## What changed
- Add the experimental `code_mode_buffered_exec` feature.
- When enabled, default code-mode `exec` calls without an explicit
`yield_time_ms` to 30 seconds instead of 10 seconds.
- Report the effective default in the model-visible `exec` tool description.
## Testing
- Verify that the generated tool description advertises the 30-second default
when the feature is enabled.
GitOrigin-RevId: 0eea4d0105137f09050ce31aad18c25c79df6b2a
## Why
Patch approval events can take longer than the default test event timeout.
## What changed
Wait up to 15 seconds for a patch approval request or turn completion in the
approval test helper, while preserving the existing event checks and assertions.
GitOrigin-RevId: 73a50538482b5a88ccf0be07f8fad21000e54699
## Why
Permission profiles defined by `requirements.toml` can be selected as active,
but their network configuration was not included when resolving the network
proxy specification for that selection.
## What changed
Merge managed permission profiles from `requirements.toml` with configured
profiles before looking up the active profile's network proxy settings.
## Testing
- Verify config loading retains the proxy settings of an active managed
profile.
- Verify `turn/start` accepts a managed profile with network settings.
GitOrigin-RevId: 3374f82a8fb0ce8f338fa2cea77c28ea2eb89636
## Why
System proxy discovery can block, and delegating fallback to each transport can repeat discovery or apply inconsistent environment proxy behavior.
## What changed
- Resolve unavailable system proxy decisions to an explicit environment proxy or direct route, including `NO_PROXY` settings and WebSocket-specific fallbacks.
- Add asynchronous system proxy resolution that uses cached decisions first and serializes blocking platform lookups on Windows and macOS.
- Preserve `NO_PROXY` handling for WebSocket connections, including HTTP and HTTPS proxies, and hash proxy cache keys on every platform.
## Testing
- Add coverage for explicit environment fallback, cached asynchronous resolution, and proxied or bypassed WebSocket connections.
GitOrigin-RevId: 3b7cf170dcfe639eec53c3c1514f92bfcf13e7e9
## What changed
- Add model-catalog approval message variants for `never` and `unless_trusted`.
- Select the catalog message that matches the active approval policy, while retaining the existing built-in text when that variant is absent.
- Treat an explicitly empty variant as an instruction to suppress the built-in approval text, consistent with `on_request` messages.
## Testing
- Cover variant selection, fallback and empty-message behavior, catalog deserialization, and the initial permissions message sent to the model.
GitOrigin-RevId: a0f8d41a08645f39b80093be53f200eeee18ca25
## Why
Remote compaction can process large histories. Repeatedly estimating and replacing the full history, and cloning it when tracing is disabled, adds avoidable CPU and memory overhead.
## What changed
- Estimate each history item's token count once, update the total as trailing tool outputs are rewritten, and replace history only after all rewrites are selected.
- Preserve unclamped token totals while calculating removed tokens so saturated estimates do not hide overflow.
- Snapshot compaction input history only when rollout tracing is enabled.
- Reuse the v2 request input instead of cloning it before adding the compaction trigger.
## Testing
- Cover enabled and disabled compaction trace contexts through `is_enabled()` assertions.
GitOrigin-RevId: 8de9c9704ba29532fba05430eb204bbae2e8bf83
## What changed
- Define and export skill metadata, policy, dependency, interface, and configuration rule types from `codex-skills`.
- Update core, plugin, and extension consumers to use the shared models while retaining compatibility re-exports from `codex-core-skills`.
- Share product-restriction handling between host and environment skill metadata.
## Testing
- Add unit coverage for default and explicit implicit-invocation and product policies.
GitOrigin-RevId: 9093469f8fff8e2d902dc9781235f64709827a87
## Why
Sandboxed process launch through the exec server was unsupported on Windows.
## What changed
- Add a shared native process launcher that selects the Windows sandbox session backend when required and otherwise launches PTY or pipe-based processes.
- Pass Windows permission, workspace-root, proxy, filesystem-override, and desktop settings through exec-server process preparation.
- Preserve piped stdin and inherited file descriptors across the shared launch path.
## Testing
- Verify that a remote Windows sandbox process accepts stdin while blocking a write under a read-only policy.
- Enable workspace-root sandbox tests on native Windows while continuing to skip environments that cannot emulate Windows sandbox semantics.
GitOrigin-RevId: 6719ccbe87ad0d5a57af56503e3f23f894c60a71
## What changed
- Add `iconUrlDark`, `distributionChannel`, `installUrl`, and
`pluginDisplayNames` to the experimental `app/read` connector metadata.
- Populate plugin display names from enabled plugins without starting their MCP
servers.
- Accept both dark-icon field spellings from the batch API and keep the new
upstream fields optional for compatibility with older responses.
## Testing
- Cover the serialized response fields, legacy batch responses, and plugin
display-name lookup without MCP startup.
GitOrigin-RevId: 68601a104de839c0f4f4621d33b0e4d08ad88c64
## What changed
- Render the first line of a completed hook warning in the hook header using
`says:`, with any remaining lines indented below it.
- Keep the existing hook header when no warning is present and avoid rendering
warning entries a second time in the output body.
- Update hook event snapshots and the multiline warning test for the new layout.
GitOrigin-RevId: 291b556daf3b6fb64a99b090a6df1a222a18a736
## What changed
- Remove the `spawn_agents_on_csv` and `report_agent_job_result` tools and their agent-job runtime and state models.
- Drop the legacy `agent_jobs` and `agent_job_items` tables during state database migration.
- Keep `features.enable_fanout` and `agents.job_max_runtime_seconds` accepted as no-op compatibility settings while omitting them from the generated configuration schema.
## Testing
- Verify upgrades remove both legacy agent-job tables.
- Verify the removed feature and configuration keys still parse without taking effect.
GitOrigin-RevId: 8cc3337da78c67162229f02f40a747f503542646
## Why
`SqliteConfig` stores its home as an `AbsolutePathBuf`, but its test constructor
previously accepted a `PathBuf` and checked the absolute-path invariant at
runtime.
## What changed
- Make `SqliteConfig::new_for_testing` accept an `AbsolutePathBuf` directly,
removing its fallible conversion and `expect`.
- Update SQLite test setup to convert temporary directory paths with
`PathExt::abs` at each call site.
GitOrigin-RevId: 93585b9aea805e2449b3465ac20eeb39417ed555
## What changed
- Run the bubblewrap `/proc` mount probe with the minimal read-only filesystem policy instead of the command's filesystem policy and working directory.
- Preserve the requested network namespace mode during the probe.
- Add coverage that the probe mounts a temporary root and `/proc` without binding the full filesystem.
GitOrigin-RevId: e3bc4352f510980aeebe81dedf3f9211947982ed
## What changed
- Make the MCP connection manager's event sender optional.
- Skip startup notifications when a caller has no session event stream.
- Decline interactive MCP elicitations in that mode while allowing the
underlying operation to continue.
- Update threadless resource reads, status snapshots, and connector discovery
to use the channel-free path.
## Testing
- Extend the app-server MCP resource test to verify that a threadless resource
read declines an elicitation and still returns the resource contents.
GitOrigin-RevId: 65fea152adbc02a78e09a40cae4134b6dccffd34
## What changed
- Add a shared local thread-store resolver that follows `history_base` links into ordered, bounded rollout segments, including archived ancestors.
- Support resolving a lineage at an explicit `HistoryPosition`.
- Reject malformed lineages with cycles, missing or mismatched source rollouts, non-paginated sources, or invalid cutoff bounds.
## Testing
- Add unit coverage for nested and archived lineages, explicit history positions, cycles, missing sources, and out-of-bounds offsets.
GitOrigin-RevId: a9063ad87e81d9865bd17fc588ea5d8e9ac74c0a
## What changed
- Assert truncated audio history through `ContextManager::raw_items()`.
- Initialize `additional_context_limit` in the quoted Windows hook command test.
GitOrigin-RevId: 43c19804e558b7a11fc17590506ec6ec7f3a9897
## What changed
- Change `ReviewDecision::Denied` to carry a rejection string and update the generated protocol schemas.
- Preserve specific rejection reasons through command, patch, network, MCP, delegated, and automatic approval flows so tool results can return them to the model.
- Distinguish invalid approval responses from user declines and truncate oversized rejection messages before adding them to model context.
## Testing
- Add coverage for denial serialization, invalid app-server approval responses, dropped automatic reviews, network rejection propagation, and rejection-message truncation.
GitOrigin-RevId: d48a55643aa03c7ecfa08a8e4b669913eec5464a
## Why
Mid-turn auto-compaction queued `SessionStart` hooks but continued sampling
before running them. This delayed hook-provided context and ignored requests to
stop the continuation.
## What changed
Drain pending session-start hooks immediately after successful mid-turn
auto-compaction. End the turn when a hook requests a stop; otherwise include
its additional context in the next sampling request.
## Testing
Added coverage for repeated compactions in one turn, context delivery without
leaking hooks into the next user turn, and stop requests that block sampling.
GitOrigin-RevId: c57708a792fb47d98d95c38d7d91bcd9f235be84
## What changed
- Add `additionalContextLimit` to command hook configuration for events that can emit `additionalContext`.
- Apply the limit independently to each hook's context before it is sent to the model. Unset values retain the 2,500-token default, while `0` disables spilling for that hook.
- Expose configured limits through app-server hook and requirements responses, and show them in the TUI hooks browser.
## Testing
- Cover JSON and TOML parsing, discovery and hashing, per-hook spilling behavior, app-server responses, and TUI rendering.
GitOrigin-RevId: 55ddc2dbbdf9067dceee3e1968c03617bbf6bcea
## Why
An inherited `FILE_DELETE_CHILD` grant can make a write root look stale, but
`SET_ACCESS` cannot replace an ACE inherited from an ancestor. Treating that
grant as explicit causes unnecessary ACL refresh attempts that cannot converge.
## What changed
- Add explicit-ACE filtering to the Windows DACL permission checks.
- Refresh a write root only when `FILE_DELETE_CHILD` is present in an explicit
allow ACE, while retaining effective-permission checks for required rights.
- Ignore inherited stale rights when deciding whether `SET_ACCESS` must repair
an allow ACE.
## Testing
Add a Windows regression test covering a write root that inherits
`FILE_DELETE_CHILD`, verifying that repeated refresh checks leave its explicit
write ACE unchanged.
GitOrigin-RevId: d0df9429efcf299da3ff3c1bce92942684803293
## Why
Cloning a `ContextManager` deep-copies every `ResponseItem`, even when callers only need a read-only history snapshot.
## What changed
- Store history items in an `Arc<Vec<ResponseItem>>` so cloned snapshots share their items until one is mutated.
- Reuse an owned vector when consuming an unshared snapshot and clone it only when other snapshots still reference it.
- Estimate session token counts from a snapshot after releasing the session-state lock.
## Testing
Added a regression test that verifies cloned histories share their backing items before mutation and diverge without affecting each other afterward.
GitOrigin-RevId: 24605888209ae9ff78cadb04d9c36463f7e2dbe0
## What changed
- Point the default Codex Apps MCP server at `ps/mcp` instead of the legacy Apps endpoint for both `backend-api` and `api/codex` base URLs.
- Use the same server configuration for Codex Apps and the hosted plugin runtime.
- Update MCP integration tests and test-server routing to expect the plugin-service path.
GitOrigin-RevId: 939f20dcff67ba6f79c11b328bfc624b25e3aac2
## What changed
- Add personality instruction variables for the GPT-5.6 model variants.
- Add the GPT-5.5 availability notice.
- Remove the `auto_review` and `permissions` message fields and the legacy `supports_reasoning_summary_parameter` flag from the bundled catalog.
GitOrigin-RevId: 426b0735a78232d2f658e0124db7c45889d08286
## Why
Paginated threads keep metadata updates in SQLite, while their rollout retains the initial `memory_mode`. Reconciliation could therefore overwrite the current setting with a stale value, and memory processing excluded these threads entirely.
## What changed
- Include paginated threads in stage 1 memory job selection and global memory output lookup.
- Preserve the SQLite `memory_mode` when reconciling or backfilling an existing paginated thread, while continuing to seed missing rows and restore legacy threads from rollouts.
- Omit `memory_mode` alongside Git metadata when flushing resumed paginated history.
## Testing
Added coverage for preserving disabled memory mode during reconciliation and backfill, selecting eligible paginated threads, and omitting initial metadata on paginated resume.
GitOrigin-RevId: 2a6e16068e69680728757fbec27aeefae45b8110
## Why
Audio attachments could be dropped from tool outputs and reconstructed thread history, while inline base64 payload size could overstate their model context cost.
## What changed
- Preserve remote and local audio attachments in legacy user-message events and thread history.
- Keep audio from function and custom tool outputs when the model supports it and the output fits the truncation budget; otherwise replace it with an omission marker.
- Estimate inline audio cost from its duration, with a serialized-size fallback for malformed or unsupported data URLs, so compaction and truncation use model-visible cost instead of raw base64 size.
- Replace audio in messages and tool outputs with an explanatory placeholder for models without audio input support.
## Testing
Add coverage for history round trips, dynamic tool responses, modality filtering, duration-based context estimates, output-budget truncation, and remote compaction.
GitOrigin-RevId: 5ed4628850a668f8383b7cd73db71b144cc02cd1
## What changed
- Point the packaged `rg` manifest at ripgrep 15.2.0 archives for all
supported macOS, Linux, and Windows targets.
- Update each archive's size, SHA-256 digest, and executable path.
GitOrigin-RevId: 6dcfa4fe40597a7591b8413e330d27e6b0b86627
## What changed
- Classify `multi_agent_v2` as stable while keeping it disabled by default.
- Exclude structured `multi_agent_v2` configuration from under-development feature warnings.
## Testing
- Update the warning-event test to verify that only the remaining under-development feature is reported.
GitOrigin-RevId: 12cb661dde9bc7b5aeae2f83476cb5ff75e0c959
## Why
Paginated rollouts contain only the initial Git metadata tuple. Reusing that
tuple during later reads, reconciliation, or resume can overwrite newer SQLite
values and undo explicit field clears.
## What changed
- Treat SQLite as authoritative for Git metadata once a thread is paginated.
- Apply paginated Git metadata patches directly to SQLite without appending
compatibility metadata to the rollout.
- Preserve explicit null values when reading by rollout path or reconciling the
initial session metadata, and ignore the initial Git tuple during resume.
## Testing
Added coverage for updating and clearing paginated Git metadata, preserving the
result through reconciliation and rollout-path reads, and avoiding stale Git
metadata updates on resume.
GitOrigin-RevId: 4d3e68cb24965b93d337d923092a98bac9b352b5
## Why
Building a Responses WebSocket request copied the model, instructions, input
history, tools, and other request settings into a second payload before
serialization.
## What changed
- Make `ResponseCreateWsRequest` borrow shared fields from
`ResponsesApiRequest`.
- Keep only WebSocket-specific fields owned, and allocate a separate input
vector only for incremental continuations.
- Restore response item IDs after preparing a full request so the request
retained for continuation matching keeps its original state.
## Testing
Update the serialization test to verify that the borrowed WebSocket request
produces the same wire payload as the Responses API request plus its
WebSocket-specific fields.
GitOrigin-RevId: b63d746b7f5164b5204ab95b057c914a4aeb71e1
## What changed
Remove the fallback that replaced image content in the latest tool output with
`Invalid image` and retried the model request. Invalid-image responses now use
the existing bad-request error path without modifying conversation history.
GitOrigin-RevId: f4ea782f16c5451feb261694caa73b77f24c3e4e
## What changed
- Tell the model that generated images are already displayed to the user and should not be repeated as Markdown images or file links.
- Keep generated-image artifact paths and output hints owned by the image generation extension.
- Centralize the availability check for the image generation tool across feature, provider, namespace-tool, model modality, and authentication requirements.
## Testing
- Verify artifact paths sanitize session and call identifiers.
- Verify the model-facing output hint reports that the generated image is already displayed.
GitOrigin-RevId: 6c865b6a96d6a71bb2c33246d1518a00e7c8b840
## Why
Requests longer than 10 seconds were grouped into the overflow bucket, limiting visibility into long-running request latency.
## What changed
Add second-based duration histogram boundaries at 12, 15, 20, 30, 60, and 120 seconds.
## Testing
Extend the duration histogram test to cover every new boundary and a 121-second overflow value.
GitOrigin-RevId: 063a5d397dbd37d3b3940c7a1632b7b479e874a4
## Why
Pending Guardian review state could outlive an interrupted or completed turn,
leaving stale review details visible during a following goal turn.
## What changed
- Clear pending Guardian reviews during both turn completion and finalization.
- Restore an active MCP startup header after cleanup, falling back to `Working`
when no MCP server is starting.
## Testing
Added coverage for interrupted and completed turns, including goal continuation
and overlapping MCP startup status.
GitOrigin-RevId: a15ed81a5d6005e9294089f4424a8dedbc6e9cb5
## Why
The TUI already receives skill metadata from the app-server protocol, but converted it into the core-skills model before using it.
## What changed
- Use `codex_app_server_protocol::SkillMetadata` throughout skill toggles, mentions, and input submission.
- Remove the protocol-to-core conversion and the TUI's `codex-core-skills` dependency.
GitOrigin-RevId: 43d43d74bfb67fc21ef11a0ca0592aa44b89a122
## What changed
Add a borrowed `Line` conversion and use it when rendering and measuring
hyperlink-aware history and pager content. This preserves line styling and
alignment while avoiding clones of the underlying span strings.
GitOrigin-RevId: c9cd6ff5ec63cd29719ea2ae51acede7364f2bd8
## What changed
- Add one-shot composer and status-line transitions when the active reasoning effort changes to Max or Ultra.
- Keep a tier-specific prompt accent after the transition, including a distinct Ultra prompt glyph.
- Preserve drafts and footer priority, avoid replaying effects for restored or unchanged settings, and disable motion when animations or sufficient terminal color support are unavailable.
## Testing
- Add unit and snapshot coverage for animation frames, prompt accents, status-line transitions, narrow layouts, terminal palettes, restored sessions, and reduced-motion behavior.
GitOrigin-RevId: 6da1d501d228d0cf1144b3d211d5d1304a06f447
## What changed
- Resolve the turn ID for restored token usage before sending resume and fork responses, avoiding a clone of the full thread history.
- Add `ThreadHistoryBuilder::active_turn_id` so attribution can inspect the active turn without materializing its items.
- Stop history reconstruction at the latest persisted `TokenCount` and verify that a later tail turn does not receive the replayed usage.
GitOrigin-RevId: 69c4a9fe04770d1a938470afc933431e61670094
## Why
Large command streams, including commands that never emit a newline, can make an active TUI cell grow without bound.
## What changed
- Retain live output in full up to 1 MiB, then preserve the first and last 50 completed lines plus the current partial line.
- Bound individual long lines while preserving UTF-8 boundaries, visible head and tail content, and ANSI rendering around omission markers.
- Show omitted line and byte counts in previews and transcripts. Final command output still replaces the bounded live preview when execution completes.
## Testing
Added unit and snapshot coverage for chunk boundaries, split CRLF sequences, newline-free and UTF-8 output, truncated ANSI sequences, retained-output limits, and preview and transcript rendering.
GitOrigin-RevId: 3cf3cc862db686f20ac09b1f6e7f9dc6d264e6eb
## What changed
- Encapsulate aggregated command output behind APIs for total and retained line
counts, reversible preview iteration, and full transcript iteration.
- Build truncated command previews from the retained head and tail lines while
preserving the logical omitted-line count.
- Cover head/tail rendering for long streamed agent and user-shell output.
GitOrigin-RevId: 54b23a19e4639e24c5e4778082edbbc6484ecf80
## Why
During TUI startup, `hooks/list` holds the global config queue while an
uncached `model/list` request may also take time. Waiting for
`configRequirements/read` before submitting the model request prevents these
operations from overlapping.
## What changed
Submit `model/list` and `configRequirements/read` together during bootstrap so
the model fetch can overlap the config requests. Preserve the request-specific
error context and update the session state after both requests succeed.
GitOrigin-RevId: 69a866e6d2fd4798b914dd000456194f49fcfef6
## Why
Sizing, rendering, and cursor placement can query the same chat widget layout
multiple times in one frame, repeatedly measuring active transcript cells.
## What changed
- Build one chat widget renderable tree per frame and reuse it for sizing,
rendering, and cursor placement.
- Cache each flex child's desired height by width for the lifetime of that tree.
- Reuse the bottom pane's renderable directly instead of forwarding each
renderable operation through a wrapper.
## Testing
- Verify flex layouts measure a child once across frame passes and remeasure it
when the width changes.
- Verify a chat widget frame measures its active transcript cell once.
GitOrigin-RevId: 5ad1a6711f4011c699b5d002b13dc3319cb4db8e
## What changed
- Move lifecycle and approval payloads into either the interrupt queue or the
immediate handler instead of cloning them for both branches.
- Borrow MCP elicitation request metadata while constructing the corresponding
form, and consume apply-patch changes after collecting notification paths.
## Testing
- Add a history replay test confirming deferred MCP start and completion events
remain FIFO after streaming finishes.
GitOrigin-RevId: 84ed34f6ff173463a1da016dc140405392c0ba45
## What changed
- Record when committed streaming content contains an inline visualization directive and keep using the canonical full-source renderer afterward.
- Restore directive state during full recomputes and clear it when the streaming renderer is reset.
- Cover directive tracking with and without visualization context, including a raw-to-rich render mode switch.
GitOrigin-RevId: 17786851d6dad6805240b79072fcbd0bb441874a
## What changed
- Remove unused APIs and test helpers across the Rust workspace.
- Route Linux sandbox test setup through the shared exec request path.
- Drop dependencies that became unnecessary after the cleanup.
GitOrigin-RevId: 48fd9b2b1575bad7add0819786b4d8ec88d8486f
## Why
The first session metadata record defines a rollout's history mode. If that
record uses an unsupported mode, skipping it as a parse error can cause later
metadata copied from fork history to be treated as canonical.
## What changed
- Validate session metadata history modes before deserializing rollout lines,
while continuing to tolerate unknown modes in copied fork history.
- Simplify thread status registration to pass only the thread ID instead of a
fully materialized `Thread`.
GitOrigin-RevId: 9de4313d4a76b279d32bb49b2016a4f5d07507c1
## Why
zsh prints tied scalar/array exports such as `PATH` and `path` with `-T`
metadata. The snapshot filter rejected that syntax, so commands available only
through the configured `PATH` could not be resolved from a restored snapshot.
## What changed
- Recognize writable tied exports and retain them in zsh shell snapshots.
- Skip readonly exports so restoring a snapshot does not make variables such as
`PATH` immutable.
## Testing
Added macOS coverage for restoring a tied `PATH`, handling entries with spaces,
deduplicating path entries, excluding readonly tied exports, and resolving a
command available only through the restored snapshot.
GitOrigin-RevId: 297cfae3368230c7f6fe5ef84f0a10b35a2c802c
## What changed
- On session startup, remove exact `allow` entries from `rules/default.rules`
for command prefixes that Codex no longer suggests as policy amendments.
- Record the migration in `.sandbox_migration` so it runs only once, preserving
rules created after the migration.
- Skip the migration when user and project exec policy rules are ignored.
- Expand the protected prefix list across shells, interpreters, package runners,
and destructive or privilege-related commands.
## Testing
- Cover selective removal, case-insensitive matching, one-time behavior, and the
startup path with ignored policy rules.
GitOrigin-RevId: a0c60e3f82b9630e621fd034b40462e3ab775102
## What changed
- Skip loaded-subagent backfills for fresh and forked threads, which cannot have pre-existing descendants.
- Keep the backfill after resuming a thread so agent navigation is populated immediately.
- Reuse thread status collected during backfill when opening the agent picker, avoiding duplicate `thread/read` requests while preserving liveness from live event channels.
## Testing
- Add a WebSocket-backed session lifecycle test that records app-server requests across fork, fresh-session, resume, and agent-picker flows.
GitOrigin-RevId: 604742b3cb495da12c97908d7a0ffa4ebc6a6da3
## Why
The transcript overlay cached every committed cell's height. Cells whose content changes after insertion, such as refreshed status output or newly available visualizations, could outgrow that cached height and be clipped.
## What changed
- Let history cells declare whether their transcript height is stable, and remeasure dynamic cells on later overlay renders.
- Preserve cached heights for stable cells when appending a committed cell instead of rebuilding all renderables.
## Testing
- Cover status output growing after a rate-limit refresh.
- Cover a visualization link replacing its unavailable placeholder.
- Verify that appending a cell preserves cached measurements for stable cells.
GitOrigin-RevId: 03e7166cb94d7353c9b89bff8abb3b7ccb84a6bb
