## Why
Enabling, disabling, or reinstalling a cached remote plugin can change its
effective MCP servers, hooks, and skills without materializing a new bundle.
These transitions need to invalidate plugin consumers just like bundle updates
and removals do.
## What changed
- Report a unified set of changed remote plugins and their affected
capabilities for materializations, removals, and installed-state changes.
- Derive enablement and reinstall changes by comparing installed snapshots,
while keeping materialization metadata separate for hook trust handling.
- Preserve removal notifications even when stale cache cleanup fails, and use
all reported changes to refresh loaded plugin state and runtime consumers.
## Testing
Added coverage for cached plugin enablement, disablement, removal, and
reinstallation, including removal when cache cleanup cannot complete.
GitOrigin-RevId: 86d9c1356f7cf70b3c85effebc3e5986b931bdba
## Why
Managed configuration can disable resumable unified execution while leaving shell tools enabled. This should not remove command execution entirely or grant access to retained processes.
## What changed
- Keep `exec_command` available in completion-only mode when `unified_exec` is disabled.
- Remove `tty`, `yield_time_ms`, `session_id`, and `write_stdin` from that mode, and add a `timeout_ms` argument.
- Terminate one-shot commands on timeout or turn cancellation, reporting timed-out commands with exit code 124.
## Testing
Added integration coverage for tool availability, timeout cleanup, and interruption cleanup under managed configuration.
GitOrigin-RevId: a382d34f922fdc4ba52889272c6db985abd42e47
## What changed
- Add the `codex-guardian-context` crate for assembling structured context shared by synchronous Guardian review and asynchronous scoring.
- Preserve transcript entry roles and original byte counts, and let section contributors declare shared or consumer-specific scope.
- Collect applicable sections in registration order, skip optional sections, and fail the collection when required evidence is missing.
- Provide UTF-8-safe prefix/suffix truncation with approximate omitted-token accounting.
## Testing
- Cover scoped registry collection, ordering, optional sections, required-evidence failures, truncation markers, and UTF-8 boundaries.
GitOrigin-RevId: 19d8458403c470c9b413992dbff925a12259c3c5
## What changed
- Generate a UUIDv7 turn ID for each Guardian classification and preserve it across retries.
- Record the owning turn as `parent_turn_id` and propagate its trusted `root_turn_id` when known, including through the tool lifecycle extension API.
- Omit ambiguous or unknown root lineage and keep classification identities distinct when reusing Guardian sessions and WebSocket connections.
## Testing
- Cover classification metadata across retries, connection reuse, concurrent samples, auth refresh, and consecutive owning turns.
GitOrigin-RevId: 78757f2ad031844e4b545802f634c678a9ee2298
## What changed
- Add an optional `timeoutMs` parameter to `thread/shellCommand` and propagate it to user shell execution.
- Preserve the one-hour default when the parameter is omitted or `null`, allow longer deadlines, and treat `0` as an immediate timeout.
- Reject negative or otherwise invalid values before execution. Timing out an auxiliary shell command leaves its active turn running.
## Testing
- Cover default, extended, immediate, invalid, standalone, active-turn, and interruption behavior.
GitOrigin-RevId: d72f67793d9766b7383f28d30140270a4a76a495
## What changed
- Limit the proactive-mode override to earlier developer instructions that require an explicit delegation request.
- Preserve user requests as higher-priority guidance.
- Encourage root agents and subagents to delegate parallel work when doing so could improve speed or quality.
## Testing
- Update the multi-agent world-state snapshot for the revised proactive-mode message.
GitOrigin-RevId: 0b61701980e0306dd37a925aceb23c75bf3c1af7
## Why
Fetching plugin recommendations while building the first-turn context delays that
context when the recommendation catalog is not already cached.
## What changed
- Start fetching recommendations during session initialization, in parallel with
MCP and plugin setup, and reuse the in-flight request when building context.
- Centralize the feature checks that enable plugin recommendations.
- Prevent an invalidated recommendation request from repopulating the cache after
a newer request has completed.
## Testing
Added coverage for startup feature gating, reuse of the in-flight fetch, legacy
behavior when recommendations are unavailable, and cache invalidation races.
GitOrigin-RevId: 708774b9bfc472c9847973ae8b2774b66e7321e4
## What changed
- Include Windows safety guidance in the `exec_command` tool description when a
single ready environment reports Windows as its executor platform.
- Omit that guidance when the selected executor reports a non-Windows platform.
- Preserve the host-based behavior when platform information is unavailable or
the tool can target multiple environments.
## Testing
- Cover Windows, non-Windows, unknown-platform, and multi-environment tool plans.
- Verify remote environments expose guidance that matches their target OS.
GitOrigin-RevId: 95e09790446c684ffaff2a0be077db426fff4618
## What changed
Update the `send_user_message_async` tool guidance to reserve it for questions
that require missing information, preferences, constraints, clarification, or
approval. Ask models to keep these questions concise, understandable, and at
an appropriate level of abstraction for the user and task.
GitOrigin-RevId: 4a16754401fb5fc106bbf65759ac4a7a78b6c7b3
## What changed
Add an end-to-end core test that compresses a parent and forked child rollout,
then resumes the child through paginated history loading. Verify that the
resumed model input:
- retains the persisted compaction checkpoint and inherited history;
- includes child history through the fork boundary;
- excludes obsolete pre-checkpoint and post-fork parent messages; and
- reads the compressed ancestor without materializing its JSONL file.
GitOrigin-RevId: 1624c4d844345fe2671436f37bc39b3e59fb3dea
## What changed
- Scan regular files under `CODEX_HOME` in a background blocking task when
standalone app-server metrics are enabled.
- Record `codex.app_server.codex_home.size_bytes` for the entire home and the
`sessions` and `archived_sessions` subdirectories.
- Avoid following symlinks, discard incomplete scans, and cancel scanning during
shutdown.
GitOrigin-RevId: 236fc0a5c156dbd4a99e9835135de194809b4a7b
## Why
Rollout compression skips files that participate in paginated fork lineages because lineage readers rely on byte offsets into the original JSONL. This prevents shared rollout history from being compressed.
## What changed
- Add a seekable rollout reader that preserves logical JSONL offsets for plain and zstd-compressed files without materializing compressed data beside the rollout.
- Use the reader for lineage validation, model-context replay, and SQLite history projection, including when compression races with a read.
- Add the opt-in `local_thread_store_shared_compression` feature. The default mode continues to leave shared lineages uncompressed for compatibility, while the new mode includes referenced sources and fork pointers.
- Record the uncompressed size in new zstd frames so common prefix-bound checks can avoid decoding the full file.
## Testing
Add coverage for seekable compressed reads, prefix validation across zstd frame shapes, compressed lineage replay and projection, concurrent forks, and compression-mode behavior.
GitOrigin-RevId: 83e519f7eabf7448977eff37c55899858cc17a07
## Why
Manual terminal-input approvals shell-quote the proposed input, which cannot
preserve NUL bytes for an accurate review.
## What changed
Reject terminal input containing a NUL byte before creating an approval request
or sending any of the input to the terminal.
## Testing
Extend the unreviewable-input integration test to cover both oversized input
and NUL-containing input, and verify that rejected input is not executed.
GitOrigin-RevId: e738c59ba5570012663bf6168ad052be237ca226
## What changed
- Assign type-prefixed IDs to Guardian classifier input items before sending the request, preserving existing IDs and reusing generated IDs across WebSocket retries.
- Add a prefixed ID to the memory phase-one user message.
## Testing
- Verify retried Guardian requests contain the same unique, prefixed item IDs.
- Verify the memory phase-one request includes a prefixed user-message ID.
GitOrigin-RevId: e01da49d9a661b6bfd2088f091a3ddf69d601e40
## Why
An MCP binding can wait for one server while another connection updates a
shared tool catalog. Preferring the catalog captured before that wait can omit
newly available tools from the first model request.
## What changed
- Recheck the current MCP tool catalog before using the captured fallback.
- Keep the captured catalog when it expires during the wait, while honoring a
server that disables catalog caching.
## Testing
- Cover a shared Apps cache populated by a peer while binding capture waits.
- Cover cache opt-out while capture waits for another server.
GitOrigin-RevId: e4389c7ba23e4d8397dbae0930c9e36ba0a172fb
## Why
Binding capture can wait for multiple MCP servers concurrently. During that wait, a tool catalog may be published by another connection or expire after it has already allowed a server to remain dormant. Re-reading only the current cache could omit those tools or trigger unnecessary startup.
## What changed
- Retain the cached catalog that permits lazy startup for the duration of binding capture.
- Recheck the cache after the optional startup grace so catalogs published while waiting are included.
- Add regression coverage for catalogs published during startup and catalogs that expire while another server is resolving.
GitOrigin-RevId: b5c66831db8f9ebecc79ff25c5b4b0182ad68a2f
## What changed
- Mark `clock.curr_time` and `clock.sleep` as built-in control tools so their calls emit control-tool analytics.
- Cover completed, failed, and interrupted calls, verify analytics omit tool arguments and output, and keep the calls out of the dynamic-tool count.
GitOrigin-RevId: 489c8015f836be146ccc2c7e96ef7243b0599e16
## Why
An existing terminal keeps the sandbox, network settings, and grants it had at launch. Later `write_stdin` calls must account for that retained authority, including when permissions change between turns.
## What changed
- Capture each terminal's effective launch permissions and compare them with the current environment before accepting non-empty input.
- Request approval when the terminal bypassed the sandbox, retains additional grants, strict auto-review is active, or its sandbox or network settings differ from the current policy.
- Reject input when current denied-read or environment-owned network restrictions cannot be enforced by the existing terminal, while keeping empty polls and non-TTY interrupts available without review.
- Include the effective sandbox mode and user-visible additional grants in `write_stdin` approval actions without exposing host-owned grant paths.
## Testing
Added unit and end-to-end coverage for retained cross-turn grants, strict auto-review, sandbox and network policy changes, denied reads, Windows sandbox backends, and private grant redaction.
GitOrigin-RevId: e2d71daf9358474c885998786c4928069100711c
## What changed
Update the Guardian v2 required-model-switch test to continue the thread with
empty input. This keeps authorization unchanged, so the test specifically
verifies that the required-model check prevents reuse of a cached approval for
the sandboxed command.
Continue asserting that a Guardian review starts and that the command skips a
second classification.
GitOrigin-RevId: 5ce914cb6e66f0527d9ec154331cc356104bfb72
## Why
The overload integration test coupled terminal error behavior to live retry timing and tracing events.
## What changed
- Verify retry attempts and exponential backoff directly with `run_with_retry` under paused Tokio time.
- Keep the overload integration test focused on request count and terminal error events.
- Add timeouts while waiting for retry telemetry, resumed requests, and turn completion so failures cannot hang the test.
GitOrigin-RevId: 571a6a6a36a6aded038000a63bff1958c7209913
## Why
After switching a thread to a model that requires automatic review, a
standard-scope approval could reuse a cached low-risk score for an action that
skipped classification. This allowed the action to bypass the required full
review.
## What changed
Check the thread's current model requirements before making a fast Guardian
decision. Defer required models to the full review path instead of approving
from cached scoring state.
## Testing
Add coverage for switching to a required model and reviewing a sandboxed
command without running another classification.
GitOrigin-RevId: e79e6a5df19c46acd5c4176f522a859793177639
## What changed
- Share the root thread's current service tier across its agent tree.
- Apply that tier to each subagent request when the selected child model supports it, including remote compaction requests.
- Propagate root service-tier updates to existing and newly spawned subagents without rewriting child-owned settings.
- Remove per-spawn and role-level service-tier overrides so routing remains controlled by the root thread.
## Testing
- Cover supported and unsupported child models, full-history forks, live tier changes, remote compaction, new subagents, and reloaded subagents.
GitOrigin-RevId: 6c299e224e90b41a558788fb7aa4d5f336fa383a
## What changed
- Convert history backend `images` into `input_image` function-call output items alongside encrypted or plaintext history content.
- Keep image data out of logged output and post-tool-use hook responses.
- Reject malformed image attachments instead of silently dropping them.
## Testing
- Cover encrypted and plaintext outputs, supported image detail values, malformed attachments, and propagation into the next model request.
GitOrigin-RevId: a8eee10eb09c637bcfb06a759dd6313caf707e51
## What changed
- Build per-context conflict checks from the shared runtime action registry instead of maintaining separate action lists.
- Preserve the existing context validation order and error messages.
- Add regression coverage for the first reported conflict across editor, Vim, pager, list, agents, and approval contexts.
GitOrigin-RevId: 665839bb9094ac573ab1494248c69c4000998483
## Why
History and notes results are already limited by the backend using the requested
output budget before encryption. Applying another client-side limit can reject
or truncate an already bounded response.
## What changed
- Return encrypted history and notes results without an additional size check.
- Preserve fallback JSON results instead of truncating them again when building
the tool response.
GitOrigin-RevId: 97e7a59a13e8d485cc3c613d0fdf8073f6eb67ab
## Why
Realtime voice calls can span zero or many backing turns, so they need the
saved thread source to identify their origin consistently.
## What changed
- Add `thread_source` to `x-codex-turn-metadata` when starting realtime
WebSocket, existing-call, and WebRTC transports.
- Encode the metadata as header-safe ASCII and omit absent, invalid, or
larger-than-256-byte source values.
## Testing
- Cover source propagation, escaping, and size limits across realtime
transports, including child threads that share a parent session.
GitOrigin-RevId: b5a9eef8dfdbb83387bd0e364a4922b15030f4d6
## What changed
- Add a stable `sleep_tool` feature that can enable or disable registration of the built-in sleep tool independently of the clock tool.
- Support structured `features.sleep_tool` configuration with `model_driven` and `always_on` modes. The default `model_driven` mode preserves the existing model and `current_time_reminder` behavior, while `always_on` registers sleep whenever the feature is enabled.
- Preserve nested sleep-tool configuration across CLI overrides, config merges, and feature toggles, and expose the settings in the generated config schema.
## Testing
- Cover feature-map overrides, both selection modes, legacy clock settings, config merging and editing, and invalid mode rejection.
GitOrigin-RevId: 18eadc582f8a4445958c29d063e5be495276703e
## What changed
- Emit turn-scoped authentication recovery started and completed events when a model provider refreshes expired credentials.
- Add stable `modelProvider/authRecoveryStarted` and `modelProvider/authRecoveryCompleted` app-server notifications with the thread, turn, provider, and user-facing message.
- Show recovery progress in the TUI and `codex exec`, including Amazon Bedrock session reauthentication.
## Testing
- Cover provider recovery success and failure events, app-server routing, client rendering, and Amazon Bedrock credential refresh.
GitOrigin-RevId: 3010c38d0676f18bced27761cf86dd38344d09f9
## What changed
Return consistent `Unable to perform operation:` messages for provider,
authentication, request construction, transport, and response parsing failures.
Do not include the underlying error details in these user-facing messages.
GitOrigin-RevId: c23fe7713f9d61e5f013ef217e1c0bdc2b824d83
## What changed
- Add the under-development `powershell_shell_version` feature flag.
- When enabled for a single local PowerShell environment, query the selected
shell executable and include its major/minor version in
`<environment_context>`.
- Cache version lookups, bound command execution and output, and report when a
previously visible version becomes unavailable.
## Testing
- Cover environment-context diffs when the shell version appears or disappears.
- Verify on Windows that the version is model-visible only when the feature is
enabled.
GitOrigin-RevId: 3ec8e80425ec3c193134dc1b69f61ef09ede375e
## What changed
- Count loaded-plugin cache requests by `hit`, `hit_after_wait`, or `load` outcome.
- Record time spent waiting for the load semaphore and loading plugins.
- Count cache clears and capacity evictions.
- Remove the unused force-reload path from `PluginsManager::plugins_for_config`.
GitOrigin-RevId: db7829e7bd9b1414a263d9a7100c05dfdd832777
## What changed
Apply the legacy plugin app/MCP routing policy even when no authentication mode is available. Unauthenticated plugin reads now omit unavailable app declarations while retaining their MCP server alternatives.
## Testing
Extend plugin manager coverage for unauthenticated reads and provide ChatGPT authentication in the app-server bundle-details test.
GitOrigin-RevId: d90d29d0702488eb7413c6779254a9e6aae615c0
## Why
Microsoft Store PowerShell executables under `WindowsApps` can be inaccessible to the dedicated elevated sandbox account.
## What changed
- For local elevated sandbox commands that target Store PowerShell, select the first compatible `pwsh.exe` or `powershell.exe` from `PATH` or the standard fallback locations.
- Continue adding `-NoProfile`, and preserve the selected shell for remote execution and other sandbox modes.
## Testing
Added unit coverage for Store-path filtering, fallback ordering, local replacement, and remote-path preservation.
GitOrigin-RevId: 8ab3d0db925e886879701445040f13902fb1ad24
## What changed
- Include `recencyAt`, derived from the newest non-archived assigned thread, in project responses.
- Let `project/list` sort by `position` or `recencyAt` in either direction, with recency as the default descending order for that key and projects without activity placed last.
- Encode the sort key, direction, and millisecond-precision anchor in new cursors while retaining existing ascending-position cursors.
- Add an index for project recency lookups and document the API's sorting and pagination behavior.
## Testing
- Cover recency updates when threads are assigned, archived, unarchived, or deleted.
- Cover ordering, null handling, tie-breaking, cursor pagination, invalid sort combinations, and cursor mismatches.
GitOrigin-RevId: 6165e2b9976b0b45cfa16bcd42978b67811f6df5
## What changed
Resolve the review model's token budget from the parent turn's original
configuration and model-default setting when deciding whether a Guardian
follow-up review needs context rollover. This preserves explicit token-budget
preferences instead of unconditionally applying the review model's defaults.
GitOrigin-RevId: cb3adaea7dad22f4975d84c6ee62dfc2bfbddbb9
## Why
Remote environment registration can return an explicit `503` `registration_conflict` after the registry finishes its own recovery attempts. Retrying ambiguous failures is unsafe because a timed-out request may still have replaced a newer registration.
## What changed
- Retry only explicit `registration_conflict` responses with jittered backoff, both during initial registration and after a rejected relay connection.
- Apply the existing connection timeout to the complete registration attempt, including response-body diagnostics.
- Keep retry waits and in-flight requests cancellable while preserving the Noise identity and initialized session across successful recovery.
## Testing
Added unit coverage for retry eligibility and timeout handling, plus relay tests for identity and session preservation and prompt shutdown during registration.
GitOrigin-RevId: 374e6fdac4755f377cbea9400c76a228b790945a
## What changed
- Reuse the mention codec's linked-tool parser and mention-name helpers when extracting skill mentions.
- Remove the duplicate parser and environment-variable filter from the skill mention code.
## Testing
- Cover Unicode byte offsets, whitespace-trimmed paths, both supported sigils, and malformed linked mentions.
GitOrigin-RevId: 4ed3627f0caf90d49bcbc0681ff9d9b43b3736e3
## Why
Long-lived Guardian sessions can exhaust the review model's context window. After
a rollover, a transcript delta is insufficient because the new window no longer
contains the original transcript or persistent review instructions.
## What changed
- Compact an exhausted Guardian session before submitting the next review while
respecting the review model's token-budget configuration.
- Detect whether the full transcript remains in the reviewer history. If it does
not, reset transcript and Node REPL evidence cursors so the next review replays
the complete transcript and associated image evidence.
- Restore the follow-up reminder and Node REPL policy when compaction removes
them.
## Testing
- Cover rollover both when the reviewer window is already exhausted and when a
follow-up reminder crosses the limit.
- Verify that multimodal reviews preserve the fallback buffer and replay browser
evidence after rollover.
GitOrigin-RevId: f2a02a84e1f25d15c6e2b8eddb93e2d1805449f5
## What changed
- Register both `clock.curr_time` and `clock.sleep` when a model lists `clock` in `experimental_supported_tools`.
- Keep current-time reminder injection controlled separately by the `current_time_reminder` feature.
## Testing
- Cover a model-advertised clock with reminders disabled, verifying that both tools are available and no reminder is added.
GitOrigin-RevId: c39b5cb6e2da397daaaa4e0ad81f0763b8dd91e9
## Why
Read-deny policies must use the target executor's path convention so URI-based
policy checks and native filesystem enumeration enforce the same rules.
## What changed
- Prepare deny roots and glob matchers from `PathUri` policy context, including
executor-relative working directories and home-relative patterns.
- Match Windows globs case-insensitively with normalized separators, while
preserving byte-oriented POSIX matching for non-UTF-8 paths.
- Fail closed for malformed paths, incompatible path conventions, unresolved
home-relative patterns, and invalid globs.
- Make Windows deny-read discovery use case-insensitive ripgrep glob matching.
## Testing
Added coverage for Windows URI conventions, executor home expansion,
case-insensitive `.env` discovery, malformed paths, non-UTF-8 names, and
canonical directory-link targets.
GitOrigin-RevId: 36001219a2e9b36acfce8972dc1d3bbc304271c5
## What changed
- Load the effective plugin configuration for every requested `cwd` in
`plugin/list` and `plugin/installed`.
- Combine local marketplaces in request order, keeping the first source for
duplicates while merging installed and enabled state across repositories.
- Report invalid project configurations through `marketplaceLoadErrors` without
dropping valid local catalogs or globally enabled remote catalogs.
- Use the combined marketplace context for forced and background cache refreshes,
while keeping remote plugin feature gating request-wide.
## Testing
- Cover distinct and duplicate marketplaces, cross-repository enabled state,
invalid project configuration, local cache refresh, and global versus
project-local feature flags.
GitOrigin-RevId: 53c8e0e9c760dc2b969ab10d0e63fb439673c230
## What changed
- Add the optional `platformOs` field to exec-server environment metadata, with
local exec servers reporting `std::env::consts::OS` and legacy servers
remaining compatible when the field is absent.
- Preserve the reported OS in each selected turn environment, including the
local environment.
## Testing
- Extend protocol serialization and environment-selection coverage for remote
and local platform values.
GitOrigin-RevId: 0f325f6cb83cd8f8c7ec1ce2a4041f45a29e8f38
## What changed
- Add an optional `multi_agent_reasoning_effort` field to model metadata.
- When `Ultra` is selected, use a supported catalog override when present. Otherwise, prefer `max`, then the highest supported non-`ultra` effort, and finally `medium` when the model has no reasoning levels.
- Preserve existing request behavior for other reasoning efforts, including mapping `persistent` to `disabled`.
## Testing
- Cover configured, missing, unsupported, and empty-catalog fallbacks for both parent and spawned-agent requests.
GitOrigin-RevId: b7ef576f32e0ea95ef38bf95c25789ed16aae94e
## What changed
- Count remote executor MCP discovery attempts, including refreshes, with `success` and `error` outcomes.
- Count each discovered server by name and report whether it was found or unavailable before policy and startup filtering.
GitOrigin-RevId: b912aa64b606e6243cce79df0fe85d485a2c3547
## What changed
- Report the executor's user home in environment metadata.
- Cache the reported home on selected turn environments and include it in
filesystem sandbox contexts, including `apply_patch` requests.
- Preserve compatibility with executors that do not report a home directory.
## Testing
- Cover local, remote, inherited, serialized, and `apply_patch` sandbox
contexts.
GitOrigin-RevId: ea65e68e33a80bc9bcfc7c53b1486ab944699470
## What changed
- Add `ToolLifecycleContributor::on_mcp_tool_result` with access to the
executed MCP tool context, rewritten arguments, extension data stores, and
mutable server result.
- Run contributors before publishing MCP completion and before preparing the
result for the model, so extensions can inspect or replace successful and
error results.
## Testing
- Cover unchanged, replaced, and error results for direct MCP calls, plus
replaced and error results through Code Mode.
- Verify completion waits for result processing and the processed result flows
to both completion events and subsequent model input.
GitOrigin-RevId: ece7acfe9c5eed4778cb78a29c3ffd06a74e1bcf
## What changed
- Add `mcp_optional_startup_grace_ms` with a default of 1,000 ms to control how long tool catalog capture waits for optional MCP servers.
- Treat a value of `0` as disabling the shared grace, so optional servers use their configured `startup_timeout_sec` instead.
- Apply updated grace values during runtime and MCP configuration refreshes, and reset cached startup deadlines when the configured duration changes.
## Testing
- Cover custom, disabled, refreshed, and shared startup grace behavior in MCP connection and core integration tests.
GitOrigin-RevId: 936196215ed6e0a29c9ec72cfee17663978df8ef
Expand Windows deny-read globs robustly across protected directories while preserving filesystem enforcement. Preserve structured MCP tool and resource errors through app-server responses. Bind cached Guardian classifications to current local and root authorization state so stale scores cannot approve actions.
Add regression coverage for the sandbox resolver, structured protocol errors, and authorization changes.
GitOrigin-RevId: 4b80ed724d869afeca79204222d8465fa99d3a24
## Why
Planning tools for a candidate model must not change the selected model or its
Responses Lite tool inventory. Otherwise, preparing a fallback can overwrite
metadata before the current model's request is sent.
## What changed
- Build tool registries and visible specs from an explicit `ModelInfo`.
- Store the finalized tool mode, code-mode names, namespace inventory, and
terminal and child-management capabilities on `ToolRouter`.
- Publish namespace inventory only after selecting the step's router, and use
that router's mode when dispatching and recording tool calls.
## Testing
Added coverage for candidate-model isolation, fallback compaction metadata,
code-mode worker selection, normalized tool names, and router capabilities.
GitOrigin-RevId: 56df2de8f42c7e93a4c4dcd117720e40a0cca3db
