing assessment, circuit breaker, and retry tests into the new crate, including coverage for transient errors, cancellation, and deadlines.
GitOrigin-RevId: 96ec9989a0066acb012bf3bec0b9f7d8bf11a4ef
## Why
An error searching a single compressed rollout could abort the entire search and prevent results from other rollouts from being returned.
## What changed
Log a warning with the rollout path and error, skip that compressed rollout, and continue searching. Apply this behavior to both the compressed scan used alongside `ripgrep` and the fallback scan used when `ripgrep` is unavailable.
GitOrigin-RevId: 07a0d7109456a374244bb84f32cf18e60a69c206
## Why
The existing request duration metric starts at dispatch, leaving out time spent waiting before dispatch.
## What changed
Add `exec_server_request_total_duration_seconds` to measure time from decoded receipt until response enqueue or disconnection, including queueing. Record it alongside the existing duration metric with the same `method` and `result` labels for success, error, and disconnection paths. Preserve existing dispatch and queue timing semantics.
## Testing
Add coverage for successful requests, route errors, unknown methods, response delivery failures, and disconnection during execution. Extend admission-wait coverage to verify total timing includes queueing and route setup without double-counting, and assert each completion and duration is recorded once.
GitOrigin-RevId: ed0f719751b821b8f9f9556914b0f67dc6c5ad8f
## Why
Model switches within a turn can leave extension context using stale model metadata. Skill catalogs, context windows, and metric attribution need to match the model captured for each sampling step.
## What changed
- Pass captured model metadata and step-specific telemetry to world-state contributors.
- Use that metadata for skill catalog budgets and usage instructions, preserving configured budget overrides.
- Supply the captured model's usable context window to turn-context contributors, including when rebuilding context.
## Testing
Add regression coverage for model switches during skill discovery, catalog budgets and metric attribution, and extension context windows after `new_context` and retained-step context rebuilds.
GitOrigin-RevId: c51b40b739a380b9767a54ea8c8ee6de45bf01c8
## Why
Model changes during an active turn could leave context using the turn's initial model and settings, even after requests switched to the new model.
## What changed
- Build model, personality, approval, collaboration, app, plugin, and multi-agent instructions from captured step settings.
- Resolve multi-agent policy from the step's effective reasoning effort.
- Use the captured step's context window for full-context metadata and retain that step when rebuilding initial context for compaction.
## Testing
Add regression coverage for model switches updating context without duplicating updates, delegation policy following selected or model-default reasoning effort, and compaction preserving retained context after a model change.
GitOrigin-RevId: 3615c7a3f13f15fdcce9ad13d23311340e0aea88
## Why
Rendering the voice strip over sparkle stars can leave voice indicators with a star's color even when their glyphs are unchanged.
## What changed
Render the voice strip after effort ignition and before sparkle effects so stars skip its text and preserve its styles.
## Testing
Add a regression test comparing all nonblank composer cells, including their styles, across 80 sparkle frames while confirming stars still appear. Add a snapshot of the active voice composer.
GitOrigin-RevId: 48774408646c1281df3a6f7b1e12ab9d03a1babd
## Why
Cancelling a session after persistence opens but while MCP startup is still pending can leave its writer held, blocking a subsequent resume.
## What changed
Create `LiveThreadInitGuard` inside the persistence startup future so it protects the live thread while `tokio::join!` waits for other startup work. Dropping the startup future then schedules writer cleanup.
## Testing
Add a regression test that cancels a resume while MCP startup is blocked, verifies the local writer is released, and resumes the same thread successfully.
GitOrigin-RevId: 57d8624e783a7b993579a6386aea35ba845da0e3
## Why
Guardian request and section cost distributions need shared bucket boundaries across synchronous and asynchronous reviewers so their measurements align.
## What changed
- Add `histogram_with_boundaries` to session telemetry and extension metrics while preserving session attribution.
- Use shared request-token buckets up to 2,000,000 tokens and section-cost buckets up to 16,777,216 for both review paths.
## Testing
Extend telemetry tests to verify explicit bucket boundaries and sample counts, and Guardian integration coverage to check exported request and section metric bounds for both synchronous and asynchronous reviews.
GitOrigin-RevId: e34f6972b3418bac3b061939b62eeeccaec5a299
## Why
Session transitions need stack headroom to rebuild configuration and the chat widget. Tokio stores task outputs inline even when it boxes the task future, so returning a large `Config` adds stack pressure during task allocation.
## What changed
Box the configuration build result inside the runtime worker and unbox it after joining. Box the `App::run` future to keep the large event loop out of enclosing startup futures.
## Testing
Extend the production-stack TUI integration test to enter `/side`, verify the process remains running, and return to the saved conversation with Ctrl+C.
GitOrigin-RevId: 44975300bb837bf3645ff762cf7bcdc44a277f41
## Why
Diagnostic attachment reads assume the queued file path still exists and contains plain bytes. Compressed rollouts can therefore be omitted when only the logical `.jsonl` path is available, or attached as compressed data when a `.jsonl.zst` path is supplied.
## What changed
- Read rollout attachments through a bounded decoder that resolves plain or compressed representations without materializing a durable JSONL file.
- Use canonical `.jsonl` filenames for attachments and app-server report metadata, while preserving filename overrides.
- Apply size limits to decoded bytes and preserve JSONL prefix truncation.
## Testing
Add regression tests for compressed attachments, representation changes after queuing, plain-sibling preference, filename overrides, decoded size limits, truncation, nonregular files, and unrelated `.zst` attachments.
GitOrigin-RevId: b30f7dd08a741b0c99283460a1ce8933d2920ddf
## What changed
Add `ComposedContext::enforce_budget` to fit evidence within the input token allowance after reserving existing context. Preserve required content and message boundaries, reserve an omission notice, and return an error if required evidence cannot fit.
Carry retention policies through transcript rendering and composition. Protect user messages, protected messages, and the newest five tool entries. Remove oversized optional items first, then evict commentary, older tool evidence, and images in priority order. Add image admission support and record omitted content in truncation observations.
## Testing
Add tests for existing-context reservations, required-message preservation, image omission, framing costs, and protection of the newest five tool entries in both context profiles.
GitOrigin-RevId: 8a0f245bad82f562323b4533442385ddbcb6e541
## What changed
- Record per-section text bytes, estimated text tokens, image bytes, and image counts for synchronous reviews and asynchronous scoring without logging evidence payloads.
- Emit estimated request tokens through `codex.guardian.context.request_tokens`. Synchronous estimates include assembled history, instructions, tool definitions, and output format, and measure the full logical request before WebSocket delta generation. Asynchronous estimates cover the assembled input.
- Add shared context budgeting helpers, including conservative image token reservations independent of encoded payload size and model-aware input limit calculation.
## Testing
Add coverage for separate text and image accounting, image estimates independent of encoded size, and section estimates that bound delivered messages. Extend asynchronous scorer and app-server tests to verify cost metric emission.
GitOrigin-RevId: aed45ecd9c23706f88caa51f2a4f2c77872d3bdb
## Why
Cold rollout files can still have active writers. Compression must not replace their files while writes are pending, including background work that outlives the local thread store.
## What changed
- Share cross-process writer locks between the rollout compressor and local thread store. Skip busy threads and recheck file state under coordination before publishing compressed files.
- Retain writer ownership through recorder background I/O and compressed rollout materialization. Wait for writers to exit on shutdown or discard, and stop queued file work before deletion.
- Route rollout metadata patches through owned recorders, reject competing writers before updating SQLite, and serialize metadata appends with shutdown.
## Testing
Add regression coverage for compression with live and detached writers, parallel cold-file compression, publication locking, metadata ownership conflicts, updates to compressed rollouts, concurrent shutdown, and discarding deferred items.
GitOrigin-RevId: c52f4037eb2019e8095b554e29bf531af95aa4bd
## Why
Dormant MCP servers with usable cached tool catalogs prevented binding reuse, causing each model step to capture a new binding even when the catalog was unchanged.
## What changed
Track dormant catalog revisions alongside ready client revisions so bindings can be reused until the catalog changes or the server starts. Explicit server and plugin requirements still trigger binding capture and startup. Check cached catalog availability without cloning tool definitions.
## Testing
Extend the lazy-startup integration test to verify that two model steps share one binding while the server stays dormant, and that starting the server invalidates that binding.
GitOrigin-RevId: ef0469a85000f84a928946c24350e3bda1522fb5
Guard the `voice_args` array expansion to avoid unbound-variable errors
under `set -u` when packaging bundles without voice arguments. Preserve
quoted arguments for the `primary` bundle.
GitOrigin-RevId: 677a565d1678dc1e218cd9a98298aa6a26873643
## Why
Configured HTTP credential destinations need dummy credentials translated inside `CONNECT` and SOCKS5 tunnels, where credential interception previously detected only TLS.
## What changed
- Detect plaintext HTTP for configured credential destinations and proxy requests with URL-scoped credential substitution.
- Keep requests bound to the authorized tunnel destination, rejecting mismatched authorities and nested `CONNECT` requests.
- Support HTTP/2 clients and preserve HTTP upgrades, including `h2c`, and opaque traffic in full mode.
- Enforce limited-mode method restrictions on plaintext HTTP and reject upgrades and opaque traffic.
## Testing
Add end-to-end tests for both tunnel transports covering credential URL scope, request bodies, HTTP/2, destination mismatches, upgrades, lossless opaque forwarding, and limited-mode enforcement.
GitOrigin-RevId: 89262c996ea4ee156c628f9ed8e72b52504b2f6b
## Why
Applications need to deliver content from other agents, tools, or services with tool-level authority, without treating it as user input or granting authorization.
## What changed
- Export `ExternalMessage` for sync and async `run(...)` and `turn(...)`, accepting text or structured content with a tool name and optional namespace. Send it through `toolOutput` and require CLI 0.151.0 or newer.
- Support starting a turn or joining an active regular turn while preserving external content as function output in history. Keep external messages separate from user-input lists and `steer(...)`.
- Give turn handles independent subscriptions, replaying completed items and latest usage to joining handles. Release consumed transient events and clean up subscriptions on closure, failure, or cancellation.
- Document the authority boundary and add sync and async examples.
## Testing
Add coverage for wire representations, input validation, runtime compatibility, tool authority across resume, active-turn joins, and tool-output truncation. Add subscription tests for replay, concurrent consumers, early completion, cancellation, and cleanup.
GitOrigin-RevId: 6106327085fd9c4bd11b71e20b3d8e74738b8bb5
## Why
Python callers need control over response history loading and a way to override the service tier for one turn. These options also need runtime compatibility checks to prevent older CLIs from silently ignoring them.
## What changed
- Add `include_turns` to sync and async thread resume/fork methods. Omission preserves server defaults; `False` skips response history loading without changing model context.
- Add `turn_service_tier` and `source` to sync and async `run()` and `turn()`, and generate both methods together to keep their options aligned.
- Require CLI `0.151.0` or newer when sending the new options, with lazy schema checks for unversioned local builds.
- Pin the bundled runtime dependency to `0.153.4` and reject unsupported runtime versions during SDK packaging.
## Testing
Add coverage for option forwarding, history flag omission and inversion, runtime version checks, cached schema probing, and packaging compatibility. Extend app-server and installed SDK smoke tests to exercise the new options.
GitOrigin-RevId: 4bcc9cff687b0651e67852e7c080df7fadac6d76
## Why
GNU tar extraction and signing require writable copies of Bazel outputs.
## What changed
Recursively add owner write permission to the staged macOS voice runtime before creating the unsigned release archive.
GitOrigin-RevId: 1d415251b26303ac17cf886f35ae18277e9e6032
## What changed
Move TLS prefix detection into `brokered_tunnel` and carry explicit `BrokeredProtocols` requirements from the credential broker through HTTP `CONNECT` and SOCKS5 tunnel handling. Dispatch on `TunnelProtocol::Tls` or `TunnelProtocol::Opaque`, preserving the initial-read timeout, fragmented TLS header detection, and replay of all inspected bytes.
## Testing
Move the fragmented TLS regression test alongside the detector, retaining coverage for delayed header fragments and lossless prefix replay. Update broker and proxy assertions for the protocol-aware types.
GitOrigin-RevId: 6e5a92d8bdf6db5b59cff7f210c17b01d93ee23b
