refactor(native-chat): structured agents declare their capabilities instead of shared code naming Claude and Codex

[brennanb2025]

	Files	Added	Deleted	Net
Test	80	$\color{#1a7f37}{\Huge{\mathbf{+}}}$​746	$\color{#cf222e}{\Huge{\mathbf{−}}}$​92	$\color{#1a7f37}{\Huge{\mathbf{+}}}$​654
Prod	52	$\color{#1a7f37}{\Huge{\mathbf{+}}}$​594	$\color{#cf222e}{\Huge{\mathbf{−}}}$​223	$\color{#1a7f37}{\Huge{\mathbf{+}}}$​371

Stacked on #24991. Review only what this PR adds on top of it: git diff origin/brennanb2025/acp-a1-opaque-handle 288f72cd43b. The A2 commits are 2667173fed9, 63508a51c34, 88a09c4e4c6, 38ac9109088 and the review-round commits 19c8cb5f417 (required agent registry) and c5aaee8ede9; the rest are merges of #24991 and main.

ELI5

Orca's structured native chat (the chat view that drives an agent through its own protocol instead of scraping a terminal) only knows two agents, Claude and Codex, and that fact was spelled out in many places of the shared code: "if the agent is Codex do this, else assume Claude". Adding a third agent meant editing all of those places. This PR gives each agent a single description of itself (what it can do, and how its options behave while no chat is running) and makes the shared code read that description instead of asking "is this Claude or Codex?".

There is no user-visible change: Claude and Codex chats behave exactly as before.

What Changed

The problem. Before this PR:

• The router that sends each chat to the right agent backend was typed as exactly { claude, codex }, so a new backend was a type edit, not a registration.
• What an agent could do (rewind, compact, set a goal, report context usage) was answered by "does this backend happen to implement that optional method". A client cannot know that before a chat is running, and the router's own compact method always existed, so "can compact" was true for any agent.
• Option reads at rest (no agent process running), frame classification, the create/visibility gates and the renderer's label and validation code each had their own === 'claude' / === 'codex' branches. Some defaulted silently: the chat header label was agent === 'codex' ? 'codex' : 'claude', so any other agent would have been called Claude.
• The stored account home typed its variable as 'CLAUDE_CONFIG_DIR' | 'CODEX_HOME' inline.

Mechanism now.

• Agent definition (StructuredAgentDefinition): each agent's own module declares one (claude-structured-agent-definition.ts, codex-structured-agent-definition.ts): its capability record and its at-rest option rules (which option keys it accepts, its fallback model list, whether an unpicked effort means "the model's default").
• Capability record (AgentSessionCapabilities, plain JSON in src/shared so the next PR can send it to clients unchanged): rewind, compact, threadGoal, contextUsage, imagePrompts, steering: 'inject' | 'queue', approvalEnforcement: 'provider' | 'orca'. The values reproduce today's behavior exactly (a test pins them). A running chat may narrow a declared capability, never widen it: Codex rewind is still narrowed to unsupported for a thread with legacy history.
• One registry answers "what is this agent" (StructuredAgentRegistry, structured-agent-registry.ts): built once at runtime setup from the { definition, adapter } registrations, and a required dependency of both the host and the router. It answers definition(agent), definitions() and capabilities(agent). It refuses a duplicate registration, and a registration whose definition declares compact, threadGoal or rewind without the adapter method that does it, so a declaration can't drift from what the adapter can do. Every reader asks it: option reads at rest (which options a stopped chat accepts, its fallback model list, its default effort), compact admission, thread goal and rewind. There is no second map of agents: Claude's and Codex's definitions are listed only where the runtime builds the registry, so a definition the runtime did not register cannot affect any read.
• Router: routes by the registry and refuses an unregistered agent. The per-capability supportsThreadGoal / recordsContextUsage methods and the "method exists" checks are gone, and so are the adapters' own copies of those answers. A running chat may only narrow: the router's rewindSupport applies the declared rewind itself and then the adapter's narrowing, so no caller can widen it.
• Compact admission: /compact is admitted by the agent's declared compact capability (before, the router's always-present compact method admitted it for any agent). An agent that declares no compaction gets the existing "command refused" answer.
• Frame classification keeps the same per-agent tables but reaches them through a lookup by agent id; Claude's message:result rule moved into Claude's entry. (This is still a third per-agent table beside the definition; see the temporary list.)
• Account home: new src/shared/agent-session-account-home.ts owns the concept (the environment variable that points an agent at its config directory). Stored values are byte-identical ({"variable":"CODEX_HOME","path":…}); the stored-record check still admits only a variable some agent declares, because that variable becomes a child process's environment. The Claude and Codex launch checks and the model-catalog fingerprint read it from there.
• Older clients' turn status text (the line a client that predates the turn item shows) names the session's agent from the catalog instead of === 'claude' ? 'Claude' : 'Codex'. Claude and Codex text is byte-identical; a row with no record and no legacy: id would now say "Agent" instead of guessing "Codex", which no real chat reaches because the record store never deletes records.
• Renderer: labels come from the agent catalog by agent id (same labels as before), including the cut-turn notice main added. Tab/launch validation and the "show structured chat" choice use the one shared agent guard. The Codex-only "the host's model listing names the configured model" rule is now a field on Codex's option catalog (hostListingNamesConfiguredModel) instead of agent === 'codex'.
• Gate for adapters with no per-agent check (structured-agent-session-provider-support.ts): the fallback no longer assumes such an adapter is Codex. In production the host always holds the router, which has a per-agent check, so this only affects single-adapter test doubles.

Deliberately left as Claude/Codex lists (each one place, not a branch):

• AGENT_SESSION_PROVIDER_HANDLE_PROVIDERS / AgentSessionHandleProvider: what stored records, RPC params (including the account-home variable enum) and persisted tab state validate against. Widening it is a wire and stored-data change, which feat(native-chat): open structured chat's wire and stored records to registered agents, behind a negotiated capability #25159 does behind a new host capability (agent-session.structured.registered-agents.v1). temporary
• The variable table in agent-session-account-home.ts (which environment variable points each agent at its config directory). It is not derived from the registered definitions yet because its one safety-relevant reader is the stored-record check (isPersistedAgentSessionRecord), a pure shared function that the record store runs while loading rows, before any adapter or router exists. That same check still hard-codes the stored agent list itself (claude | codex). Making it read the registrations means passing the runtime's declared variables into record loading, which belongs with widening that stored agent list. temporary: the next PR, which widens the stored agent list behind a host capability, and the protocol-agent launch description, which declares the variable with the agent, move both checks onto the registered definitions.
• The model-catalog subsystem (agent-model-catalog-*) still types its agent as Claude/Codex: it is probe infrastructure that gets an agent-neutral shape when the first protocol-driven agent reports its own model list. temporary: feat(native-chat): open structured chat's wire and stored records to registered agents, behind a negotiated capability #25159 opens its types to any registered agent.
• The Claude/Codex account resolution and location rule in the runtime's create path (orca-runtime-get-structured-agent-session-create-support.ts; the location gate picks Claude's or Codex's rule by name because the agent picker asks before the registry exists). temporary: feat(native-chat): open structured chat's wire and stored records to registered agents, behind a negotiated capability #25159 moves account resolution and location support onto each agent's runtime registration (which exists before the host starts), with Claude and Codex as ordinary entries.
• Codex-specific journal identity branches in rewind/recovery and the unhandled-frame reader: they belong to the shared timeline assembler (Admit one provider event's journal writes as one queued operation, decided when it runs #25141 / Add a shared timeline assembler for structured agent chats (not wired yet) #25064), which owns item identity. temporary
• The per-agent frame classification table (provider-frame-disposition.ts): classification belongs to the protocol (every ACP agent shares one) or to the definition. temporary: an agent without an entry gets the safe default (its frames stay visible), which is what the ACP adapter PR (feat(native-chat): Grok as a structured chat over the Agent Client Protocol #25225) relies on; moving classification onto the definition or the protocol is a follow-up.
• Adoption of an existing conversation from history (structured-agent-session-create-adoption.ts) stays Claude/Codex: only they have transcript importers. intended for now.
• Terminal (transcript-scraping) chat code in use-native-chat-session-options.ts and the option discovery files is a different chat system and out of scope.

Why

Adding an agent should be "write its adapter and its definition, register it", not a sweep through shared types, and a client should be able to learn what an agent supports before a chat runs. A capability record declared per agent and narrowed by the running session is the common pattern: a static declaration readers use instead of the agent's name, with the live session allowed to narrow it but never widen it.

Alternatives considered:

• Keep "capability = optional method exists". Rejected: it can't be answered before a session runs, can't be sent to a client, and the router (which defines every method) answered "yes" to compact for every agent.
• Put the whole definition in src/shared so the renderer can read it directly. Rejected: clients must learn capabilities from the host they talk to (an older or remote host may differ), which the next PR does by publishing this same record. Only the account-home variable table is shared, because the stored-record check needs it.
• Make variable any environment-variable name. Rejected: the stored variable is set on a child process, so it stays limited to names an agent declares.

Differences from the common pattern

• Capabilities are not published to clients yet; the renderer still knows which agents exist from the shared list. temporary: the next PR in this stack publishes the record per host behind a capability.
• Narrowing is still one adapter method per capability (only rewind narrows today), not a narrowed record from a startup handshake. temporary: the ACP adapter PR (feat(native-chat): Grok as a structured chat over the Agent Client Protocol #25225) narrows from the agent's startup reply.
• The registry is a dependency of the host and the router, not methods on the adapter contract (matches the common pattern, where the registry is separate from the adapters).
• imagePrompts, steering and approvalEnforcement are declared, but nothing in this PR reads them. temporary: feat(native-chat): Grok as a structured chat over the Agent Client Protocol #25225 reads imagePrompts; nothing in the current stack reads steering or approvalEnforcement yet. They stay so the published record (feat(native-chat): open structured chat's wire and stored records to registered agents, behind a negotiated capability #25159) already says how a protocol agent's follow-ups and permission requests are handled. For protocol-driven agents approvalEnforcement will be orca: Orca answers every permission request the agent sends, by the session's setting. The agent's own settings still decide when it asks, so orca never means a sandbox.

Linked Issue

N/A (internal stack: native chat for more agents).

Visual Proof

N/A: no user-visible change; labels resolve from the same catalog entries as before.

CI status (October 5)

Merged current main. On head 288f72cd43b every CI check passes: typecheck, static analysis, all five unit-test shards, cross-version wire compatibility, relay integration, and packaging for macOS and Windows.

Testing

• I manually tested these changes locally
• Automated tests added/updated, or explained why not below

What I verified:

• New tests: the registry (routes a registered agent the router has no code for, refuses an unregistered one and a duplicate, refuses a registration that declares compact, thread goal or rewind without the adapter method, answers capabilities from the registry, lets a session narrow but never widen rewind, including Claude's undeclared rewind at rest and live), the Claude and Codex definitions (pinned to today's behavior), and the account home (stored values byte-identical, an undeclared variable refused at load).
• structured-agent-session-registered-definition.test.ts: a router registered with non-default Claude rules routes to its adapter, answers that registration's capabilities, accepts that registration's option keys at rest (and refuses model, which Claude's own module accepts), and reads that registration's fallback models and default effort. A router that did not register Claude refuses its option picks and reads no rules for it, even though this build ships Claude's definition. A pick for a session with no record is refused, as before.
• Compact admission by declaration, the cut-turn notice label and the older-client turn text each have a test that fails without the fix. Test doubles build the same registry the runtime does.
• Updated tests that used the old router constructor or the removed per-capability methods.
• Ran 766 test files in native chat, structured runtime/RPC, Claude/Codex structured adapters, renderer native chat and shared agent-session code (7,592 tests), plus orca-runtime.test.ts and the runtime's other suites (59 files): all pass.
• Changed-code quality gate (lint, casting, React Doctor, design system) passes since the stack base; oxlint/oxfmt clean on changed files.

What I didn't verify:

• Review rounds: two round-1 reviews (correctness, and an architecture challenge that recommended the required registry), a round-2 review of the fixes, and a final readiness review: no P0-P2 issues open. Local runs on the final code: 76 A2 test files (637 tests) and 51 runtime suites (387 tests) pass; local typecheck only reports modules missing from the out-of-date shared dependency install.
• No app run or real agent CLI run: this is a refactor with no UI or wire change, and its behavior is pinned by the tests above.

Review

Agent skill upstream boundary

• Not applicable, or this change follows docs/reference/agent-skill-sharing-upstream-boundary.md and copies or mechanically translates no upstream skill-installer source, tests, fixtures, registry entries, path tables, comments, or documentation.

Notes

• Security: the stored account-home variable is still limited to declared names. Cross-platform, SSH/remote: no path, process or platform behavior changed; definitions live on the runtime that runs the agent. Mobile and older clients: no wire change. Backwards compatibility: stored records are unchanged. Overlap with in-flight lifecycle PRs (fix(native-chat): the agent's exit ends its record, and an unconfirmed stop is joined instead of held #24862, fix(native-chat): never drop an agent process end Orca could not confirm #24507, feat(native-chat): the chat reads "Stopping…" from your Stop until the turn actually ends #24369, fix(native-chat): a failed start is shown on its message and retried, and later messages go ahead #24340) is limited to single-line or import hunks in shared files.

Checklist

• This PR is small and focused
• I explained what changed and why (ELI5, the user-facing before/after, the mechanism, and why over the alternatives)
• Before/after screenshots or videos attached for UI changes, or N/A with reason
• Self-reviewed for correctness, security, and performance
• Cross-platform, SSH/remote, and path/shortcut impact considered (or N/A)
• pnpm lint, pnpm typecheck, pnpm test, and pnpm build pass (or CI will cover; local preferred)

[brennanb2025]

Review summary (head 288f72cd43b)

The problem. Structured native chat only knew Claude and Codex, and shared code was full of "if Codex do this, else assume Claude" branches. Adding an agent meant editing all of them. Some defaulted silently: any other agent would have been labeled Claude. "Can this agent compact" was answered by whether a method existed, so it was true for every agent.

User-facing change: none. Claude and Codex chats behave exactly as before. Each agent now describes itself once (what it can do, and how its options behave while no chat is running), and shared code reads that description instead of the agent's name.

What the review changed

• One required agent registry. Before, the agent definitions were reachable only through optional router-only methods on the adapter interface, so any host built without the router would silently lose option rules and capabilities. Now one StructuredAgentRegistry is a required dependency of both the host and the router. It refuses a registration that declares compact, thread goal or rewind without the adapter method that does it.
• /compact is admitted by the declared capability. Before, the router's always-present compact method admitted it for any agent.
• Rewind check in one place. The router applies the declared rewind itself, so no caller can widen it. Claude (no rewind) is refused at rest and while running.
• No more silent name defaults. Two that remained or came in from main now use the agent catalog: the cut-turn notice label (was "Claude" for any non-Codex agent) and the status text older clients show (was "Codex" for any non-Claude agent). Claude and Codex text is unchanged.
• Small fixes. A leftover inline account-home table now uses the shared helper. A stale comment and a test-only export are cleaned up. The PR description now matches the code and names each temporary item's follow-up PR.

Deferred

• steering and approvalEnforcement are declared but read by nothing in the current stack; imagePrompts is read by feat(native-chat): Grok as a structured chat over the Agent Client Protocol #25225. They are labeled temporary in the description. Whether to keep or drop the two unread fields is an open product call.
• Frame classification is still a per-agent table beside the definition.
• rewind is a yes/no flag rather than a richer fork mode.
• Checking a record's account variable against its own agent is tracked on feat(native-chat): open structured chat's wire and stored records to registered agents, behind a negotiated capability #25159, which rewrites that check.

Verified

• Two independent round-1 reviews (correctness, and an architecture challenge that designed its own solution first; verdict keep, no split), a round-2 review of the fixes, and a final readiness review. No P0–P2 issues open.
• Local runs: 76 test files (637 tests) for this PR's code and 51 runtime suites (387 tests) pass. New tests fail when their fix is reverted.
• CI on this head: everything passes, including typecheck, all five unit-test shards, cross-version wire compatibility and packaging.

Not verified

• No app or real agent run. There is no user-visible change, so there was no live QA.
• No real remote or SSH runtime was exercised; definitions live on the runtime that runs the agent, and no wire or stored data changed.

[brennanb2025]

Merged current main into this branch twice (now 3b0c7aba7e1). The first merge brings in the A1 update; the second brings in main's #24862.

• fix(native-chat): never delete chat history on read; a chat Orca can't load says why once #24576: main removed historyFilePath from the structured adapter interface, so I dropped this branch's registry-routed copy in the adapter router.
• fix(native-chat): the agent's exit ends its record, and an unconfirmed stop is joined instead of held #24862: main replaced claude-unproven-stop-send.test.ts with claude-stop-exit-ends-record.test.ts, so I took the deletion. In codex-stop-row.test.ts I combined this branch's agents dependency with main's recording logger.
• Outside the conflicts: four of main's new tests built a host or turn context without the agents registry this branch makes required (newer-content-host, unopened-tab, stop-note-withdrawn-send, claude-stop-exit-ends-record). They now pass it, following this branch's pattern for every other test.
• Verified: 203 of 210 related test files pass locally (2202 tests); 7 can't load on stale local dependencies. The scoped typecheck has only that stale-dependency error. The line-limit lint (oxlint 1.86, the version CI runs) passes on every file the branch changes, as merged with main.
• CI: not run yet. The first job waited about 30 minutes for a runner during today's backlog and was cancelled, so it shows as failed; the run is queued again.

[brennanb2025]

Earlier CI merges missed main's restored provider-handle import and the test-only parser dependency still needed by old release checkouts. Merged main 08a970a3f117; the branch's own changes are preserved, the submission-position test matches main with exactly one used import, and main's package/lock fix is inherited through the merge with no branch-authored dependency edits. The prior broad run passed 1,429 tests; fresh-main checks passed 118 tests, with remaining local failures limited to stale stream-json/stream-chain dependencies, and scoped oxlint passed; new-head CI is pending.

[coderabbitai]

📝 Walkthrough

Walkthrough

The change introduces registry-backed structured-agent definitions and routes session operations through registered adapters. Agent capabilities and resting-option rules now come from definitions. Shared account-home helpers replace provider-specific variable handling in affected APIs. Renderer and legacy status text use agent identity and display names more directly. Test hosts now receive explicit agent registries.

Priority: ⬇️ Low

Merge Risk: 🔵 Low · up to db71a

The remaining test-fixture cleanup does not block merging; the test checks the intended behavior.

Security Architecture Review

Security architecture risk: 🔵 Low · up to db71a

The change centralizes security-sensitive routing and operation eligibility. Inspected controls remain in place, but incomplete coverage of identity and permission boundaries leaves limited residual uncertainty.

Retained concerns
No architecture-level concerns identified.

Security review details

Security Blast Radius

• observed — Within inspected production composition, registration authority remains host-owned and the selectable execution backends remain the same two providers. Unknown identities are refused rather than dynamically creating an adapter or authority profile.

Trust Boundaries and Controls

• observed — Attach input carries provider and agent separately: handle admission checks provider, while journal identity and adapter selection use agent. The attach and RPC paths are unchanged, and the base router already selected by agent. This inspection did not establish a newly reachable or exploitable cross-provider path.
• observed — Persisted account-home variables remain restricted to CLAUDE_CONFIG_DIR and CODEX_HOME. Provider launch checks retain their respective pinned-variable requirements, now obtained from the shared mapping rather than literals.

Resilience and Maintainability Implications

• observed — Delayed option persistence uses the admitted runtime fence. Replacement refuses a moved fence or an unsuitable lease state, preserving protection against stale writers after ownership changes.

🚥 Pre-merge checks | ✅ 4 | ❓ 1

❌ Failed checks (1 inconclusive)

Check name	Status	Explanation	Resolution
Docstring Coverage	❓ Inconclusive	Docstring coverage is 45.71% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 35 functions across 50 files. (80 skipped…	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly summarizes the main change: structured agents declare capabilities instead of relying on shared Claude/Codex-specific logic. It is somewhat long, but remains specific and relevant.
Description check	✅ Passed	The description covers the required ELI5, changes, rationale, linked-issue status, visual-proof status, testing, agent-skill boundary, notes, and checklist sections. It explains that visual proof is n…
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

Full details: Docstring Coverage

Explanation

Docstring coverage is 45.71% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 35 functions across 50 files. (80 skipped: 80 over the file limit.)

✨ Finishing Touches 💡 1

📝 Generate docstrings 💡

• Commit to this branch
• Create a new PR

• Autopilot · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

🧹 Nitpick comments (1)

src/main/native-chat/agent-session-wire/structured-agent-session-mutation-owed-import.test.ts (1)

93-96: 🎯 Functional Correctness | 🔵 Trivial | 💤 Low value

Remove the unused adapter capabilities stub.

The test host already uses claudeAndCodexDeclared(). performThreadGoalChange checks ctx.agents.capabilities(ctx.agent), not an adapter property. The stub does not enable the goal path and misstates where capability declarations come from.

Suggested cleanup

-import { CODEX_STRUCTURED_AGENT } from '../../codex/codex-structured-agent-definition'
...
     Object.assign(host.deps.adapter, {
-      changeThreadGoal,
-      capabilities: () => CODEX_STRUCTURED_AGENT.capabilities
+      changeThreadGoal
     })

---

ℹ️ Review info

⚙️ Run configuration

• Configuration used: Repository UI
• Review profile: CHILL
• Plan: Advanced
• Run ID: 87506e7a-89f9-4c92-a357-94b0a08fd4a6

📥 Commits

Reviewing files that changed from the base of the PR and between 08a970a and db71a89.

📒 Files selected for processing (132)

• src/main/claude/claude-api-retry-idle-sweep.test.ts
• src/main/claude/claude-context-usage-unloaded-turn.test.ts
• src/main/claude/claude-structured-agent-definition.ts
• src/main/claude/claude-structured-effort-default-at-rest.test.ts
• src/main/claude/claude-structured-launch-resolution.ts
• src/main/claude/claude-structured-session-adapter.ts
• src/main/claude/claude-structured-session-test-support.ts
• src/main/codex/codex-provider-retry-idle-sweep.test.ts
• src/main/codex/codex-structured-agent-definition.ts
• src/main/codex/codex-structured-launch-resolution.ts
• src/main/codex/codex-structured-question-order.test.ts
• src/main/codex/codex-structured-session-adapter.ts
• src/main/codex/codex-structured-session-close.test.ts
• src/main/native-chat/agent-model-catalog/agent-model-catalog-fingerprint.ts
• src/main/native-chat/agent-model-catalog/agent-model-catalog-service.ts
• src/main/native-chat/agent-session-wire/provider-frame-disposition.ts
• src/main/native-chat/agent-session-wire/structured-agent-definition.ts
• src/main/native-chat/agent-session-wire/structured-agent-registry.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-accept-then-deliver.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-adapter-router-registry.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-adapter-router-test-support.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-adapter-router.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-adapter-router.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-adapter.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-adopted-import.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-append-delivery.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-at-rest-commands.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-claude-compact-stop.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-claude-echo-working.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-claude-option-queue.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-claude-queued-stop.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-claude-root-exit.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-claude-stop-ends-session.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-claude-stop-exit-ends-record.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-claude-stop-turn-end.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-close-verdict.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-codex-stop-row.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-command-turn.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-conversation-stop.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-crash-mid-start.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-crash-turn-end.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-delivery-loop.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-failed-create-owner-verdict.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-failed-create-sink-release.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-grouped-prompt.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-host-delivery.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-host-mutations.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-host-test-abandon.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-host-test-harness.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-host-types.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-host.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-import-mismatch.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-late-settlement.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-legacy-handoff-record.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-main-agent-working-agreement.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-mutation-admission.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-mutation-context.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-mutation-owed-import.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-mutation-turn-context.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-newer-content-host.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-operation-settlement.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-operation-start-refusal.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-option-settlement.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-options-read.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-options-read.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-owed-work-release.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-prompt-cancel.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-provider-child-record.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-provider-restore.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-provider-started.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-provider-support.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-queued-message-rig.test-fixture.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-recovery-exits.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-refusal-retry.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-registered-definition.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-repeat-press.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-repeated-stop.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-rest-test-rig.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-restart-interruption-test-harness.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-restart-status-publication.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-rewind-at-rest.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-rewind.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-rewind.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-send-idempotency.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-send-open-stale-turn.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-send-preparation.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-send-restarts-failed-start.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-settled-attach-retry.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-start-failure-writer.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-startup-reconcile-failure.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-stop-note-opened-turn.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-stop-note-withdrawn-send.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-surface-lifetime.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-thread-goal.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-thread-goal.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-turns.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-turns.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-unopened-tab.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-view-start-after-failed-start.test.ts
• src/main/native-chat/agent-session-wire/structured-agent-session-wedged-profile-migration.test.ts
• src/main/native-chat/agent-session-wire/structured-conversation-command.test.ts
• src/main/native-chat/agent-session-wire/structured-conversation-command.ts
• src/main/native-chat/agent-session-wire/structured-conversation-compaction.test.ts
• src/main/native-chat/agent-session-wire/structured-provider-session-ownership.ts
• src/main/runtime/orca-runtime-get-structured-agent-session-create-support.ts
• src/main/runtime/rpc/methods/structured-agent-session-adoption-replay.test.ts
• src/main/runtime/rpc/methods/structured-agent-session-at-rest.test.ts
• src/main/runtime/rpc/methods/structured-agent-session-hold.test.ts
• src/main/runtime/rpc/methods/structured-chat-tab-table.test.ts
• src/main/runtime/structured-agent-session-runtime.ts
• src/main/runtime/structured-claude-pending-rewind.test.ts
• src/renderer/src/components/native-chat/NativeChatStructuredSession.tsx
• src/renderer/src/components/native-chat/use-host-model-catalog-upgrade.ts
• src/renderer/src/components/native-chat/use-structured-agent-session-options.ts
• src/renderer/src/components/native-chat/use-structured-agent-session.cut-turn-notice.test.tsx
• src/renderer/src/components/native-chat/use-structured-agent-session.ts
• src/renderer/src/lib/structured-agent-session-launch-label.ts
• src/renderer/src/lib/structured-agent-session-launch-persistence.ts
• src/renderer/src/lib/structured-agent-session-provisional-tab.ts
• src/shared/agent-session-account-home.test.ts
• src/shared/agent-session-account-home.ts
• src/shared/agent-session-capabilities.ts
• src/shared/agent-session-option-catalog-claude-codex.ts
• src/shared/agent-session-option-catalog-types.ts
• src/shared/agent-session-record.ts
• src/shared/agent-session-turn-record.test.ts
• src/shared/agent-session-turn-record.ts
• src/shared/agent-turn-lifecycle-text.ts
• src/shared/structured-agent-session-mutation.ts
• src/shared/tui-agent-display-names.ts
• tests/e2e/cross-version-wire/cross-version-agent-session-wire.unit.test.ts
• tests/e2e/structured-chat-owner-status-activation.unit.test.ts

💤 Files with no reviewable changes (2)

• src/main/codex/codex-structured-session-adapter.ts
• src/main/claude/claude-structured-session-adapter.ts

Included review availability: This review used your included allowance. Your plan provides up to 10 included reviews per hour; 5 remain after this review.