## What changed
- Expose `send_user_message_async` to root agents whenever the selected model advertises support for it.
- Retain `send_async_message` as a removed compatibility flag so existing configuration is accepted without controlling tool availability.
- Update the async message test to cover model-driven tool exposure without enabling the legacy flag.
GitOrigin-RevId: 558b427c2b203a057dd7c836e0ec702ae6134346
## Why
Bedrock sessions that use the AWS SDK credential chain need a way to recover when credentials expire during a request.
## What changed
- Add `aws.auth_refresh` provider configuration with an `aws` command, arguments, and a configurable timeout.
- Run the command for refreshable Bedrock authentication failures, reload the SDK credentials, re-sign the request, and retry it.
- Share refresh state across matching provider configurations so concurrent failures invoke the command only once. Bearer tokens, command auth, and static environment credentials do not use this recovery path.
## Testing
- Add coverage for configuration validation, refreshable error classification, concurrent refresh sharing, and an end-to-end retry signed with refreshed credentials.
GitOrigin-RevId: 0302fe3aabdbc1097e7bd62a74d407ba38a3cc57
## What changed
- Add environment-provided MCP restrictions for configured and plugin-provided servers.
- Disable attachment-scoped servers while their environment configuration is pending, failed, or unselected, while preserving selected-plugin access and the controller-owned Apps server.
- Apply the resolved policy to runtime startup, model tool exposure, telemetry, OAuth flows, and skill dependency installation.
## Testing
- Cover policy filtering for configured and plugin servers, environment state transitions, Apps ownership, and managed OAuth rejection.
GitOrigin-RevId: 7ca5d4dda14068e758497f88835d5cba685e159e
## What changed
- Execute hook-triggered MCP calls through the runtime's latest connection set instead of preparing a separate catalog-bound call.
- Keep hook calls from waiting for server startup or reconnecting, and reject disconnected servers immediately.
- Apply the shorter of the hook-requested timeout and the server's configured tool timeout.
## Testing
- Update connection manager tests to cover non-waiting calls and disabled tool filters.
GitOrigin-RevId: 3166b2d1e26d9ae2b8b3a648b4d0897cc260eb93
## What changed
- Add `send_user_message_async` for root agents when the async-message feature is enabled and the model advertises tool support.
- Emit the supplied text as an asynchronous agent message and immediately return an accepted result so the turn can continue.
- Keep the user-visible update out of the model's input context.
## Testing
- Add an integration test covering tool registration, item events, continued turn execution, and model context.
GitOrigin-RevId: 13d5168cb8cd6742db17bb74a15c910994e9c98b
## What changed
- Add an optional `delivery` field to agent message events and app-server
`agentMessage` items, with `"async"` identifying a user-visible message sent
without ending the current turn.
- Preserve the marker through legacy event conversion, thread history
materialization, replay, and generated JSON and TypeScript schemas.
- Require delivery of async agent message completion notifications on the
in-process app-server transport.
## Testing
- Cover preservation of async delivery metadata in thread history.
- Cover delivery-required classification for async agent messages.
GitOrigin-RevId: f92ba25de1293ee271404badfdf0d117d6530329
## Why
An unfamiliar executable can ignore its arguments, so trust in an apparent
inner command must not implicitly trust the executable that runs it.
## What changed
- Evaluate unfamiliar shell executables alongside their parsed commands when
determining unified exec approval requirements. Inner commands can add
restrictions, but cannot grant trust to the executable.
- Include the executable in reusable approval keys so approval for one custom
shell does not apply to another.
- Parse literal PowerShell scripts without launching the requested executable,
preserving command policy checks before approval.
## Testing
Add cross-platform coverage for spoofed shell paths, allowed and forbidden
inner commands, explicit custom-shell approval, and session approval isolation.
GitOrigin-RevId: 0dd2c7e9a2ac30965ef5fe5de1a8d2968bb5f9d0
## What changed
- Carry selected plugin identities from MCP discovery into per-turn extension data, limited to capability roots that are ready for the turn.
- Annotate executor skill catalog entries with the matching plugin ID and `user` scope.
- Include that attribution in `skill_invocation` analytics for both explicit and implicit executor skill invocations.
## Testing
- Extend the executor skills app-server test to verify plugin ID, skill scope, and invocation type for explicit and implicit analytics events.
GitOrigin-RevId: 9345600ae29bcb8eb894fffd2d6f7b05f22ebf74
## Why
Agent roles should customize a child agent without expanding the authority or changing the provider configuration inherited from its parent session.
## What changed
- Apply only supported role overrides for model behavior, developer instructions, personality, service tier, and capability reductions.
- Preserve parent-owned permissions, model providers, endpoints, MCP servers, notifications, and other unrestricted configuration.
- Keep managed feature requirements effective when a role disables capabilities, and reject symlinked user role files.
- Use the same bounded role application path for both multi-agent implementations and resumed agents.
## Testing
Add coverage for authority preservation, managed feature requirements, symlink rejection, provider inheritance, and provider routing after resume.
GitOrigin-RevId: c528d615b691f9c02bfbc21154d514ea07743010
## What changed
- Execute `mcp_tool` hook handlers through the session's shared MCP runtime, including managed hook configurations.
- Restrict hook calls to already-connected, cataloged, and policy-allowed tools. Unavailable servers fail immediately without starting or reconnecting them.
- Pass session metadata to hook tools, cap hook timeouts by the server timeout, and invoke hooks without model-tool approval or recursive hook dispatch.
- Record handler type and execution mode in hook telemetry, and mark memory as externally influenced after successful or blocking MCP hooks when configured.
## Testing
- Added end-to-end coverage for prompt interpolation, hidden tools, blocking decisions, unavailable servers, thread metadata, approval bypass, and memory-mode handling.
- Added MCP runtime coverage for startup readiness, tool filters, catalog revisions, and managed connector policy.
GitOrigin-RevId: 1647cb801ebc5fa385c7bda9d9c266a389c611b8
## Why
When a local proxy request disconnects before network approval completes, the
owning tool call needs a model-visible explanation instead of remaining tied to
the abandoned request.
## What changed
- Track disconnect timing while plain HTTP and CONNECT requests await policy
decisions.
- Cancel the owning execution and report how long the request waited when it
disconnects before approval completes.
- Preserve an explicit approval outcome when disconnect cleanup runs afterward.
## Testing
Added unit coverage for disconnect tracking and outcome precedence, plus
end-to-end coverage for plain HTTP and CONNECT requests.
GitOrigin-RevId: b354b29bbe86f38e252fcaf529541f177480136b
## What changed
- Add a public `snapshot_script` helper to `codex-shell-command` for Bash,
Zsh, POSIX shell, and PowerShell, returning `None` for Command Prompt.
- Update `codex-core` and its shell snapshot tests to use the shared helper
without changing snapshot behavior.
GitOrigin-RevId: 779b9dc964635b717922ea9f5646b94c4dbb0da1
## Why
Thread settings could replace an owner-provided environment configuration with
`FromThread`, allowing the thread to take ownership of that configuration.
## What changed
Reject environment settings updates that change an existing owner-provided
configuration to `FromThread`.
## Testing
Cover preview and turn settings updates for pending, ready, and failed
owner-provided environments.
GitOrigin-RevId: c516954819447ceb29e2ff9f29fe781ca101ae54
## What changed
- Let model providers identify recoverable authentication errors and refresh their own authentication state before falling back to `AuthManager` recovery.
- Apply provider recovery to both HTTP and WebSocket response requests, with at most one provider-owned recovery attempt per request.
- Preserve existing `401 Unauthorized` handling as the default for providers that do not customize recovery.
## Testing
- Add coverage for successful and failed provider recovery, including the single-attempt bound and preservation of the original unauthorized error.
GitOrigin-RevId: 9a737e3be96354699586f97dd88e544a45082969
## What changed
- Add a dedicated developer policy for Guardian reviews of Node REPL browser and computer-use actions.
- Inject the policy once before the first eligible Node REPL approval and retain it when the reviewer session is reused.
- Include Node REPL policy eligibility in the reviewer reuse key so sessions are not reused across incompatible model settings.
- Keep the approval request itself on the generic planned-action format.
## Testing
- Cover first-action injection, reuse after a shell review, repeated Node REPL reviews, and ineligible parent models through the production approval path.
GitOrigin-RevId: bfce190ff8599788aa85e388cc6a959d27fb489c
## Why
A stored permission grant may allow access to a parent while explicitly denying
a child path. A later request for that child must not be treated as already
approved by the broader parent grant.
## What changed
- Compare materialized permission profiles without intersecting away denied or
reopened paths before deciding that a request is preapproved.
- Execute preapproved commands with the stored grant itself so its denied paths
remain enforced.
- Fail closed when permission profiles cannot be materialized.
## Testing
Added unit and integration coverage for turn and session grants across
`exec_command`, `shell_command`, and `apply_patch`, including approval-disabled
and `Never` approval modes.
GitOrigin-RevId: 5455880328a89c7958f859c7ce87805dff9704fb
## What changed
- Return retained unified-exec output as one contiguous byte buffer.
- Use the buffer directly for sandbox-denial checks, removing intermediate
chunk aggregation and an unnecessary owned string conversion.
- Update the output-retention test to assert the flattened snapshot.
GitOrigin-RevId: 84eacee0166b86d3cb2d6a37e96125977fad0267
## What changed
- Reconnect frameless bidirectional WebRTC sideband sockets after unexpected transport loss with capped exponential backoff.
- Preserve pending text or handoff output and bounded transcript state across reconnects.
- Treat `404 Not Found` and `410 Gone` handshake responses as terminal session completion, and preserve HTTP status information from failed WebSocket handshakes.
- Bound Realtime delegation input and transcript fields while retaining the most relevant portion of each.
## Testing
- Add unit coverage for reconnect backoff, terminal status handling, transport error classification, transcript continuity and truncation, and pending outbound messages.
- Add an end-to-end WebRTC conversation test covering an unclean disconnect, reconnect, transcript reconciliation, outbound progress during sustained inbound traffic, and terminal session completion.
GitOrigin-RevId: 544fd7e542c3434d0bb42cfbcc7143604419a037
## What changed
- Add an optional `connectorId` to `mcpServer/resource/read`.
- Forward the selected connector through `x-codex-turn-metadata` for threaded and threadless resource reads, restricting hosted app resources to that connector.
- Update the generated protocol schemas and app-server documentation.
GitOrigin-RevId: 5cc94d703abbbb6086665910c1c9ed5dc7075210
## Why
`node_repl` screenshots are nested tool results and do not appear in the parent
conversation history, so Guardian v2 transcript image collection could miss
them.
## What changed
- Allow Guardian v2 to enable bounded, thread-scoped `node_repl` image capture
when transcript images are configured.
- Add retained `node_repl` screenshots to review requests when transcript
sources include tool outputs, while deduplicating images and keeping them out
of the parent turn input.
## Testing
- Add an integration test covering screenshot capture without the synchronous
Guardian transcript feature flags and verifying that screenshots stay out of
the parent request.
GitOrigin-RevId: 5f70c5f2e80389aa8b2a286c3b7f9b1104865486
## What changed
- Pass metrics bound to the active turn's model to approval-review contributors.
- Add counter support to `ExtensionMetrics` while preserving host session attribution.
- Record Guardian v2 tool-call score lag and count stale-score fallbacks with the `score_lag` reason.
## Testing
- Cover counter forwarding and session metadata tags.
- Verify Guardian v2 metric samples for current and stale scores.
GitOrigin-RevId: 0ec34736d29d2e37ed217a27e773b4f6230ae38a
## Why
Compaction can remove the tool-call events that associate an MCP app widget with
the tool, account, and URI needed to authorize later resource reads.
## What changed
- Store a bounded MCP resource-origin checkpoint with each compacted rollout.
- Restore the checkpoint when resuming a session so existing widgets remain
readable after compaction and restart.
- Discard saved origins when rollback rewrites history across the compaction
boundary, and reject oversized or invalid checkpoints during restoration.
## Testing
- Extend the app-server MCP resource-origin test to compact a paginated thread,
continue the conversation, and read the original widgets before and after a
restart.
- Verify rollback clears provenance that no longer matches retained history.
GitOrigin-RevId: 7f91c6202fe8d9be21c78dda42fa205937c44263
## Why
App widgets need to read resources using the app and account context of the
tool call that produced them, including after a thread is restored.
## What changed
- Add an optional `originCallId` to `mcpServer/resource/read` requests and
return it on successful scoped reads.
- Track bounded provenance for successful app tool calls in thread history and
use the current tool binding, account link, and app policy when reading the
associated widget resource.
- Reject missing, mismatched, failed, or ambiguous origins, while keeping
app-only tools available for widget reads without exposing them to the model.
## Testing
Add app-server coverage for scoped widget reads across both history modes,
ephemeral and persistent threads, server restarts, app-only visibility, and
invalid origin cases.
GitOrigin-RevId: 11eaefc066b2a3b639599e4cbdef680233d150a2
## What changed
- Exclude descendant-authored `MESSAGE` progress updates from the history retained after remote compaction v2.
- Continue retaining descendant-authored tasks while omitting progress updates and final answers from subsequent requests.
## Testing
- Extend the remote compaction follow-up test to cover descendant progress updates and descendant-authored tasks.
GitOrigin-RevId: 3e3e60bb108d27e99e2a4c222eed4b9d02fd3829
## What changed
Start sampling for automatic idle `UserInput` requests without adding an empty
message to the input queue. Continue queuing automatic response-item input.
GitOrigin-RevId: 5d304aed9e7e165380b29b5b59d9245e97b40659
## Why
Collaboration guidance can change without changing the active mode or model, such as after a model catalog refresh. Tracking only the mode and model leaves the earlier guidance in retained history.
## What changed
- Include a hash of the rendered collaboration instructions in the world-state snapshot so changed guidance is appended to retained history once.
- Emit one empty collaboration fragment when guidance is removed, preventing stale instructions from remaining active without repeatedly clearing them.
- Refresh legacy snapshots that lack an instruction hash once on resume.
## Testing
Cover same-mode catalog updates, clearing missing or explicitly empty guidance, retained-history behavior, and cold resume from legacy snapshots.
GitOrigin-RevId: 69f690adeccd1fb9794a6b15647e6efae3fec4f6
## Why
Cold resumes and forks could lose the thread's active permission profile and
fall back to the current configured default.
## What changed
- Restore the latest persisted approval policy, approvals reviewer, and active
permission-profile ID when resuming or forking a thread.
- Re-resolve persisted profile IDs through current configuration and
requirements. Missing or invalid profiles fall back to the configured
default, while threads without a persisted profile continue to use current
configuration.
- Keep explicit permission and approval overrides ahead of persisted settings.
## Testing
Added coverage for legacy and paginated histories, settings updates, explicit
overrides, removed or changed profiles, requirement fallbacks, and forks.
GitOrigin-RevId: 2289af2f3206bf131a6ffe0e8447e8d402e76c47
## What changed
- Add the optional `active_permission_profile` field to `TurnContextItem` to record the built-in or named profile that produced the effective permission profile.
- Populate the field when serializing a turn context while preserving compatibility with records where it is absent.
## Testing
- Add coverage verifying that turn-context serialization stores the active permission profile.
GitOrigin-RevId: d69142ab1b26595ab406235f89818ef41370e963
## What changed
- Fetch recommendations from the Codex-specific `/ps/plugins/suggested/codex` endpoint and parse its compact response shape.
- Fetch the selected plugin's details before presenting an install request, using them to verify availability and populate connector metadata.
- Skip install elicitation when the selected recommendation is no longer available, and return a retryable response when its metadata cannot be verified.
## Testing
- Cover the new recommendation route and response shape.
- Cover metadata hydration for available plugins and rejection of unavailable plugins.
GitOrigin-RevId: 2b0e2d70572aae2b7cd8e458b42e9dd14be3dbaf
## What changed
- Remove `experimental_thread_config_endpoint` from the configuration model and schema.
- Stop app-server and the in-process client from constructing a remote thread config loader from local configuration.
- Keep the thread config loader supplied at app-server startup fixed for the lifetime of the config manager.
GitOrigin-RevId: e7907fee2d631e7ecf89ec6bac8f22bc2c221350
## What changed
- Mirror execution, permission, and patch approval requests into active realtime conversations with a prompt to review them in the app.
- Mirror user-input and elicitation requests with a prompt to respond in the app.
- Include the serialized request event so the realtime handoff retains the request details.
GitOrigin-RevId: 56fefa70b9ecaaad2fcf83592f20e8b4514a004a
## What changed
- Give `PluginsManager` a shared `AuthManager` instead of a separately mutable authentication-mode snapshot.
- Read the current authentication mode and credentials from that shared manager for plugin discovery, startup tasks, CLI commands, MCP setup, and external-agent migration.
- Update test helpers and coverage to exercise plugin projections and curated marketplace selection as authentication changes.
GitOrigin-RevId: 600c94de5130eda2da5727e1a0b4d39083fefc56
## Why
Filesystem permission paths can use a convention that differs from the host
running Codex. Converting them immediately to native absolute paths can change
the meaning of ambiguous paths such as `/C:/secret` or Windows UNC paths.
## What changed
- Store literal filesystem permission paths as `PathUri` values through the
runtime policy and execution protocol.
- Keep legacy string-based serialization at explicit protocol boundaries and
reject conversions that cannot be represented losslessly.
- Encode native paths as opaque URIs when a normal file URI would imply the
wrong path convention.
## Testing
Added coverage for cross-platform and ambiguous path round trips, UNC path
variants, permission-profile serialization, and deny-policy enforcement.
GitOrigin-RevId: 5247713796d1f2bb4e02f94eb9fc82d4698060f0
## What changed
- Add spans for environment resolution, readiness waits, and environment info,
including fields for the selected environment and resolution state.
- Propagate the active span and tracing subscriber into spawned resolution,
startup, and shell-snapshot tasks.
- Verify that blocking snapshots retain the expected nested environment spans.
GitOrigin-RevId: 3c47e9ec1ebf45dbe1250b9ff76e2983d2a8df2a
## What changed
- Add `EnvironmentNetworkPolicy` as a serializable view of portable domain,
Unix-socket, upstream-proxy, and local-binding restrictions without exposing
controller-owned proxy runtime settings.
- Add an optional `network_policy` field to `EnvironmentConfig` and re-export
the policy and related permission types through the core APIs.
- Reject owner-provided network policies during environment preview and
readiness until runtime enforcement is implemented, while preserving the
existing controller policy when the field is absent.
## Testing
- Cover rejection through both environment preview and readiness, including
preservation of the existing environment selection.
GitOrigin-RevId: 80ad4cf4d4a45632daa7a823e6cf568eb0e8bb80
## Why
Creating the telemetry shutdown thread during shutdown can fail under resource
pressure, including when the native thread guard page cannot be allocated.
## What changed
- Start and verify a dedicated shutdown worker when `OtelProvider` is created.
- Send the provider to the prepared worker for bounded shutdown, while preserving
timeout behavior and avoiding a potentially blocking destructor if worker
preparation failed.
- Rename the fallible provider constructor to `try_new`.
## Testing
Add Unix regression coverage that injects guard-page allocation failures, plus
coverage for worker preparation failure, successful shutdown, and timeouts.
GitOrigin-RevId: 3656298078a800a7fa392437c2ee4a68753092e3
## What changed
- Register `in_app_chat` and `in_app_dictation` as stable, default-enabled feature gates for desktop apps.
- Allow requirements to disable either capability and expose both keys in the generated configuration schema.
## Testing
- Add configuration tests that verify enterprise requirements can disable each gate.
GitOrigin-RevId: 5bba35d710fb0b6924aa55df98b2e013fa288f6d
## What changed
- Add `cli_auth_credentials_store` and `chatgpt_base_url` as exact managed requirements loaded from local requirements configuration.
- Apply these requirements to runtime and bootstrap authentication configuration, overriding user-configured values.
- Expose the managed values through `configRequirements/read` and reject attempts to change them through config write APIs.
- Ignore these local-only authentication requirements in cloud-managed requirement layers.
## Testing
- Cover requirement loading and composition, runtime and bootstrap overrides, app-server responses, and read-only config writes.
GitOrigin-RevId: 8ea9c096386072036ca5441f3d48af8707988a65
## Why
Unified exec could drop network approval registration when a selected environment used a Windows working directory on a non-Windows host because the trigger path could not be converted to a host absolute path.
## What changed
- Keep network approval trigger working directories as `PathUri` values and render them with their inferred native path syntax for Guardian requests.
- Fall back to the selected turn environment only when a trigger path can be converted to a host absolute path.
- Reject selected environment working directories larger than 8 KiB at the thread-manager validation boundary.
## Testing
- Cover managed-network approval registration for a foreign Windows working directory.
- Verify that oversized working directories are rejected as invalid `thread/start` requests.
GitOrigin-RevId: 646fbf12115a35a7e5cb04b511464797bee6647b
## Why
Resumed and forked threads need their persisted Guardian risk state available when routing later approval requests.
## What changed
- Initialize a session's `SecurityRiskScore` from the latest score in its rollout history when no score was supplied explicitly.
- Preserve the inherited score when resuming or forking a thread so low-risk actions can skip redundant Guardian reviews.
## Testing
- Add app-server coverage for inheriting the latest Guardian score when resuming and forking threads.
GitOrigin-RevId: 95a6c803f8dae75516d1da9d66a722379a0a7ef2
## What changed
- Define MCP server identities, matchers, requirements, and per-plugin MCP requirements in the new `codex_protocol::mcp_policy` module.
- Keep requirement validation and matching against `McpServerConfig` in `codex-config`.
- Re-export the moved types from `codex-config` to preserve existing consumers.
GitOrigin-RevId: ad8a107e3085191241a5766f061dd004103e21e6
## Why
Guardian V2 should not override a managed approval-reviewer policy that excludes
the user reviewer.
## What changed
- Force `features.guardianv2` off when `allowed_approvals_reviewers` does not
include `user`, for both local requirements and cloud config bundles.
- Preserve Guardian V2 when the user reviewer remains allowed or when only
legacy Guardian feature settings are present.
- Return the effective disabled value through `config/read`.
## Testing
Added coverage for cloud requirements and the app-server config RPC.
GitOrigin-RevId: 82b6d0aaf47fbaa58ae349d4e56777cd442f5807
## What changed
- Add optional `guardian_v2` model-message defaults for classifier instructions,
review thresholds, reasoning effort, transcript selection and limits, and
action and compaction token limits.
- Apply those defaults when Guardian v2 samples a tool call, while preserving
explicit `[features.guardianv2]` settings over catalog values.
- Preserve the new catalog configuration when applying model overrides.
## Testing
- Cover model-catalog serialization and model override behavior.
- Verify that sampling uses catalog defaults while retaining local overrides.
GitOrigin-RevId: 4ae9ec235c647acbca05a20b7c4b87a4e1331161
## What changed
- Add `[skills].max_context_tokens` to override the token budget used to render the available-skills catalog.
- Require a positive value and cap configured budgets at 10,000 tokens.
- Preserve the existing default of 2% of the model context window when the option is unset.
## Testing
- Cover configuration parsing, override behavior, the 10,000-token cap, and catalog rendering within the configured budget.
GitOrigin-RevId: f29ddcb57fc2d800c0006d03e1c33cbe3ab6272d
## What changed
- Add an optional restrictive execution policy to `EnvironmentConfig` and merge it over the active command policy for shell execution and approval decisions.
- Reject environment policies that contain allow rules, so environment configuration can only tighten command access.
- Include the environment policy fingerprint in cached approval keys so policy changes require fresh session approvals.
## Testing
- Cover restrictions overriding saved prefix approvals, policy changes invalidating session approvals, and environment overlays remaining effective when model-specific prefix filtering is active.
GitOrigin-RevId: e94f877549ed5e0f577b8cb52f3c1872c58f029d
## Why
Permission profiles written before the rename to `:workspace_roots` can still
contain `:project_roots`. Treating that token as unknown ignores its entries and
can drop filesystem restrictions.
## What changed
- Parse `:project_roots` as an alias for `:workspace_roots`.
- Cover deny rules and read-only subpath carveouts in legacy profiles.
GitOrigin-RevId: 10aafc9e15372d1a8bcc2dbcf3ad7cc987f4e67e
## What changed
- Carry `ShellEnvironmentPolicy` in each resolved `EnvironmentConfig` and use
the selected turn environment's policy for shell commands, user shell tasks,
and unified exec.
- Infer the policy from the thread configuration when an environment does not
provide its own resolved configuration.
- Redact the policy from `EnvironmentConfig` debug output because it can contain
explicit environment variable values.
## Testing
- Verify shell handlers and unified exec filter inherited variables according
to the selected environment while preserving its explicit overrides.
GitOrigin-RevId: 9f6a52aa4af60e5f17251a0e1b11e1926b779055
## Why
Callers need to associate host-owned state with a thread before Core starts it.
## What changed
- Add `ThreadManager::reserve_thread_id` and allow new threads to start with the reserved ID. Reject reserved IDs when resuming an existing thread.
- Let thread stores stage metadata for a reserved ID. The local store merges staged values into the first successful metadata update and then removes them.
- Clear staged metadata when an unmaterialized thread shuts down or is discarded, and require the state database for staging.
## Testing
- Cover reserved ID generation and resume validation.
- Cover metadata persistence, merging, rollout-compatible updates, cleanup, and invalid staging requests.
GitOrigin-RevId: 3f9bc48f0f7764011cc90ea01349e497dda0a9e3
## Why
Thread settings can change while a turn is still active, including between sampling requests. Those updates should apply to the next turn instead of changing the model configuration partway through the current turn.
## What changed
- Snapshot the model, reasoning settings, service tier, approval settings, and model-attributed telemetry in `StepContext`.
- Build prompts, model requests, world state, tracing, and startup prewarms from that step-scoped snapshot.
- Share immutable model metadata between turn and step contexts.
## Testing
Added an integration test that pauses an active turn, updates its thread settings, and verifies that all requests in the active turn retain the original settings while the next turn uses the updates.
GitOrigin-RevId: c51352218e463dcc9b7338566d16742292c49fbf
