mirror of
https://github.com/stablyai/orca.git
synced 2026-10-30 16:47:34 +00:00
c20971dbd109b096c4cfd8aa1ab5ae06e129fbcf
25
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 Brennan Benson
|
c20971dbd1 |
fix(native-chat): send launch prompts and worker tasks through the retrying start (#33832)
* Make opening launch prompts own the retrying chat start * test: observe launch inventory through surface publication * fix: keep blank chat launch prompts unsent * fix: bound launch seed filtering and update CLI replay fixture * fix: preserve encoded launch selections after filtering |
||
 OrcaWin
|
2847bbe9e8 |
[comment audit] Consolidate completed code comment cleanup
Combine the completed comment cleanup from 319 stacked PRs into one change against `main`, so the full CI suite validates the combined result once. This replaces the previous audit stack; no additional audit work is included. The change shortens redundant comments across 2,753 source and shell-snapshot files (25,235 net lines removed). Conflicts preserve current-main executable code. Runtime behavior, test assertions, directives, licenses, generated notices, SAFETY rationales, and compatibility markers are preserved. The audit tracker remains local, outside the repository. Validation: - Equivalent source/emitted AST and executable tokens across all 2,750 source files, with separate checks for embedded shell, SQL, JavaScript, CSS, and generated fixtures. - Identical executable lines in the three shell snapshots; protected comments unchanged. - Changed-code quality gate passed with no new findings. - 358 tests passed, 2 skipped across 23 focused test files. - `git diff --check` passed. |
||
|
Brennan Benson
|
6e973fa6d4 |
A launch can name chat or terminal: agent.launch gets a surface pick and the CLI gets --surface chat|terminal (#31191)
* feat(agent.launch): a launch can name chat or terminal, and the CLI gets --surface chat|terminal agent.launch gains an optional `surface` field, read by hosts advertising agent.launch.surface.v1. `terminal` always runs the agent's CLI; `chat` runs a chat even when the user's default is a terminal, and one the host can prove it cannot open is refused with a typed no-effects error before anything is created, so a retry answers with the same refusal. The pick is part of the replay fingerprint. `orca agent launch` and `orca worktree create --agent` accept --surface chat|terminal and refuse it on a host that would drop it, including the legacy worktree.create fallback. * test(agent.launch): pin the surface pick beside the other pre-start failures * fix(agent.launch): settle requested surfaces from resolved workspace facts * fix(agent.launch): describe fallback boundaries and correct test types |
||
|
Brennan Benson
|
251516de60 |
feat(native-chat): choose which agents open as a chat (#29071)
* feat(native-chat): choose which agents open as a chat Settings → Chat gets a Providers list. Each agent the host can open as a chat has a switch; off opens that agent in a terminal instead. Only choices the user makes are saved. Unchosen: Claude and Codex on, others on when their CLI is installed on this computer. The renderer's route and the host's launch-mode decision both read one rule, so the tab menu, shortcuts, CLI, orchestration and relayed phone launches agree. * fix: keep chat provider settings with the other chat settings types * test: route input carries this computer's detected agents * fix(chat): resolve provider choices from their owning computer * test(chat): use checked native inventory fixtures * Fix launch preparation lifetime and execution host startup * fix(chat): keep unchosen providers on without discovery * test(chat): type provider routing fixtures against store contracts * test(chat): align shared launch fixture with private settings reader * test(chat): give the structured-session RPC fixture the private settings reader The host now reads saved chat choices through requireStore().getSettings(); the shared RPC fixture only had getClientSettings, so the launch-mode reader saw no settings and create support refused Claude. The fixture now follows each test's client settings. * fix: preserve provider choice and cached native inventory |
||
|
Brennan Benson
|
635d3853f1 |
fix(agent.launch): avoid blank shells when launching new workspaces (#28451)
* fix: avoid blank shells beside new workspace chats * Carry promised agent chat through worktree creation * Use live agent launch inventory for new workspace surfaces * Type live agent launch owner from runtime contract * Assert one recovery shell after failed agent launch * Keep worktree activation adapters out of creation flow * Restore ordinary workspace setup handoff after launch correction * Cover the phone's agent.launch workspace create in the first-surface flow test |
||
  Jinwoo HongandClaude
|
5f134d8f01 |
feat(mobile): show and use the desktop's server workspaces on the phone (#26501)
* test(e2e): add a paired phone client helper and phone-mirror baseline specs
Later phone-mirror slices need one oracle that drives a real mobile-scoped device with the
phone's own capabilities against a serve host or a paired desktop. Pins #26022 across a serve
restart and today's gap: a desktop's phone does not list the desktop's server workspaces.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* feat(mobile): show SSH host health on the workspace row host badge
Rows on an SSH host the desktop cannot reach looked healthy on the phone. The badge now
adds the desktop sidebar's own word ("devbox · Disconnected"), derived by the same shared
function from the connection status ssh.listTargetSummaries already sends; old desktops keep today's badge.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* test(e2e): compose the phone-mirror topology and check the cold snapshot
Slices need one composer for host, desktop and phone with a single teardown, so specs stop
copying setup. The serve-restart spec now reads the first tab-stream snapshot after the cold
start, which is what the phone builds its tab strip from.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* feat(mobile): name an unhealthy SSH host even in a single-host list
The desktop card shows SSH status on any host count, so the phone now reads target summaries
whenever a repo is on SSH and badges an unhealthy host alone too. Labels and health share one
target-to-host keying, and a hook-level test pins the reply-to-row wiring.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* test(e2e): keep phone-mirror teardown errors from masking test failures
Teardown failures become a test attachment instead of replacing the real failure, and the
cold-snapshot poll reports stream errors together with the PTY list.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* fix(mobile): clear stale SSH host badges and announce them without the dot
A catalog that no longer spans hosts or holds an SSH repo now clears the health and label maps,
so a host that left cannot keep badging rows. The badge gets a screen-reader label ("devbox,
Disconnected") instead of reading the middle dot.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* feat(runtime): let a paired desktop register its phones as delegated mobile devices
A desktop paired to this runtime can now call pairing.delegatedMobileDevice.sync to make each of
its phones an ordinary mobile-scope device it owns, so the phone mirrors that host through the
desktop. Children are never pending, inherit the parent's reach, and die with the parent's grant.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* refactor(runtime): move the mobile device revoke primitive below dispatch
Dispatch binds the caller-bound delegated-device context, so the revoke path it needs now lives in
a layer under it as a pure move; this drops the abstract method seam. No behaviour change.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* fix(runtime): keep a runtime grant until its delegated phones are revoked
A failed child cleanup now leaves the parent in place so a retry can still find the children, and
revokeRuntimeAccess lives beside the revoke it cascades to. Adds restart, rename and bounds coverage,
shares credential minting, and drops the timing-dependent census.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* fix(runtime): fail a delegated sync whose dropped phone cannot be revoked
Matches the parent revoke: a child whose cleanup could not be saved is kept and the sync errors
before minting anything, so the desktop retries instead of believing the server matches its list.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* feat(mobile): relay a paired phone's server-workspace requests through the desktop
A phone mirrors its desktop, including workspaces on the desktop's configured servers. The desktop
now forwards a request that names `executionHost: runtime:<env>` to that server over one socket per
(phone, server), signed in as the phone's delegated device; replies pass through untouched except
stream ids. Hidden until the phone client ships: the relay capability is not advertised yet.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* test(mobile): type the single-host health fixtures inline so the test tsc passes
Fixtures hoisted into consts widened their health maps to string, which the now-required
hostHealthById rejects under the tests-typecheck ratchet; passing them inline keeps the union.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* docs(mobile): state the desktop relay contract in its entry module
main dropped docs/reference, so the relay's contract lives beside the code. Also pins that a
server whose delegated sync fails (e.g. delegated_device_revoke_failed) is unavailable until a
later request syncs it again.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* test(mobile): pin the relay's configured-server adapter
The production hosts adapter was only typechecked; this pins that it resolves configured servers
by exact id only (never a name or a removed server), fences by pairing revision and runtime
identity, and stops reporting retirements once unsubscribed.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* test(mobile): re-record the repo-metadata goldens for SSH host health
The settled metadata state now carries hostHealthById, and an SSH-only catalog now reads the
target summaries so a lone unhealthy host can be badged. Scoped re-record via rpc-recording.mts.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* refactor(mobile): tighten the desktop relay after elegance review
Reuses the shared E2EE socket opener, folds relay state into fewer fields, never sends a request
whose link ended while opening, passes failure codes through, relays only this desktop's own
phones, and drops the unused host-state reader until S2 needs one.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* fix(mobile): bound relay frames and stop sticky relay verdicts
Restores the 8 MiB frame cap on every remote runtime socket, re-checks an update-needed server on
each new link so an in-place upgrade relays, answers a phone the server just declined without
syncing again, and ends a link whose mirrored capability update cannot be queued.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* fix(mobile): run targeted desktop-owned calls on the desktop
The phone tags every call on a workspace screen with its execution host, so a
paired-desktop method carrying `runtime:<env>` now runs locally instead of
failing 'forbidden'. The Mac stays the only reader of the routing table.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* feat(mobile): list the desktop's configured servers and their last workspace rows for the phone
The phone needs the same host list the desktop shows, including servers that are offline or
too old to relay to. mobileRelay.hosts.list/worktrees answer from the desktop's own status and
its own worktree.ps fetches, cached per server and fenced by pairing, never from relayed replies.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* refactor(mobile): list server rows through one desktop fetch, with no side effect on listing
The phone now always lists a server's rows through mobileRelay.hosts.worktrees, so the
desktop's own fetch is both the live source and the cache, and stamping stays in main.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* fix(mobile): fetch a server's full workspace list within a bound, and forget removed servers
The desktop now asks for the phone's full-catalog limit, answers within 5s with the last rows
when a server hangs, drops a server's rows when it is removed or re-paired, and checks the
server's identity after the fetch so a reply from before a re-pair is never served.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* feat(mobile): name a server workspace's host on the phone's calls
A phone reaches a server's workspace through its desktop only if each call names the host; the
direct, cloud-relay and shell-bridge transports now carry `executionHost`, streams keep it for
replay and unsubscribe, and an older shell (no `executionHost` grant) refuses rather than strips it.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* fix(mobile): list managed servers as the desktop sidebar does, and hide rows mid-delete
A server Orca manages on an SSH host now lists once under that host's name, or not at all while
the host is reached over the SSH relay, using the sidebar's own folding. Rows the server is
deleting are left out at reply time, as a phone asking the server itself would see.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* fix(mobile): keep a disconnected server's last rows, and never join a re-paired server's fetch
Retirement also fires on a manual disconnect or managed-server update, which wrongly emptied
the phone's list; rows are now pruned only once a server is no longer configured. Fetches are
shared per server identity, so a poll after a re-pair asks the new server right away.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* feat(mobile): list the paired desktop's server workspaces on the phone
Fixes the phone missing a desktop's `orca serve` workspaces: rows come from the desktop's
mobileRelay.hosts.* with each server's health ("Update needed", "Disconnected"), open through the
relay, follow the desktop's hidden hosts, and the desktop now advertises the relay capability.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* test(mobile): pin that a desktop failing the hosts list leaves only its own rows
The page asks a relaying desktop for its servers; a refusal such as method_not_found must leave
today's list with no server rows and no catalog error.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* fix(mobile): mirror the desktop's shown hosts exactly, one host included
Jinwoo's call: the phone follows visibleWorkspaceHostIds as the desktop sidebar does, so a
one-host list narrows the phone too instead of being read as a single-host focus.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* fix(mobile): scope a server workspace at its route, never the desktop
Every screen under a server workspace's route now reads one context and
gets that server's client and a router that keeps pushes on it, refusing
to fall back to the desktop; pins write to the row's owner like the Mac.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* fix(mobile): never open an unreachable server workspace, and prove the relay in e2e
A server workspace the phone can't reach now gets one explicit screen at its route
and no row tap or action-sheet entry opens it; native routes read the route's host,
a census pins every workspace route, and the e2e reads files and git on the server.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* feat(mobile): gate a server workspace on that server's own features
A targeted status.get now relays to the server, so a server workspace's route
provides that server's status to every screen under it and its action sheet,
while whether the phone can reach the server stays the desktop's answer.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* fix(ci): run the remote-runtime WebSocket opener test on Node
Under the bun vitest project, `ws` resolves to Bun's built-in client, which
ignores maxPayload, so the frame-cap test hung. The opener only ships on
Node/Electron, so run its test in the Node runtime project like its siblings.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* fix(mobile): give the file preview the workspace router, back to 128 page assets
The preview was the one workspace screen on the plain router, so the scoped router
was reached by a route set of its own and esbuild split it into an extra chunk
(129 assets); every workspace screen now uses it and it shares the route's chunk.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* refactor(mobile-relay): derive the phone allowlist from the routing table
Adding a phone method now takes one edit (its route). A server-targeted call on a
legacy non-E2EE socket is refused instead of running locally, and only
execution-host methods enter the relay.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* refactor(mobile-relay): drop unread freshness fields from server workspace lists
The phone reads only the rows, so stale, fetchedAt, totalCount and truncated
were dead wire weight. The desktop still serves a server's last rows while
the server is unreachable.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* refactor(mobile): read only the rows of a server's workspace list
The desktop no longer sends a freshness flag, so the phone's schema drops it.
The relay capability list names the delegated-devices constant instead of
repeating its literal, which also retires a now-circular type check.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* fix(mobile): show a server workspace as disconnected while its server is down
The phone read the desktop's connection state for a server workspace, so a stopped
server kept a green dot and "Loading tabs" spun forever. The workspace route now
narrows that state by the server's health from mobileRelay.hosts.list, reusing the
existing disconnected session UI and refetching tabs when the server returns.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* refactor(mobile): share one desktop server-list feed between the host list and workspaces
A server workspace polled mobileRelay.hosts.list beside the host list's own read, so
the phone could ask the desktop twice per tick and the row badge and the screen could
disagree. One feed per desktop client now owns that read; both readers join it.
Claude-Session: https://claude.ai/code/session_01JNnE9qzUZMMnqpZWCqM3nb
* test(e2e): read a server's reachability from the hosts list, not a dropped stale flag
|
||
|
Brennan Benson
|
39a01f9c0b |
feat(agent-launch): create a folder workspace and start its agent in one host operation (#26083)
* feat(agent-launch): accept a create-folder-workspace target agent.launch can now create a folder workspace and start its agent there, the same operation it already runs for a new git worktree. The host creates the folder workspace (metadata only, no startup terminal), resolves it the way an existing target is resolved, then settles the mode and builds the surface as for any existing folder workspace. Hosts advertise agent.launch.create-folder-workspace.v1 so a client offers the target only to a host that accepts it. * fix(agent-launch): authorize, classify and navigate folder creates like their create method A create-folder-workspace launch could do three things its two-call equivalent never did: - A phone could create folder workspaces. The mobile allowlist admits agent.launch but not folderWorkspace.create, and is checked by method name only. A create target now needs the authorization of the create method it stands for, read from the same allowlist, and a phone is refused with agent_launch_target_forbidden before admission, so no ledger row and no create. create-worktree maps to worktree.create, which phones already may call, so nothing changes for it. - A create refused before anything was stored (group gone, folder missing, SSH host unreachable) replayed as agent_session_operation_unknown, and a retry under the same operation id was refused until the row expired. runtime.createFolderWorkspace now raises those refusals as FolderWorkspaceCreateRefusedError, with the same message text, and the launch records it as failed for any target. A failure after the create still replays unknown. - A paired caller's folder create activated the new chat on the host window and did not select the caller's own tab. Only create-worktree, whose host activation runs setup, keeps the host navigation now; a folder create navigates like an existing workspace. Tests cover the phone refusal, replay of a success, a refused create settling failed and replaying that failure, an after-create failure staying unknown, an SSH connection reaching the agent's view mode, paired navigation, and an older host refusing the target at parse. * fix(agent-launch): replay a refused folder create with its whole path A folder create refused before anything was stored is recorded as the launch's failure, and a retry under the same operation id answers from that record. The record went through the 128-character cap every launch failure code gets, and folder refusals carry the folder's absolute path, so a retry could show the user a path cut short: a different folder. The refusal is now recorded whole when it fits a path as long as Linux's PATH_MAX after its code. Past that, which no folder a stat accepts reaches, only the code before the path is kept, so the replay degrades to the bare code instead of a partial path. Also: the controller test asserts the exact refusal message, the create-folder-workspace capability comment says a phone needs its own gate, and the forbidden-target code says why it is not `forbidden`. * test(folder-workspace): type the controller test's store stub without an assertion The changed-code quality gate refuses the `as never` cast on the stub store. It is now a RuntimeStore whose unreached members throw. * test(agent-launch): run the folder-create launch tests in the Node runtime project The operationId replay tests open the real SQLite record store, so the file belongs with the other ledger-backed agent.launch tests. * test(agent-launch): pin that a paired folder create's terminal stays off the host window The folder-create test now asserts surfaceOwner: false on the terminal a paired caller's launch creates, rather than matching any options. The caller-selection header now says only an in-process worktree create keeps worktree.create's navigation. * test(agent-launch): drop the per-connection view-mode expectation main removed |
||
|
Brennan Benson
|
22974c5809 |
Default experimental native chat to structured chat (#22933)
* refactor(native-chat): make structured chat the only Chat UI runtime Retire the experimentalStructuredNativeChat opt-in. Chat UI on is now the one gate for structured chat: new agent tabs that default to Chat UI start a structured session wherever the launch is supported, and unsupported launches keep falling back to terminal-backed chat. - Shared isNativeChatEnabled predicate replaces prefersStructuredNativeChatByDefault; host admission, renderer routing, orchestration mode, startup projection, tab sync and the resume offer all read it. - The retired key is stripped on load and on save, dropped from the settings type, defaults, the published client-settings projection and telemetry. - Settings no longer offer the structured toggle; resume-after-restart and shell-environment rows show under the Chat UI default view. - Startup structured-tab projection moves beside its sync hook so both gates read the same predicate. * docs(native-chat): describe structured chat as the Chat UI runtime New local Claude and Codex chats no longer sit on a PTY, so the page stops saying the terminal is always the source of truth and scopes the chat/terminal toggle to terminal-backed chats. * fix(native-chat): stop promising a Windows fallback in the Chat UI scope note A host refusal after the route picks structured chat still ends as a failed launch. * fix(native-chat): keep open structured chats reachable when Chat UI is off Turning Chat UI off hid open Claude/Codex chat tabs while their provider processes kept running, because the host and the desktop mirror gated serving existing sessions on the same setting that decides how new launches open. The host now asks two separate questions. Creating a new structured session (agentSession.create, createSupport, ensure) needs the negotiated capability plus the Chat UI setting. Serving a session that already exists (listing, reading, sending, resuming, closing, and the mobile projection and restore) needs the capability only. The desktop mirror, startup restore and resume-on-restart offer no longer read the setting, and the toggle-off teardown and the mobile re-publish on setting change are removed as dead. * fix(native-chat): run the local structured tab sync only where chats may exist The previous commit started the every-worktree session-tab census and subscription, the startup restore, and the resume-offer host read for every desktop user. Gate them on a derived fact instead: Chat UI is on, or the local host holds at least one structured session record. With Chat UI off nothing can create a session, so the answer can only go from yes to no. The host answers from its installed record store, or from whether the store file exists when the host is not installed, so a machine that never had a structured chat pays one existence check and never starts the host. Chat UI turning on starts the sync; turning it off never stops it. * fix(native-chat): answer an empty restart offer without starting the chat host With every Chat UI user now on structured chat, the launch-time resume offer read built the host even on machines with no stored chat, which on Linux scans every process environment. No store on disk means no offer, so the listing answers empty without installing the host. * fix(native-chat): release the structured tab sync when it never subscribed A failed capability probe reads as unsupported, so the sync returned without subscribing while still holding the one-sync slot. Nothing could start another, and chat tabs stopped live-updating until relaunch. The start now reports whether it subscribed, and a sync that did not releases the slot so the next Chat UI change retries. * fix(native-chat): let the host answer the restart offer whatever Chat UI says The renderer asked main whether a structured store existed before reading the restart offer, so it would not start the chat host for users without one. The host now answers an empty offer without starting itself, which made that probe an extra round trip. In paired web it was also wrong: the probe is hard-coded to no and the gate read the browser's own Chat UI setting, so a paired host's chats were never offered back once that was off. * refactor(native-chat): drop the structured tab sync generation fence The fence existed so a Chat UI toggle-off could retire in-flight inventory replies before it cleared the mirror. That teardown is gone: turning Chat UI off keeps open chats, and only a test reset still bumped the generation, so every production check compared equal. The sync's own disposal still fences its subscription, and the startup restore latch and publication cursors stay. * fix(native-chat): retry the structured tab sync when the capability probe fails A failed runtime status probe maps to an empty capability list, which the sync read as "unsupported" and gave up on. The release only let a later Chat UI change try again, so one transient startup failure left chat tabs without live updates for the whole session. The sync now tells an unknown answer apart from a host that answered without the surface, re-asks unknown answers on its existing subscribe backoff, and stops only on a definite no. * refactor(native-chat): begin single-switch structured launch routing * Make Chat UI the single structured launch default * Remove retired chat settings copy from localization catalogs * Keep locale checks aligned with the single Chat UI switch * Keep explicit chats accessible and older client defaults terminal * Align newly landed Settings tests with the Chat UI switch * Expect terminal fallback on older paired structured hosts * Correct terminal fallback launch expectation * Update tests that still expected the retired chat settings * Keep chat composer options off terminal launches and restore route coverage Chat composer model and effort no longer become CLI flags on terminal launches, since a new launch can no longer open in the terminal chat view; structured launches seed their own options. Restores dropped route coverage, including that an explicit resume-in-new-chat request stays available while Chat UI is off. * Replace the stale remembered-options test with terminal fallback coverage The test expected chat composer model and effort on a codex launch with Chat UI on, but its mocked host has no structured capability, so that launch is the terminal fallback that must not carry them. It now checks the fallback with Chat UI on and off, with shipped and stored arguments; structured creates are covered where the seed is applied. * Lower the terminal pane listener budget to the measured 15 The explicit-chat change retired the per-pane Chat UI setting read, so each mounted pane opens one fewer store listener. Pin the budget to the measured count and count that read among the saved listeners. |
||
|
Brennan Benson
|
c71601f51c |
Add Pi structured native chat through its RPC mode (#25851)
* End a running call as its turn's journal row ends
A call still running when its turn ends takes the state of that turn's
row: a row another writer settled first (a person's Stop) stands, so its
calls read interrupted whatever the provider's later end reports. The
no-ending path that settled calls from the Stop row is gone, since a Stop
now leaves running calls to the provider. Adds the two Spanish strings.
* Say why a Grok turn failed, and keep task rows in Grok's own words
A failed Grok turn ended with no reason on screen: the translator dropped
every copy of Grok's message. The failed turn now gets one status row in
Orca's existing "provider did not accept this message" words with Grok's
reason, read from whichever copy arrives first (the given-up retry, the
turn's end, the prompt's completion notice, or the prompt's error answer);
later copies only fill a reason the row still lacks.
A running background command no longer reads "Background task <id>
started": a task's summary is mapped only once it has settled. A monitor
stays a monitor when the agent reads its output: a frame that names no
kind keeps the known one, and a "[monitor" command is a monitor.
A prompt's turn is marked started, so a late frame for an ended prompt
neither reopens it nor becomes the active turn. A tool's turn is held in
one place at a time.
* Read a monitor from Grok's exact output prefix
* Word a failed Grok turn in Grok's own text, not as a refused message
A turn that started and then failed was told "The provider did not accept
this message", Orca's sentence for a message refused before its turn. The
row now reads as a Codex turn-ending error does: an error status row with the
provider's own words. With no words, the dialect names the failure ("Grok
ended this turn with an error." / "Grok usage limit reached."), else the
agent's display name does.
* Settle a stopped turn's running call as its turn row ended after a restart too
The restart sweep ended every running call by the death evidence alone, so after
a person's Stop with no proof the child died the call read failed under a turn
that read interrupted. The sweep and the live dead-generation settlement now ask
the same rule the assembler does: a call in a turn already settled ends as that
row ended; only a turn still running leaves its calls to the evidence.
* Keep the dead-generation settlement under the line cap
* Register the ACP schema verify step in the PR preflight phase test
* refactor(agent-session): one required agent registry; declarations admit what they claim
A StructuredAgentRegistry, built once from the {definition, adapter}
registrations, is now a required host dependency and the router routes with
it. The adapter interface loses its optional router-only capabilities?() and
definition?(); every reader (options at rest, thread goal, rewind, the
/compact handover) asks the registry. A live session still narrows rewind
through the adapter, and the host combines declared and narrowed in one
helper. The registry refuses a registration that declares compact, a thread
goal or rewind without the adapter method behind it.
/compact is admitted by the declared capability, so an agent that declares
compact:false gets the commandRefused fact instead of a thrown error.
Also: the cut-turn notice names the agent from the catalog, create-support
builds the account home through agentSessionAccountHome, and the turn
status text older clients read names the session's agent instead of
defaulting to Codex (Claude/Codex text unchanged).
* Read ACP permissions, session events and prompt errors through the protocol client's own types
The translator now reads a permission request with the client's lenient reader, a session update
with its session-event reader, and takes only the agent's own error answer as a failed prompt's
reason, so an Orca-side error never reads as the provider's words. Tests cover protocol values
newer than this build.
* refactor(agent-session): the router applies the declared rewind itself
The router holds the registry, so its rewindSupport answers the owner's
declared rewind narrowed by the adapter, in one place no reader can bypass.
The host-side combining helper is gone; readers ask the adapter they hold.
* test(agent-session): register the agents the merged-in tests now need
The adoption replay test builds its host over this build's registered agents
instead of an adapter definition method, and the stored-form test passes the
stored agents the record guard now requires.
* chore(agent-session): the registry is the one lookup; drop the test-only empty capability record
* fix(agent-session): keep "dismiss all" restart offers dismissed for the desktop
The desktop renderer calls the restart methods as a paired runtime client
without the registered-agents capability, so it was handed a Claude/Codex
audience and its "dismiss all" took the scoped path: no dismissal fence and
no clearing of unwritten teardown witnesses, so a late teardown write could
bring a dismissed offer back.
The audience is now derived against this host's registry: a client that can
show every agent the host registered gets none, and dismisses exactly as
before (one fence for every offer). A client that truly cannot show some
agent still gets a final dismissal for what it sees: each cleared session is
fenced on its own (bounded, superseded by a dismiss-all fence) and only the
witnesses of agents it sees are dropped.
One predicate now answers which agents a client renders, for both tab
projection and restart offers; the Claude capability rule is folded in.
* fix(agent-session): a changed agent definition never hides that agent's chats
A record was readable only if every handle used the transport its agent's
definition declares today, and its account variable was declared by some
registered agent. A later build that drives the same agent over another
protocol, or under a renamed variable, would have set every existing chat of
that agent aside: no tab, no history, no error.
Readability now asks only that the record agree with itself: its agent is
registered, every handle shares one namespace owned by that agent, and the
account variable is a well-formed name. Whether this build can drive it (the
chain's transport is the one the agent speaks, and the account variable is
that agent's own, since it becomes the child's environment) is decided in
performAttach, the one admission every start of every agent passes through,
and refused there as hostUnsupported. A record pinning another agent's
variable is refused the same way rather than passing because some agent
declares it.
* refactor(agent-session): each agent's registration says where it runs and which account it pins
createSupport, create and the model catalog's account read still chose a
location rule and an account resolver by name (Claude, Codex, else none),
so a new agent would have needed a third branch beside the registration
list that already decides storage, routing and publication.
Each runtime registration now carries supportsLocation(location) and
resolveAccountHomePath({ launchEnv, location, purpose, workspacePath }),
and Claude's and Codex's rules move into their entries unchanged: a read
still syncs no home and starts no bridge, a Codex launch still trusts the
workspace first, and WSL locations resolve as before. The list exists
before the host is built, so none of these reads installs the host, and an
agent the list does not hold is answered no without opening the journal.
* test(agent-session): drop a duplicate registry key; keep fences out of the capsule state
A spread input already carries the registry, so the explicit key was
overwritten (TS2783). The session-fence helper now returns only the fence,
not the capsule state it was given.
* fix(agent-session): a scoped dismiss-all persists no per-session fence
The per-session dismissal fence added a forever-persisted capsule field
with an arbitrary cap that served no caller: only the local desktop calls
the restart methods, and on a host whose agents it all shows it gets no
audience and takes the unscoped, fenced dismissal. A scoped dismiss-all
(a caller that cannot show some registered agent) keeps that agent's
offers, clears only its audience's unwritten witnesses, and writes no
fence; a late write from this process is serialized behind it. The field
never left this unreleased branch.
* fix(agent-session): refuse an attach whose agent is not the session's own
The start check judges the record's agent while the router starts the adapter of params.agent, and the attach wire let the two differ. Admission now refuses agent !== provider as requestMalformed, so the agent checked is the agent started. Every host-built attach already sets them equal.
* fix(agent-session): offer to start a chat only when the start would accept it
The restart offer, a failure's retryable flag and the pre-send check asked only whether the adapter runs the chat's location, while the start also refuses a record this build cannot drive; such a chat was offered Resume and its Retry failed forever. One host predicate, hostCanStartRecord (location support and agentDrivesSession), now answers all four; adapterSupportsRecord stays the reader gate, and an undrivable chat's offer is kept, not retired.
* fix(agent-session): the model-catalog probe uses a record's account only if this build drives it
A record-scoped catalog read started the agent's lister under the record's account path whatever variable the record pinned. It now uses that path only when agentDrivesSession holds, and otherwise resolves the account as for a read with no record.
* refactor(agent-session): the record store admits agent ids; comments say where transport is checked
The store only ever asked whether an agent is registered, so its admission list is now the registered ids; the definition is the one home of an agent's transport and account variable. Comments that said the record store checks transport, and one that cited a nonexistent function, now point at agentDrivesSession and the attach admission. The isPersistedAgentSessionRecord(value, agents) call shape and the test fixture the cross-version probe imports are kept.
* docs(agent-session): the record store admits the registered agents' ids
* refactor(native-chat): Grok's registration declares where it runs; ACP no longer borrows Codex's location rule
The rule a self-supervised agent child runs under (this machine, no WSL, Windows only with process
start-time proof) is its own module that Codex and the ACP adapter both use. Grok's registration
takes the full account-home resolver signature, and D3's tests build hosts with the agent registry.
* fix(native-chat): Grok follows the ACP runtime's request contract and the managed process's close
A request the agent or a Stop cancels is answered with the agent's own cancelled reply by the code that
owns it (the runtime no longer answers a silent handler), so a Stop needs no separate decline pass. A
permission answer still being saved when the agent stopped waiting is reported unconfirmed, since the
protocol already answered it cancelled. Cancelling the agent's own turn is the plain cancel. Request
rows are matched under their generation-scoped ids. A refusal's reason comes from the dialect's wording
path. The child drops its own stderr tail and close policy for the managed process's, and a close
whose process tree was not proven gone is reported as the adapter contract asks.
* fix(native-chat): a Grok chat Orca already holds resumes without writing what Grok replays
A chat with a saved Grok session reattaches with session/resume where the agent offers it, else
session/load. Either way the call runs inside the translator's load window, so what Grok sends while
it reattaches (its saved exchange, a task the dead process left running, ended by the restart) opens
no turn and writes no row; only context usage reads on. A reply an Orca or Grok crash cut short is no
longer completed from Grok's saved history: it reads like a Claude or Codex chat's, with the existing
notice. The attach window also closes after a failed attach, and a created session that session/resume
reports missing is replaced like one session/load reports missing.
The replay reconciliation is removed: the lane no longer reads the journal, and D3's replayed-input
grammar test and completed-turn check in the assembler go with it.
* refactor(native-chat): a failed Grok reattach needs no window close of its own; its lane is replaced
* test(native-chat): D3's merged tests use the shipped declarations and the launch options main requires
* fix(native-chat): typecheck fallout of the base merges; any agent's empty chat is reusable
Main's idle-empty-chat lookup and launch join now take any registered agent, as the rest of the
launch path does. The refusal check moved into the prompt turns and the prompt-block conversion beside
the turns that send it, keeping both files in their line limit.
* fix(native-chat): a Grok Stop ends the process once Grok settles its turn; the next send resumes
Grok's session/cancel ends only the running turn: work it already moved to the background keeps
running and can begin a turn of its own after the person pressed Stop. Stop is now a session
boundary, as it is for Claude: the cancel answers open requests and lets Grok end the turn, the host
waits a bounded grace for that, then ends the process; the next send relaunches and resumes.
The adapter's own bounded close of a turn Grok began is gone. Its named-turn check stays: the host
ends the session unless the provider declines a Stop naming a turn that has since ended.
* test(native-chat): a Grok Stop ends the process only after Grok answered the cancel
* fix(native-chat): Steer on a Grok card cancels the running prompt, then sends it
A send that reached Grok while a prompt ran was held in the adapter until that turn ended: Steer
on a queued card took the card out of the host's editable queue and meant 'send after this turn'.
It now cancels the running prompt (session/cancel; the session stays) and sends as the next prompt
once Grok answers the cancel, as the common pattern does; a steer behind another cancels it in
turn, so the last one runs. The adapter holds a send only while that cancel lands, so its general
held-send queue and its holdsDispatch report are gone (every send it holds has its turn open in
the journal). An older client's mid-turn send takes the same path. capabilities.steering is
unchanged and still unread.
* refactor(native-chat): a close or Stop cancels a start through the acquire's own abort signal
The host owns the acquire it runs, so it now owns its cancellation: each attach's acquire gets an
AbortSignal, aborted from outside the session's queue by a close and by a Stop admitted now (the
same admission rule as before). The optional abandonStart adapter hook, the router's fan-out to
every adapter and the ACP adapter's session-keyed start map are gone; the ACP adapter keeps an
unkeyed set of starts only so quit can prove their children gone, and keeps a failed start's
unproven child until its exit is proven.
The hook also let a later close ask that child again. The host now does that from state it holds:
a close of a chat with no live child whose record still names an owner process with no death
evidence asks the adapter to release it. The answer is not recorded as proof (the lease probe
does that), so an owner pid an earlier Orca left is never killed or marked gone. Claude and Codex
ignore the signal and hold no such child; their release is a no-op (tested).
* fix(native-chat): a Grok crash that closes stdout before its exit still ends with Grok's last words
On macOS and Linux the agent's stdout ends before its exit is observed, with or without the
supervisor's EOF forwarding, so the connection's loss closed the journal first and its error text
became the session's ended reason, dropping Grok's stderr. The reason is now read at the proven
exit: the agent's last words when it left any, else why the connection closed. The failure already
carried them. Comments that assumed the exit comes first, that early frames past the cap refuse the
start, and that dispatch re-checks image support are corrected.
* fix(native-chat): nothing Grok sends while a held chat reattaches is written, marked as replay or not
The reattach window relied on the dialect's replay verdict, and Grok's frames read as live unless
they carry isReplay, so an unmarked chat frame during session/resume opened a turn that never
ended. D3 now marks every frame inside the window as replay before the translator reads it, so the
translator keeps only context usage whatever the agent marked; options and commands are still
adopted. The translator's load semantics are unchanged.
* test(native-chat): a Stop after a resume finds no turn an unmarked old reply opened
* test(native-chat): a resumed Grok chat keeps its last context reading; the resume refreshes only the window
* test(native-chat): a Grok background task a Stop ended reads as stopped reporting
* refactor(native-chat): quit's stop of each start answers through one promise kind
* fix(native-chat): quit aborts every start the host has in flight before draining attaches
A Grok that never answered its handshake held quit until the start's own 60 s bound, past the
20 s quit deadline. The host's teardown now aborts each in-flight acquire (and any the drain
still begins), so the adapter's own quit controller and its map of starts are gone: a start
has one canceller, the host's signal.
* fix(native-chat): a Grok start's abort stops reaching its child once the start has returned
The listener stayed on the host's signal until the attach finished committing, so a Close in that
window killed the now-live child behind the host's back and it read as Grok crashing. The start
now detaches it when it ends; a later Close goes through the session's own stop.
* fix(native-chat): a close or Stop during any attach phase stops the start before it launches
The attach began its abort controller only after reconciling leases, resolving recovery and
probing the previous owner, so a close or admitted Stop in those phases reached nothing and Grok
launched anyway. The controller now begins first, and the acquisition checks it before asking the
adapter to start.
* test(native-chat): a close during the attach's owner probe asks no adapter to start
Also renames the close test after the hook it no longer exercises.
* test(native-chat): a close's re-ask closes a Claude or Codex child a failed cleanup left
The re-ask is not a no-op for them: when the adapter still holds the child its cleanup could not
prove gone, the close stops it again as a requested close, and Claude persists the handle of the
conversation it ran so the next send resumes it. Corrects the tests' and comment's wording; the
close awaits the re-ask, bounded by each adapter's kill ladder.
* fix(native-chat): Steer during a turn Grok began itself cancels it and sends once it ends
A send while Grok ran a turn of its own (a background task waking it) went straight to Grok, which
queued it behind that turn where Orca could no longer withdraw it, while Stop treated the same turn
as the running reply. The send now waits as a steer, the turn is cancelled once, and the message
goes when the turn ends; a Stop withdraws it and an exit rejects it as never sent.
* test(native-chat): a steer whose cancel Grok never answers ends Grok and is rejected as never sent
Pins the bounded steer cancel kept from the runtime: past the bound the connection closes, the
running reply reads unverifiable, Grok's end reads as its exit, and the waiting steer is rejected
as never sent.
* fix(native-chat): a Grok crash stays a crash when a stop lands before its exit is proven
After the connection broke and the close could not prove Grok's exit, any later stop Orca asked
for (the next start, a Stop, a Close) marked the child as closed by Orca, so the crash read as a
requested close and Grok's last words were dropped; a send meanwhile was recorded unconfirmed.
The connection loss now decides the cause, and a send on that session is rejected as never sent.
* test(native-chat): fixtures this PR's registered Grok and desktop capability made stale
CI's unit shards failed on tests outside the PR's own lists. Each encodes something this PR changes
on purpose: Grok is now a registered agent (the seam test's unregistered agent is now Cursor); the
desktop now advertises registered agents (the restart-offer tests' older client drops that
capability explicitly); the attach context carries the start's abort controllers (the forget-status
double gains them); and the ACP real-host test rig sends to the host directly (listed beside the
other real-host rig in the send ratchet).
* fix(native-chat): a start quit stops is not the queued message's start failure
With quit now aborting a start it would have waited for, the delivery step recorded the aborted
start as the message's failure ("couldn't restart"). After quit has stopped delivery, the step
leaves the message to quit, which settles it as a close does ("The chat closed before this message
was sent."). The test that pinned quit waiting for that start and stopping its child now pins that
nothing is launched behind quit.
* fix(native-chat): a message sent after a Stop or close aborted a start gets its own start
A start the host aborts (an admitted Stop, a close, or quit) returned its refusal to the delivery
loop, which then rejected whatever was queued at that moment with "couldn't restart", including a
message the user sent after the Stop. The attach now reports that the host aborted it, and the loop
re-derives from the journal instead: what the Stop or close withdrew is already settled, a message
accepted since gets a start of its own, and quit's next step stops the loop. This replaces the
quit-only carve-out with the same rule for every abort and every agent.
* test(native-chat): the message sent after an aborted start is answered, so no settlement outlives the test
* fix(native-chat): a Grok model pick Grok never answers no longer holds Stop or Close
The pick runs on the session's queue. It now registers in the host's out-of-queue
abort registry beside a start, so a close, an admitted Stop or quit abandons it, and
the ACP adapter bounds it at 30 s like Claude and Codex. A late answer is still adopted.
* fix(agent-launch): a phone's launch opens a terminal for an agent whose chat it cannot show
agent.launch now reads the caller's capabilities by the rule tabs and restart offers
use (clientRendersStructuredAgent). A phone without registered-agents.v1 gets Grok as
a terminal again, as on main; the host's own callers and desktop clients are unchanged.
* fix(acp): strip every agent hook variable from the ACP child, from the shared list
ACP_CHILD_ENV_TO_DELETE was a second copy of the hook runtime keys that missed
ORCA_AGENT_HOOK_TRANSPORT; it now spreads AGENT_HOOK_RUNTIME_ENV_KEYS beside the pane
identity keys.
* refactor(native-chat): the mutation context carries the provider-wait registry itself
Keeps the host file within its line limit; one field instead of two closures over it.
* fix(agent-launch): agent.launch.v2 still vouches for Claude and Codex chats
The caller rule from the previous commit also turned Claude and Codex into terminals
for a client advertising only agent.launch.v2, whose contract says it opens a chat
(mobile retry-authority tests). Only an agent beyond those two now needs the client to
read it (clientRendersStructuredAgent); the test fixtures go back to what they were.
* refactor(native-chat): drop saved-history adoption from the timeline assembler
The common pattern discards the history a provider replays while loading a
session, so the assembler has no use for an input.history event.
* refactor(acp): drop session/load history adoption from the translator
The common pattern discards the history an agent replays during session/load,
keeping only what it says about the context window. Remove the adoption path
(acp-history-adoption.ts, the adopt option, and the historical background-task
liveness rewrite it fed) so load replay is always dropped except usage.
* refactor(native-chat): a pending input is only Orca's send now
Review follow-up to the adoption removal: drop the comment naming the
provider's saved message, and make requestedAt required since every pending
input comes from input.accepted.
* test(acp): keep the task-result status table on live frames
Review follow-up to the adoption removal: the result-status mapping was only
tested through adopted history, so run the same table on live frames, and
cover an unmarked task notice during a load being dropped.
* test(acp): a frame helper for a shell command Grok is running
* fix(acp): a Grok crash settles through the host's provider-exit batch, scoped to the turn it ended
A Grok crash ended the journal unverifiable before the adapter reported the exit, so the host's
provider-exit settlement found no running turn and wrote nothing: the adapter's failure (with
Grok's last words) never reached the journal, and a later stale-session pass wrote a bare,
thread-scoped cut-short row, so the partial reply was not folded as Claude's and Codex's are.
At a proven exit the ACP lane now ends its running turn interrupted at the exit instant, as the
host's exit contract expects of a child's own translator (Codex's does the same). When Grok's
stdout closed first (every POSIX crash), the turn is unverifiable only until the exit is proven:
the host's provider-exit settlement now takes the exit as proof naming the child's fence and
revises what that child left unverifiable in the same batch, with the turn-scoped row and the
adapter's failure. Claude and Codex write no unverifiable turn of a live child except a command
whose hand-off is in doubt; that turn is now revised at the exit instead of at the next open.
* test(acp): a crash seen first leaves the host no Grok turn to revise
* refactor(native-chat): what a gone generation left unfinished gets its own module
The settlement file passed 300 lines with the exit-proof revision. The unfinished-work reads
(capture, interrupted-by-the-exit, in-progress) are their own concept and move out unchanged,
apart from the exit proof they now take.
* refactor(native-chat): a watched exit revises what its child left unverifiable without reading Stop marks
An exit's own instant is the turn's end, so the revision needs only each row's fence: the
settlement's journal type gains itemFence alone, and the host test fakes say so.
* test(native-chat): drop the duplicate itemFence on the fake that already had one
* test(claude, codex): an exit whose stdout ended first still reports as it always did
The provider supervisor now ends Orca's stdout when the agent's ends, so on every crash EOF
arrives before the exit is seen. Claude's and Codex's connections report nothing at EOF and
report the exit, with its usual reason, once it is seen.
* fix(acp): reopen a chat with session/load, as the common pattern does
An agent that offers both now reloads its session instead of resuming it; the
reattach window still discards what it replays except context usage.
* fix(acp): drop the 60 s handshake bound; an abort fails the start's waits at once
Neither common design bounds an ACP handshake: Close, Stop and quit end a start
that never answers. The abort now also closes the connection, as a kill there
does, so the start settles even before the child's exit is proven. The
host-stopped start refusal only this bound produced goes with it; the idle
sweep keeps its words.
* fix(acp): a Stop naming an ended turn follows Claude's rule
It still stops nothing while another turn is live, but in the gap before a
follow-up's turn opens, which no client can name, it now stops what is in
flight and the session ends, as a Claude Stop does.
* fix(native-chat): a close no longer re-asks a failed start's unproven child
Neither common design retries that stop at Close, and Orca's Claude contract
re-asks only at the next start and at quit. The ACP adapter keeps the child
until its exit is proven and asks it again there, as Claude does.
* fix(acp): a message sent during a turn the agent began itself goes at once
Both common designs send it straight to the agent with no cancel; only Orca's
own running prompt is steered (cancelled, then re-prompted).
* test(native-chat): dismiss-all through a remote client's audience keeps a newer Orca's offer
Uses an audience production sends (one that cannot show every agent), per review.
* fix(acp): launch Grok as `grok agent stdio`, without the update and leader flags
The common pattern passes neither --no-auto-update, --no-leader nor
GROK_DISABLE_AUTOUPDATER; full access still adds --always-approve.
* fix(acp): an agent that ends its stdout, or answers unreadably, is not a lost connection
As in the common pattern, only a broken stdin (or Orca's own close) ends the
agent; one that closed its output but can still be written to stays until a
Stop, a close or its exit. The provider supervisor goes back to its base
content, so Claude and Codex no longer get the forwarded stdout end either.
* fix(native-chat): a person's close joining a failed one still binds the turn its child end cuts
On main every close of the chat writes its own Stop and settle. Here a later close joins the
first and writes no row, and the first's settle closed when its kill failed, so a turn that opened
in between and was cut by the next close read as failed. A person's close joining a person's close
whose Stop opened a settle now reopens that settle until its attempt is done.
Tests: a close whose kill failed still closes its settle; a turn opened between a failed close and
the next reads as the person's cancellation (each fails without its half of the fix).
* test(native-chat): Grok opens as a chat only behind the structured-chat setting
agent.launch and orchestration worker-start read the same setting as the
renderer route; pin both states for Grok on each. The setting's description no
longer names only Codex and Claude, in every catalog.
* docs(acp): generic ACP comments say what holds for every agent, not Grok
Stop ends the session for every ACP agent, as in the common pattern; the
adoption hook comment goes (adoption is not planned); a failed start's child is
retried at the next start or quit.
* test(claude, codex): type the EOF-before-exit test's streams; the supervisor no longer forwards EOF
The Claude test wrote to the child's stdout and stderr through their Readable
type, which the node typecheck rejects; it now holds its own PassThrough
streams. The comments no longer credit the reverted supervisor change.
* feat(acp): a steer's cancel asks once and never ends the agent
The runtime had one cancel: send session/cancel, wait at most 10 s for Orca's prompt to settle,
then close the connection, which ends the agent. A steer used it too, so a slow agent lost its
process just because the person added a message. requestSteerCancel() now sends session/cancel
once per prompt, cancels the agent's open requests and answers later permissions cancelled, and
never bounds or closes: the prompt's own reply ends it and the steer's prompt follows. cancel()
stays the Stop: bounded, then close. A Stop after a steer still bounds and closes. Both cancel
paths move into acp-prompt-cancel.ts over one cancel channel.
* chore(native-chat): keep the record store and recovery capsule under max-lines after the main merge
* fix(acp): a repeated steer shares the cancel in flight; say what the caller owns
Per review: a second steer before the first write lands returns that write instead of resolving
early. The steer's JSDoc says the wait for the prompt's reply is unbounded and that a prompt that
fails instead must not take the steer until the caller rebuilds the session; the Stop's says a
prompt that settles in time leaves the agent for the Stop's owner to end. The steer test now gives
the runtime a handler that would allow: the open permission's signal aborts and the late one never
reaches it.
* fix(native-chat): drop the stopDelivery the A3 merge doubled
* fix(acp): a steer's cancel asks Grok once and never ends it
A steer now uses D1's notify-only cancel. Two messages sent during a reply Grok began itself
cut that reply, as the common pattern does, and then both run; before, the queued first
message could not answer the bounded cancel and Orca ended Grok although Grok answered.
A Stop keeps the bounded cancel and its 4 s grace.
* fix(acp): a permission Grok asks with no prompt of Orca's running is declined
During a turn Grok began itself nobody asked it to act, so the request is answered
cancelled at once instead of opening a card that waits, as the common pattern does.
* fix(acp): a Grok that dies while starting is reported with its own last words
A dying process's stdout ends before its exit is seen, so the start failed as a closed
connection and Grok's stderr was lost. A start whose connection closed now waits, bounded by
the Stop grace (or a Close/Stop), for the exit before it is told.
* test: a Stop after a steer sends its own cancel; drop the import the A3 merge doubled
* test(native-chat): main's Stop-note test builds its turn context with the agent registry
* test(claude): say why the close test's fake child cast is safe
* test(native-chat): build the Stop-opened-turn test's identity and turn context the current way
The test (#25056) landed before the opaque provider handle (#24991), so main still built
the old {kind, threadId} handle; the turn context also needs this branch's agent registry.
* test(native-chat): build the Stop-opened-turn test's identity with the opaque handle
The test (#25056) landed before the opaque provider handle (#24991), so main still built
the old {kind, threadId} handle.
* test(ratchet): require src/main/provider-process now that it has landed
* test(native-chat): keep main's opaque-handle import in the Stop-opened-turn test
Main's #25706 and this branch both added the import at different lines; the merge kept both.
* test(native-chat): keep main's opaque-handle import in the Stop-opened-turn test
Main's #25706 made the same fix as this branch at a different line; the merge kept both imports.
* feat(native-chat): record a fresh provider conversation that replaced one the agent could not restore
A chat whose saved conversation the agent cannot reopen can now continue in a
fresh one: the handle chain records the new conversation as a creation that
replaces the lost one (which, why, and when), keeping every earlier link.
Rows keep a shape older builds read: the stored chain starts at the latest
replacement and carries the earlier links inside it.
* test(native-chat): build this stack's journal identities with main's opaque provider handle
Main's #24991 replaced the {kind, ...} handle with {transport, agent, nativeId}; three test
files from this stack still wrote the old shape. Same lines the downstream ACP branch uses.
* docs(acp): every reattach drops the agent's replay, not only for a chat the journal holds
* refactor(native-chat): store a replaced conversation flat; refuse it where older builds read the row
Older builds only read Claude and Codex records, so the nested stored form
protected rows no replacement can reach while adding a cap mismatch after a
downgrade. Store the chain as held, refuse a replacement in a Claude or Codex
chain until one has a stored shape older builds read, and refuse a supersession
key on a replacement that names no creation in the chain.
* refactor(native-chat): read hosts' structured agents from the app-shell services
Main grew the startup hydration hook to its line limit; the host agents sync is an app-lifetime subscription like the structured session tabs sync beside it, so it moves there.
* Use current provider handles in transition tests
* Use current provider handles in timeline fixtures
* test(native-chat): prove replacement rows survive downgrade and re-upgrade
* Require the ACP directory in the runtime import check
* test(ratchet): require src/main/acp now that this PR lands it
* feat(acp): a saved session the agent cannot reopen continues in a new one, with one warning row
When session/load (or session/resume) of a saved ACP session fails, the chat starts a new session and records it as a creation that replaces the lost one (#25747's 'replaces' link), and writes one warning row that the agent no longer remembers the earlier messages. A created session the agent reports missing is still superseded silently; a signed-out agent or a start that is over (Close, Stop, a lost agent) still fails the start.
* chore(acp): rewrap the acquire header comment
* test(acp): a start closed while the agent reopens fails without opening or announcing a new session
* Let ACP connections own their supervised agent process
* Preserve ACP cleanup evidence and isolate exit observers
* Expose ACP cleanup observations and type the permission fixture
* refactor(native-chat): the registered-agents capability lives in its own module
Main's growth put protocol-version.ts one counted line over its 300-line limit once the capability
was added; like main's other per-feature capabilities, it now has its own module, and importers read
it from there.
* refactor(acp): one connection owns the Grok process and its protocol
D3 now opens each ACP agent through createAcpAgentConnection (ACP-ALIGN #25810): one object spawns the
process on the execution host, owns its stdio and protocol, and reports its proven exit. It is built and
tracked before the handshake, so a start's abort (Close, Stop, quit) still reaches it, and a failed start
keeps that same connection for the next close to retry rather than spawning another process.
Deleted: the spawnAcpStructuredChild wrapper and its test, the raw-stream runtime assembly, the caller's
exit -> runtime.close wiring, the stdout-EOF heuristic (the connection no longer treats stdout EOF as
exit), and the 10 s steer/Stop cancel bound with requestSteerCancel. Reader control maps to
pauseReading/resumeReading; a close is connection.close after the host's existing 4 s Stop grace.
The adapter owns what the protocol no longer does: one session/cancel per running prompt however many
steers arrive (cleared with that send's settlement, retried after a failed write), and a Stop or steer
answers every open agent request the person has not already answered with the agent's own cancelled
reply. An answer already being saved when the Stop lands is sent.
Tests: blocked cancel write never holds Stop's grace, two quick steers send one cancel, a failed cancel
write is retried, a real process exiting while a child holds its stdout ends the session, and the
existing start-abort, retention, crash, connection-loss and reload-failure suites on the new rig.
* fix(acp): Grok signs in on its own machine with its API key or cached sign-in
When Grok reports that it needs authentication, Orca now names a sign-in method on the machine Grok runs
on, read from the same environment Grok was launched with: xai.api_key when XAI_API_KEY is set there and
Grok offers that method, else cached_token when Grok offers it, else none and the chat keeps the existing
not-signed-in refusal. The rule lives in Grok's launch spec; the adapter applies any agent's rule for new
and reopened sessions through the protocol client's caller-named method (authenticate, then retry once).
No new sign-in UI; interactive methods are never chosen.
* fix(acp): the adapter decides which of Grok's requests reach the person
The turn owner now admits every agent request, permission or question, from its own turn state: a
request reaches the person only while Orca's prompt runs and no steer or Stop is cutting it short (a
question may also come from a turn Grok began itself, until a Stop). Anything else gets the agent's
own cancelled reply and opens no card, so a question arriving after Stop or during a steer never
appears. A steer, like a Stop, withdraws the requests already open; an answer already being saved is
still sent. The protocol client's abort-on-cancel path is no longer used: after the connection
change its request signal aborts only when the connection closes.
* feat(native-chat): add inactive Pi RPC transport foundation
* fix(acp): a plan Grok proposes shows as a plan, with no approval card
When Grok leaves plan mode it asks the client to approve its plan (x.ai/exit_plan_mode). Orca showed a
blocking 'Approve plan / Request changes' card for it; the common pattern has no such gate. Now the
plan goes into the chat's existing Plan row (the plan-document status row Codex and ACP plan updates
already use) and the request is answered at once with 'abandoned' plus feedback telling Grok to stop and
wait for the person's feedback or a request to implement it in a later turn, so nothing is approved on
the person's behalf. Dialects gain settleRequest for requests answered without asking anyone.
* fix(orchestration): worker-start opens a Grok worker in a terminal, as before
With the structured chat setting on, worker-start decided 'structured' for Grok and then the structured
worker factory (Claude and Codex only) refused it, so the start failed; main opened a terminal Grok
worker. Worker-start now decides with no registered agents beyond Claude and Codex, so Grok gets a
terminal worker as before. agent.launch and the app's own launches still open Grok as a structured
chat. Temporary until structured workers take registered agents.
* fix(acp): a prompt answer Orca can't read ends the turn instead of hanging it
A session/prompt rejection that was not the agent's own error answer (an answer that fails Orca's
schema, or one too large to read) left the turn running: the next message became a steer with nothing
to cancel and was never sent or settled, and Stop waited its full grace. As in the common pattern, any
prompt failure now ends the turn as failed (a failed-turn row without words, since none are the
agent's) and settles the send, so the next message goes. Only a closed connection keeps the send
running, for the connection-loss path to settle.
* feat(native-chat): add bounded RPC reading control
* fix(acp): send Grok's prompt-identity extension only to agents that echo it
session/prompt carried _meta {promptId, requestId} for every ACP agent, though only Grok's dialect
echoes it (injectedPromptIdentity). Now only an agent whose dialect declares it gets the extension;
other ACP agents get a plain prompt.
* refactor(native-chat): the registered-agents capability lives in protocol-version again, as on main
This reverts
|
||
|
Brennan Benson
|
2f49377425 |
feat(native-chat): Grok as a structured chat over the Agent Client Protocol (#25225)
* Leave a stopped turn's running tools to the agent's own end
When another writer settles the open turn (a person's Stop), the assembler
now only stops that turn's text and cancels its pending prompts. Running tool
calls stay the agent's: a progress update or completion it reports after the
Stop lands as reported, and whatever is still running settles at the agent's
turn end for that turn, the next turn's open, or the session's end.
An agent's end for an earlier turn while a newer one is open no longer clears
the open turn's activity line or ends its anonymous reply. An unnamed end right
after a Stop ends the stopped turn instead of being dropped. The test rig's
restart no longer writes the dead assembler's window text, matching dispose.
* Pin that a stopped turn's running tools hold budget until the agent's end
* Type the stopped turn's tool progress update as a tool body
* List every event the assembler hands to the decision step
The type-aware lint requires an exhaustive switch with no default case.
Also retitle a Stop test to say what it asserts.
* End a running call as its turn's journal row ends
A call still running when its turn ends takes the state of that turn's
row: a row another writer settled first (a person's Stop) stands, so its
calls read interrupted whatever the provider's later end reports. The
no-ending path that settled calls from the Stop row is gone, since a Stop
now leaves running calls to the provider. Adds the two Spanish strings.
* Say why a Grok turn failed, and keep task rows in Grok's own words
A failed Grok turn ended with no reason on screen: the translator dropped
every copy of Grok's message. The failed turn now gets one status row in
Orca's existing "provider did not accept this message" words with Grok's
reason, read from whichever copy arrives first (the given-up retry, the
turn's end, the prompt's completion notice, or the prompt's error answer);
later copies only fill a reason the row still lacks.
A running background command no longer reads "Background task <id>
started": a task's summary is mapped only once it has settled. A monitor
stays a monitor when the agent reads its output: a frame that names no
kind keeps the known one, and a "[monitor" command is a monitor.
A prompt's turn is marked started, so a late frame for an ended prompt
neither reopens it nor becomes the active turn. A tool's turn is held in
one place at a time.
* Read a monitor from Grok's exact output prefix
* Word a failed Grok turn in Grok's own text, not as a refused message
A turn that started and then failed was told "The provider did not accept
this message", Orca's sentence for a message refused before its turn. The
row now reads as a Codex turn-ending error does: an error status row with the
provider's own words. With no words, the dialect names the failure ("Grok
ended this turn with an error." / "Grok usage limit reached."), else the
agent's display name does.
* Settle a stopped turn's running call as its turn row ended after a restart too
The restart sweep ended every running call by the death evidence alone, so after
a person's Stop with no proof the child died the call read failed under a turn
that read interrupted. The sweep and the live dead-generation settlement now ask
the same rule the assembler does: a call in a turn already settled ends as that
row ended; only a turn still running leaves its calls to the evidence.
* Keep the dead-generation settlement under the line cap
* Register the ACP schema verify step in the PR preflight phase test
* refactor(agent-session): one required agent registry; declarations admit what they claim
A StructuredAgentRegistry, built once from the {definition, adapter}
registrations, is now a required host dependency and the router routes with
it. The adapter interface loses its optional router-only capabilities?() and
definition?(); every reader (options at rest, thread goal, rewind, the
/compact handover) asks the registry. A live session still narrows rewind
through the adapter, and the host combines declared and narrowed in one
helper. The registry refuses a registration that declares compact, a thread
goal or rewind without the adapter method behind it.
/compact is admitted by the declared capability, so an agent that declares
compact:false gets the commandRefused fact instead of a thrown error.
Also: the cut-turn notice names the agent from the catalog, create-support
builds the account home through agentSessionAccountHome, and the turn
status text older clients read names the session's agent instead of
defaulting to Codex (Claude/Codex text unchanged).
* Read ACP permissions, session events and prompt errors through the protocol client's own types
The translator now reads a permission request with the client's lenient reader, a session update
with its session-event reader, and takes only the agent's own error answer as a failed prompt's
reason, so an Orca-side error never reads as the provider's words. Tests cover protocol values
newer than this build.
* refactor(agent-session): the router applies the declared rewind itself
The router holds the registry, so its rewindSupport answers the owner's
declared rewind narrowed by the adapter, in one place no reader can bypass.
The host-side combining helper is gone; readers ask the adapter they hold.
* test(agent-session): register the agents the merged-in tests now need
The adoption replay test builds its host over this build's registered agents
instead of an adapter definition method, and the stored-form test passes the
stored agents the record guard now requires.
* chore(agent-session): the registry is the one lookup; drop the test-only empty capability record
* fix(agent-session): keep "dismiss all" restart offers dismissed for the desktop
The desktop renderer calls the restart methods as a paired runtime client
without the registered-agents capability, so it was handed a Claude/Codex
audience and its "dismiss all" took the scoped path: no dismissal fence and
no clearing of unwritten teardown witnesses, so a late teardown write could
bring a dismissed offer back.
The audience is now derived against this host's registry: a client that can
show every agent the host registered gets none, and dismisses exactly as
before (one fence for every offer). A client that truly cannot show some
agent still gets a final dismissal for what it sees: each cleared session is
fenced on its own (bounded, superseded by a dismiss-all fence) and only the
witnesses of agents it sees are dropped.
One predicate now answers which agents a client renders, for both tab
projection and restart offers; the Claude capability rule is folded in.
* fix(agent-session): a changed agent definition never hides that agent's chats
A record was readable only if every handle used the transport its agent's
definition declares today, and its account variable was declared by some
registered agent. A later build that drives the same agent over another
protocol, or under a renamed variable, would have set every existing chat of
that agent aside: no tab, no history, no error.
Readability now asks only that the record agree with itself: its agent is
registered, every handle shares one namespace owned by that agent, and the
account variable is a well-formed name. Whether this build can drive it (the
chain's transport is the one the agent speaks, and the account variable is
that agent's own, since it becomes the child's environment) is decided in
performAttach, the one admission every start of every agent passes through,
and refused there as hostUnsupported. A record pinning another agent's
variable is refused the same way rather than passing because some agent
declares it.
* refactor(agent-session): each agent's registration says where it runs and which account it pins
createSupport, create and the model catalog's account read still chose a
location rule and an account resolver by name (Claude, Codex, else none),
so a new agent would have needed a third branch beside the registration
list that already decides storage, routing and publication.
Each runtime registration now carries supportsLocation(location) and
resolveAccountHomePath({ launchEnv, location, purpose, workspacePath }),
and Claude's and Codex's rules move into their entries unchanged: a read
still syncs no home and starts no bridge, a Codex launch still trusts the
workspace first, and WSL locations resolve as before. The list exists
before the host is built, so none of these reads installs the host, and an
agent the list does not hold is answered no without opening the journal.
* test(agent-session): drop a duplicate registry key; keep fences out of the capsule state
A spread input already carries the registry, so the explicit key was
overwritten (TS2783). The session-fence helper now returns only the fence,
not the capsule state it was given.
* fix(agent-session): a scoped dismiss-all persists no per-session fence
The per-session dismissal fence added a forever-persisted capsule field
with an arbitrary cap that served no caller: only the local desktop calls
the restart methods, and on a host whose agents it all shows it gets no
audience and takes the unscoped, fenced dismissal. A scoped dismiss-all
(a caller that cannot show some registered agent) keeps that agent's
offers, clears only its audience's unwritten witnesses, and writes no
fence; a late write from this process is serialized behind it. The field
never left this unreleased branch.
* fix(agent-session): refuse an attach whose agent is not the session's own
The start check judges the record's agent while the router starts the adapter of params.agent, and the attach wire let the two differ. Admission now refuses agent !== provider as requestMalformed, so the agent checked is the agent started. Every host-built attach already sets them equal.
* fix(agent-session): offer to start a chat only when the start would accept it
The restart offer, a failure's retryable flag and the pre-send check asked only whether the adapter runs the chat's location, while the start also refuses a record this build cannot drive; such a chat was offered Resume and its Retry failed forever. One host predicate, hostCanStartRecord (location support and agentDrivesSession), now answers all four; adapterSupportsRecord stays the reader gate, and an undrivable chat's offer is kept, not retired.
* fix(agent-session): the model-catalog probe uses a record's account only if this build drives it
A record-scoped catalog read started the agent's lister under the record's account path whatever variable the record pinned. It now uses that path only when agentDrivesSession holds, and otherwise resolves the account as for a read with no record.
* refactor(agent-session): the record store admits agent ids; comments say where transport is checked
The store only ever asked whether an agent is registered, so its admission list is now the registered ids; the definition is the one home of an agent's transport and account variable. Comments that said the record store checks transport, and one that cited a nonexistent function, now point at agentDrivesSession and the attach admission. The isPersistedAgentSessionRecord(value, agents) call shape and the test fixture the cross-version probe imports are kept.
* docs(agent-session): the record store admits the registered agents' ids
* refactor(native-chat): Grok's registration declares where it runs; ACP no longer borrows Codex's location rule
The rule a self-supervised agent child runs under (this machine, no WSL, Windows only with process
start-time proof) is its own module that Codex and the ACP adapter both use. Grok's registration
takes the full account-home resolver signature, and D3's tests build hosts with the agent registry.
* fix(native-chat): Grok follows the ACP runtime's request contract and the managed process's close
A request the agent or a Stop cancels is answered with the agent's own cancelled reply by the code that
owns it (the runtime no longer answers a silent handler), so a Stop needs no separate decline pass. A
permission answer still being saved when the agent stopped waiting is reported unconfirmed, since the
protocol already answered it cancelled. Cancelling the agent's own turn is the plain cancel. Request
rows are matched under their generation-scoped ids. A refusal's reason comes from the dialect's wording
path. The child drops its own stderr tail and close policy for the managed process's, and a close
whose process tree was not proven gone is reported as the adapter contract asks.
* fix(native-chat): a Grok chat Orca already holds resumes without writing what Grok replays
A chat with a saved Grok session reattaches with session/resume where the agent offers it, else
session/load. Either way the call runs inside the translator's load window, so what Grok sends while
it reattaches (its saved exchange, a task the dead process left running, ended by the restart) opens
no turn and writes no row; only context usage reads on. A reply an Orca or Grok crash cut short is no
longer completed from Grok's saved history: it reads like a Claude or Codex chat's, with the existing
notice. The attach window also closes after a failed attach, and a created session that session/resume
reports missing is replaced like one session/load reports missing.
The replay reconciliation is removed: the lane no longer reads the journal, and D3's replayed-input
grammar test and completed-turn check in the assembler go with it.
* refactor(native-chat): a failed Grok reattach needs no window close of its own; its lane is replaced
* test(native-chat): D3's merged tests use the shipped declarations and the launch options main requires
* fix(native-chat): typecheck fallout of the base merges; any agent's empty chat is reusable
Main's idle-empty-chat lookup and launch join now take any registered agent, as the rest of the
launch path does. The refusal check moved into the prompt turns and the prompt-block conversion beside
the turns that send it, keeping both files in their line limit.
* fix(native-chat): a Grok Stop ends the process once Grok settles its turn; the next send resumes
Grok's session/cancel ends only the running turn: work it already moved to the background keeps
running and can begin a turn of its own after the person pressed Stop. Stop is now a session
boundary, as it is for Claude: the cancel answers open requests and lets Grok end the turn, the host
waits a bounded grace for that, then ends the process; the next send relaunches and resumes.
The adapter's own bounded close of a turn Grok began is gone. Its named-turn check stays: the host
ends the session unless the provider declines a Stop naming a turn that has since ended.
* test(native-chat): a Grok Stop ends the process only after Grok answered the cancel
* fix(native-chat): Steer on a Grok card cancels the running prompt, then sends it
A send that reached Grok while a prompt ran was held in the adapter until that turn ended: Steer
on a queued card took the card out of the host's editable queue and meant 'send after this turn'.
It now cancels the running prompt (session/cancel; the session stays) and sends as the next prompt
once Grok answers the cancel, as the common pattern does; a steer behind another cancels it in
turn, so the last one runs. The adapter holds a send only while that cancel lands, so its general
held-send queue and its holdsDispatch report are gone (every send it holds has its turn open in
the journal). An older client's mid-turn send takes the same path. capabilities.steering is
unchanged and still unread.
* refactor(native-chat): a close or Stop cancels a start through the acquire's own abort signal
The host owns the acquire it runs, so it now owns its cancellation: each attach's acquire gets an
AbortSignal, aborted from outside the session's queue by a close and by a Stop admitted now (the
same admission rule as before). The optional abandonStart adapter hook, the router's fan-out to
every adapter and the ACP adapter's session-keyed start map are gone; the ACP adapter keeps an
unkeyed set of starts only so quit can prove their children gone, and keeps a failed start's
unproven child until its exit is proven.
The hook also let a later close ask that child again. The host now does that from state it holds:
a close of a chat with no live child whose record still names an owner process with no death
evidence asks the adapter to release it. The answer is not recorded as proof (the lease probe
does that), so an owner pid an earlier Orca left is never killed or marked gone. Claude and Codex
ignore the signal and hold no such child; their release is a no-op (tested).
* fix(native-chat): a Grok crash that closes stdout before its exit still ends with Grok's last words
On macOS and Linux the agent's stdout ends before its exit is observed, with or without the
supervisor's EOF forwarding, so the connection's loss closed the journal first and its error text
became the session's ended reason, dropping Grok's stderr. The reason is now read at the proven
exit: the agent's last words when it left any, else why the connection closed. The failure already
carried them. Comments that assumed the exit comes first, that early frames past the cap refuse the
start, and that dispatch re-checks image support are corrected.
* fix(native-chat): nothing Grok sends while a held chat reattaches is written, marked as replay or not
The reattach window relied on the dialect's replay verdict, and Grok's frames read as live unless
they carry isReplay, so an unmarked chat frame during session/resume opened a turn that never
ended. D3 now marks every frame inside the window as replay before the translator reads it, so the
translator keeps only context usage whatever the agent marked; options and commands are still
adopted. The translator's load semantics are unchanged.
* test(native-chat): a Stop after a resume finds no turn an unmarked old reply opened
* test(native-chat): a resumed Grok chat keeps its last context reading; the resume refreshes only the window
* test(native-chat): a Grok background task a Stop ended reads as stopped reporting
* refactor(native-chat): quit's stop of each start answers through one promise kind
* fix(native-chat): quit aborts every start the host has in flight before draining attaches
A Grok that never answered its handshake held quit until the start's own 60 s bound, past the
20 s quit deadline. The host's teardown now aborts each in-flight acquire (and any the drain
still begins), so the adapter's own quit controller and its map of starts are gone: a start
has one canceller, the host's signal.
* fix(native-chat): a Grok start's abort stops reaching its child once the start has returned
The listener stayed on the host's signal until the attach finished committing, so a Close in that
window killed the now-live child behind the host's back and it read as Grok crashing. The start
now detaches it when it ends; a later Close goes through the session's own stop.
* fix(native-chat): a close or Stop during any attach phase stops the start before it launches
The attach began its abort controller only after reconciling leases, resolving recovery and
probing the previous owner, so a close or admitted Stop in those phases reached nothing and Grok
launched anyway. The controller now begins first, and the acquisition checks it before asking the
adapter to start.
* test(native-chat): a close during the attach's owner probe asks no adapter to start
Also renames the close test after the hook it no longer exercises.
* test(native-chat): a close's re-ask closes a Claude or Codex child a failed cleanup left
The re-ask is not a no-op for them: when the adapter still holds the child its cleanup could not
prove gone, the close stops it again as a requested close, and Claude persists the handle of the
conversation it ran so the next send resumes it. Corrects the tests' and comment's wording; the
close awaits the re-ask, bounded by each adapter's kill ladder.
* fix(native-chat): Steer during a turn Grok began itself cancels it and sends once it ends
A send while Grok ran a turn of its own (a background task waking it) went straight to Grok, which
queued it behind that turn where Orca could no longer withdraw it, while Stop treated the same turn
as the running reply. The send now waits as a steer, the turn is cancelled once, and the message
goes when the turn ends; a Stop withdraws it and an exit rejects it as never sent.
* test(native-chat): a steer whose cancel Grok never answers ends Grok and is rejected as never sent
Pins the bounded steer cancel kept from the runtime: past the bound the connection closes, the
running reply reads unverifiable, Grok's end reads as its exit, and the waiting steer is rejected
as never sent.
* fix(native-chat): a Grok crash stays a crash when a stop lands before its exit is proven
After the connection broke and the close could not prove Grok's exit, any later stop Orca asked
for (the next start, a Stop, a Close) marked the child as closed by Orca, so the crash read as a
requested close and Grok's last words were dropped; a send meanwhile was recorded unconfirmed.
The connection loss now decides the cause, and a send on that session is rejected as never sent.
* test(native-chat): fixtures this PR's registered Grok and desktop capability made stale
CI's unit shards failed on tests outside the PR's own lists. Each encodes something this PR changes
on purpose: Grok is now a registered agent (the seam test's unregistered agent is now Cursor); the
desktop now advertises registered agents (the restart-offer tests' older client drops that
capability explicitly); the attach context carries the start's abort controllers (the forget-status
double gains them); and the ACP real-host test rig sends to the host directly (listed beside the
other real-host rig in the send ratchet).
* fix(native-chat): a start quit stops is not the queued message's start failure
With quit now aborting a start it would have waited for, the delivery step recorded the aborted
start as the message's failure ("couldn't restart"). After quit has stopped delivery, the step
leaves the message to quit, which settles it as a close does ("The chat closed before this message
was sent."). The test that pinned quit waiting for that start and stopping its child now pins that
nothing is launched behind quit.
* fix(native-chat): a message sent after a Stop or close aborted a start gets its own start
A start the host aborts (an admitted Stop, a close, or quit) returned its refusal to the delivery
loop, which then rejected whatever was queued at that moment with "couldn't restart", including a
message the user sent after the Stop. The attach now reports that the host aborted it, and the loop
re-derives from the journal instead: what the Stop or close withdrew is already settled, a message
accepted since gets a start of its own, and quit's next step stops the loop. This replaces the
quit-only carve-out with the same rule for every abort and every agent.
* test(native-chat): the message sent after an aborted start is answered, so no settlement outlives the test
* fix(native-chat): a Grok model pick Grok never answers no longer holds Stop or Close
The pick runs on the session's queue. It now registers in the host's out-of-queue
abort registry beside a start, so a close, an admitted Stop or quit abandons it, and
the ACP adapter bounds it at 30 s like Claude and Codex. A late answer is still adopted.
* fix(agent-launch): a phone's launch opens a terminal for an agent whose chat it cannot show
agent.launch now reads the caller's capabilities by the rule tabs and restart offers
use (clientRendersStructuredAgent). A phone without registered-agents.v1 gets Grok as
a terminal again, as on main; the host's own callers and desktop clients are unchanged.
* fix(acp): strip every agent hook variable from the ACP child, from the shared list
ACP_CHILD_ENV_TO_DELETE was a second copy of the hook runtime keys that missed
ORCA_AGENT_HOOK_TRANSPORT; it now spreads AGENT_HOOK_RUNTIME_ENV_KEYS beside the pane
identity keys.
* refactor(native-chat): the mutation context carries the provider-wait registry itself
Keeps the host file within its line limit; one field instead of two closures over it.
* fix(agent-launch): agent.launch.v2 still vouches for Claude and Codex chats
The caller rule from the previous commit also turned Claude and Codex into terminals
for a client advertising only agent.launch.v2, whose contract says it opens a chat
(mobile retry-authority tests). Only an agent beyond those two now needs the client to
read it (clientRendersStructuredAgent); the test fixtures go back to what they were.
* refactor(native-chat): drop saved-history adoption from the timeline assembler
The common pattern discards the history a provider replays while loading a
session, so the assembler has no use for an input.history event.
* refactor(acp): drop session/load history adoption from the translator
The common pattern discards the history an agent replays during session/load,
keeping only what it says about the context window. Remove the adoption path
(acp-history-adoption.ts, the adopt option, and the historical background-task
liveness rewrite it fed) so load replay is always dropped except usage.
* refactor(native-chat): a pending input is only Orca's send now
Review follow-up to the adoption removal: drop the comment naming the
provider's saved message, and make requestedAt required since every pending
input comes from input.accepted.
* test(acp): keep the task-result status table on live frames
Review follow-up to the adoption removal: the result-status mapping was only
tested through adopted history, so run the same table on live frames, and
cover an unmarked task notice during a load being dropped.
* test(acp): a frame helper for a shell command Grok is running
* fix(acp): a Grok crash settles through the host's provider-exit batch, scoped to the turn it ended
A Grok crash ended the journal unverifiable before the adapter reported the exit, so the host's
provider-exit settlement found no running turn and wrote nothing: the adapter's failure (with
Grok's last words) never reached the journal, and a later stale-session pass wrote a bare,
thread-scoped cut-short row, so the partial reply was not folded as Claude's and Codex's are.
At a proven exit the ACP lane now ends its running turn interrupted at the exit instant, as the
host's exit contract expects of a child's own translator (Codex's does the same). When Grok's
stdout closed first (every POSIX crash), the turn is unverifiable only until the exit is proven:
the host's provider-exit settlement now takes the exit as proof naming the child's fence and
revises what that child left unverifiable in the same batch, with the turn-scoped row and the
adapter's failure. Claude and Codex write no unverifiable turn of a live child except a command
whose hand-off is in doubt; that turn is now revised at the exit instead of at the next open.
* test(acp): a crash seen first leaves the host no Grok turn to revise
* refactor(native-chat): what a gone generation left unfinished gets its own module
The settlement file passed 300 lines with the exit-proof revision. The unfinished-work reads
(capture, interrupted-by-the-exit, in-progress) are their own concept and move out unchanged,
apart from the exit proof they now take.
* refactor(native-chat): a watched exit revises what its child left unverifiable without reading Stop marks
An exit's own instant is the turn's end, so the revision needs only each row's fence: the
settlement's journal type gains itemFence alone, and the host test fakes say so.
* test(native-chat): drop the duplicate itemFence on the fake that already had one
* test(claude, codex): an exit whose stdout ended first still reports as it always did
The provider supervisor now ends Orca's stdout when the agent's ends, so on every crash EOF
arrives before the exit is seen. Claude's and Codex's connections report nothing at EOF and
report the exit, with its usual reason, once it is seen.
* fix(acp): reopen a chat with session/load, as the common pattern does
An agent that offers both now reloads its session instead of resuming it; the
reattach window still discards what it replays except context usage.
* fix(acp): drop the 60 s handshake bound; an abort fails the start's waits at once
Neither common design bounds an ACP handshake: Close, Stop and quit end a start
that never answers. The abort now also closes the connection, as a kill there
does, so the start settles even before the child's exit is proven. The
host-stopped start refusal only this bound produced goes with it; the idle
sweep keeps its words.
* fix(acp): a Stop naming an ended turn follows Claude's rule
It still stops nothing while another turn is live, but in the gap before a
follow-up's turn opens, which no client can name, it now stops what is in
flight and the session ends, as a Claude Stop does.
* fix(native-chat): a close no longer re-asks a failed start's unproven child
Neither common design retries that stop at Close, and Orca's Claude contract
re-asks only at the next start and at quit. The ACP adapter keeps the child
until its exit is proven and asks it again there, as Claude does.
* fix(acp): a message sent during a turn the agent began itself goes at once
Both common designs send it straight to the agent with no cancel; only Orca's
own running prompt is steered (cancelled, then re-prompted).
* test(native-chat): dismiss-all through a remote client's audience keeps a newer Orca's offer
Uses an audience production sends (one that cannot show every agent), per review.
* fix(acp): launch Grok as `grok agent stdio`, without the update and leader flags
The common pattern passes neither --no-auto-update, --no-leader nor
GROK_DISABLE_AUTOUPDATER; full access still adds --always-approve.
* fix(acp): an agent that ends its stdout, or answers unreadably, is not a lost connection
As in the common pattern, only a broken stdin (or Orca's own close) ends the
agent; one that closed its output but can still be written to stays until a
Stop, a close or its exit. The provider supervisor goes back to its base
content, so Claude and Codex no longer get the forwarded stdout end either.
* fix(native-chat): a person's close joining a failed one still binds the turn its child end cuts
On main every close of the chat writes its own Stop and settle. Here a later close joins the
first and writes no row, and the first's settle closed when its kill failed, so a turn that opened
in between and was cut by the next close read as failed. A person's close joining a person's close
whose Stop opened a settle now reopens that settle until its attempt is done.
Tests: a close whose kill failed still closes its settle; a turn opened between a failed close and
the next reads as the person's cancellation (each fails without its half of the fix).
* test(native-chat): Grok opens as a chat only behind the structured-chat setting
agent.launch and orchestration worker-start read the same setting as the
renderer route; pin both states for Grok on each. The setting's description no
longer names only Codex and Claude, in every catalog.
* docs(acp): generic ACP comments say what holds for every agent, not Grok
Stop ends the session for every ACP agent, as in the common pattern; the
adoption hook comment goes (adoption is not planned); a failed start's child is
retried at the next start or quit.
* test(claude, codex): type the EOF-before-exit test's streams; the supervisor no longer forwards EOF
The Claude test wrote to the child's stdout and stderr through their Readable
type, which the node typecheck rejects; it now holds its own PassThrough
streams. The comments no longer credit the reverted supervisor change.
* feat(acp): a steer's cancel asks once and never ends the agent
The runtime had one cancel: send session/cancel, wait at most 10 s for Orca's prompt to settle,
then close the connection, which ends the agent. A steer used it too, so a slow agent lost its
process just because the person added a message. requestSteerCancel() now sends session/cancel
once per prompt, cancels the agent's open requests and answers later permissions cancelled, and
never bounds or closes: the prompt's own reply ends it and the steer's prompt follows. cancel()
stays the Stop: bounded, then close. A Stop after a steer still bounds and closes. Both cancel
paths move into acp-prompt-cancel.ts over one cancel channel.
* chore(native-chat): keep the record store and recovery capsule under max-lines after the main merge
* fix(acp): a repeated steer shares the cancel in flight; say what the caller owns
Per review: a second steer before the first write lands returns that write instead of resolving
early. The steer's JSDoc says the wait for the prompt's reply is unbounded and that a prompt that
fails instead must not take the steer until the caller rebuilds the session; the Stop's says a
prompt that settles in time leaves the agent for the Stop's owner to end. The steer test now gives
the runtime a handler that would allow: the open permission's signal aborts and the late one never
reaches it.
* fix(native-chat): drop the stopDelivery the A3 merge doubled
* fix(acp): a steer's cancel asks Grok once and never ends it
A steer now uses D1's notify-only cancel. Two messages sent during a reply Grok began itself
cut that reply, as the common pattern does, and then both run; before, the queued first
message could not answer the bounded cancel and Orca ended Grok although Grok answered.
A Stop keeps the bounded cancel and its 4 s grace.
* fix(acp): a permission Grok asks with no prompt of Orca's running is declined
During a turn Grok began itself nobody asked it to act, so the request is answered
cancelled at once instead of opening a card that waits, as the common pattern does.
* fix(acp): a Grok that dies while starting is reported with its own last words
A dying process's stdout ends before its exit is seen, so the start failed as a closed
connection and Grok's stderr was lost. A start whose connection closed now waits, bounded by
the Stop grace (or a Close/Stop), for the exit before it is told.
* test: a Stop after a steer sends its own cancel; drop the import the A3 merge doubled
* test(native-chat): main's Stop-note test builds its turn context with the agent registry
* test(claude): say why the close test's fake child cast is safe
* test(native-chat): build the Stop-opened-turn test's identity and turn context the current way
The test (#25056) landed before the opaque provider handle (#24991), so main still built
the old {kind, threadId} handle; the turn context also needs this branch's agent registry.
* test(native-chat): build the Stop-opened-turn test's identity with the opaque handle
The test (#25056) landed before the opaque provider handle (#24991), so main still built
the old {kind, threadId} handle.
* test(ratchet): require src/main/provider-process now that it has landed
* test(native-chat): keep main's opaque-handle import in the Stop-opened-turn test
Main's #25706 and this branch both added the import at different lines; the merge kept both.
* test(native-chat): keep main's opaque-handle import in the Stop-opened-turn test
Main's #25706 made the same fix as this branch at a different line; the merge kept both imports.
* feat(native-chat): record a fresh provider conversation that replaced one the agent could not restore
A chat whose saved conversation the agent cannot reopen can now continue in a
fresh one: the handle chain records the new conversation as a creation that
replaces the lost one (which, why, and when), keeping every earlier link.
Rows keep a shape older builds read: the stored chain starts at the latest
replacement and carries the earlier links inside it.
* test(native-chat): build this stack's journal identities with main's opaque provider handle
Main's #24991 replaced the {kind, ...} handle with {transport, agent, nativeId}; three test
files from this stack still wrote the old shape. Same lines the downstream ACP branch uses.
* docs(acp): every reattach drops the agent's replay, not only for a chat the journal holds
* refactor(native-chat): store a replaced conversation flat; refuse it where older builds read the row
Older builds only read Claude and Codex records, so the nested stored form
protected rows no replacement can reach while adding a cap mismatch after a
downgrade. Store the chain as held, refuse a replacement in a Claude or Codex
chain until one has a stored shape older builds read, and refuse a supersession
key on a replacement that names no creation in the chain.
* refactor(native-chat): read hosts' structured agents from the app-shell services
Main grew the startup hydration hook to its line limit; the host agents sync is an app-lifetime subscription like the structured session tabs sync beside it, so it moves there.
* Use current provider handles in transition tests
* Use current provider handles in timeline fixtures
* test(native-chat): prove replacement rows survive downgrade and re-upgrade
* Require the ACP directory in the runtime import check
* test(ratchet): require src/main/acp now that this PR lands it
* feat(acp): a saved session the agent cannot reopen continues in a new one, with one warning row
When session/load (or session/resume) of a saved ACP session fails, the chat starts a new session and records it as a creation that replaces the lost one (#25747's 'replaces' link), and writes one warning row that the agent no longer remembers the earlier messages. A created session the agent reports missing is still superseded silently; a signed-out agent or a start that is over (Close, Stop, a lost agent) still fails the start.
* chore(acp): rewrap the acquire header comment
* test(acp): a start closed while the agent reopens fails without opening or announcing a new session
* Let ACP connections own their supervised agent process
* Preserve ACP cleanup evidence and isolate exit observers
* Expose ACP cleanup observations and type the permission fixture
* refactor(native-chat): the registered-agents capability lives in its own module
Main's growth put protocol-version.ts one counted line over its 300-line limit once the capability
was added; like main's other per-feature capabilities, it now has its own module, and importers read
it from there.
* refactor(acp): one connection owns the Grok process and its protocol
D3 now opens each ACP agent through createAcpAgentConnection (ACP-ALIGN #25810): one object spawns the
process on the execution host, owns its stdio and protocol, and reports its proven exit. It is built and
tracked before the handshake, so a start's abort (Close, Stop, quit) still reaches it, and a failed start
keeps that same connection for the next close to retry rather than spawning another process.
Deleted: the spawnAcpStructuredChild wrapper and its test, the raw-stream runtime assembly, the caller's
exit -> runtime.close wiring, the stdout-EOF heuristic (the connection no longer treats stdout EOF as
exit), and the 10 s steer/Stop cancel bound with requestSteerCancel. Reader control maps to
pauseReading/resumeReading; a close is connection.close after the host's existing 4 s Stop grace.
The adapter owns what the protocol no longer does: one session/cancel per running prompt however many
steers arrive (cleared with that send's settlement, retried after a failed write), and a Stop or steer
answers every open agent request the person has not already answered with the agent's own cancelled
reply. An answer already being saved when the Stop lands is sent.
Tests: blocked cancel write never holds Stop's grace, two quick steers send one cancel, a failed cancel
write is retried, a real process exiting while a child holds its stdout ends the session, and the
existing start-abort, retention, crash, connection-loss and reload-failure suites on the new rig.
* fix(acp): Grok signs in on its own machine with its API key or cached sign-in
When Grok reports that it needs authentication, Orca now names a sign-in method on the machine Grok runs
on, read from the same environment Grok was launched with: xai.api_key when XAI_API_KEY is set there and
Grok offers that method, else cached_token when Grok offers it, else none and the chat keeps the existing
not-signed-in refusal. The rule lives in Grok's launch spec; the adapter applies any agent's rule for new
and reopened sessions through the protocol client's caller-named method (authenticate, then retry once).
No new sign-in UI; interactive methods are never chosen.
* fix(acp): the adapter decides which of Grok's requests reach the person
The turn owner now admits every agent request, permission or question, from its own turn state: a
request reaches the person only while Orca's prompt runs and no steer or Stop is cutting it short (a
question may also come from a turn Grok began itself, until a Stop). Anything else gets the agent's
own cancelled reply and opens no card, so a question arriving after Stop or during a steer never
appears. A steer, like a Stop, withdraws the requests already open; an answer already being saved is
still sent. The protocol client's abort-on-cancel path is no longer used: after the connection
change its request signal aborts only when the connection closes.
* fix(acp): a plan Grok proposes shows as a plan, with no approval card
When Grok leaves plan mode it asks the client to approve its plan (x.ai/exit_plan_mode). Orca showed a
blocking 'Approve plan / Request changes' card for it; the common pattern has no such gate. Now the
plan goes into the chat's existing Plan row (the plan-document status row Codex and ACP plan updates
already use) and the request is answered at once with 'abandoned' plus feedback telling Grok to stop and
wait for the person's feedback or a request to implement it in a later turn, so nothing is approved on
the person's behalf. Dialects gain settleRequest for requests answered without asking anyone.
* fix(orchestration): worker-start opens a Grok worker in a terminal, as before
With the structured chat setting on, worker-start decided 'structured' for Grok and then the structured
worker factory (Claude and Codex only) refused it, so the start failed; main opened a terminal Grok
worker. Worker-start now decides with no registered agents beyond Claude and Codex, so Grok gets a
terminal worker as before. agent.launch and the app's own launches still open Grok as a structured
chat. Temporary until structured workers take registered agents.
* fix(acp): a prompt answer Orca can't read ends the turn instead of hanging it
A session/prompt rejection that was not the agent's own error answer (an answer that fails Orca's
schema, or one too large to read) left the turn running: the next message became a steer with nothing
to cancel and was never sent or settled, and Stop waited its full grace. As in the common pattern, any
prompt failure now ends the turn as failed (a failed-turn row without words, since none are the
agent's) and settles the send, so the next message goes. Only a closed connection keeps the send
running, for the connection-loss path to settle.
* fix(acp): send Grok's prompt-identity extension only to agents that echo it
session/prompt carried _meta {promptId, requestId} for every ACP agent, though only Grok's dialect
echoes it (injectedPromptIdentity). Now only an agent whose dialect declares it gets the extension;
other ACP agents get a plain prompt.
* refactor(native-chat): the registered-agents capability lives in protocol-version again, as on main
This reverts
|
||
|
Brennan Benson
|
00e3027762 |
feat(agent-launch): show an agent's tab at once, where the caller asked (#25430)
* feat(agent-launch): host-assigned caller identity and a launch record written when the surface exists Step 1 of the agent-launch unification, on main. - The dispatcher stamps every request's caller from what its connection proved (runtime socket: the local CLI; the desktop's IPC: the desktop; a paired socket: its device). Params never set it. - The launch record is written twice: once when the tab exists (what creation settled: on the launch command, a draft, or a submit still `unconfirmed`), and again once the prompt's fate is known. A restart in between finds the running agent instead of answering "unknown". - A replay re-derives its terminal handle from the pane key in the running host, and shows `unconfirmed` only to callers that advertise agent.launch.prompt-unconfirmed.v1. - The record store opens in its own slot, without building the chat host; the chat host is built on that same store. Rebuilt from this PR's own commits ( |
||
|
Brennan Benson
|
e1e9d1c14a |
feat(agent-launch): host-assigned caller identity and a launch record written when the surface exists (#24934)
Step 1 of the agent-launch unification, on main. - The dispatcher stamps every request's caller from what its connection proved (runtime socket: the local CLI; the desktop's IPC: the desktop; a paired socket: its device). Params never set it. - The launch record is written twice: once when the tab exists (what creation settled: on the launch command, a draft, or a submit still `unconfirmed`), and again once the prompt's fate is known. A restart in between finds the running agent instead of answering "unknown". - A replay re-derives its terminal handle from the pane key in the running host, and shows `unconfirmed` only to callers that advertise agent.launch.prompt-unconfirmed.v1. - The record store opens in its own slot, without building the chat host; the chat host is built on that same store. Rebuilt from this PR's own commits ( |
||
|
Brennan Benson
|
0a26f3ec90 |
fix(native-chat): honor saved agent Command and Arguments (#25721)
* fix(native-chat): honor saved agent launch commands and arguments * fix(test): retain provider import after syncing main * fix(native-chat): preserve launch arguments and explain invalid settings * refactor(native-chat): require the current launch arguments source * fix(native-chat): require saved arguments when installing the runtime * fix(native-chat): show saved argument refusals beside launch retry * fix: ship saved-argument error messages in every locale * test: align model discovery with executable overrides * chore: drop the local pnpm install lockfile block |
||
|
Brennan Benson
|
defd85a9eb |
feat(floating-workspace): render the floating panel through the shared workspace surface (#22302)
* docs(floating-workspace): plan structured native chat in the floating panel
* feat(floating-workspace): resolve the floating workspace so a structured session can run there
* docs(floating-workspace): plan the unification onto the shared workspace surface
* chore(floating-workspace): keep the unification plan out of the branch
* test(floating-workspace): pin the structured route on capability, not workspace kind
The host half removed the 'floating-workspace' blocker; this pin still asserted it and went red
on the merge. Floating now routes like any workspace, so with capabilities un-negotiated the
resolver answers 'runtime-capability-unknown'.
* feat(floating-workspace): resolve the floating workspace's renderer row without a worktree record
The renderer had no Worktree row for the floating id: markdown creation, tab-bar path entry,
native-chat file links, attachments and model discovery all resolve a workspace through
getKnownWorktreeById and silently fail for it. The host already answers with a synthetic row
minted from the resolved floating directory; this gives the renderer the same single seam.
- The repo slice stores the host-resolved floating directory (floatingWorkspacePath); the panel's
existing cwd resolution now writes it there instead of panel-local state, so there is one copy.
- findKnownWorktreeById mints the shared floatingWorkspaceToWorktree row from that path, local
host only, cached while the directory is unchanged — mirroring the folder-workspace branch it
sits beside.
- No consumer changes needed: every row-dependent path already goes through this resolver.
* feat(tabs): scope editor and browser activation to the owning workspace
Shared editor/browser activation always wrote global selection (activeFileId,
activeBrowserTabId, activeTabType). The floating panel is an overlay above the active
workspace, and its active tab must never become global selection — the invariant
documented in agent-auto-ack-targets.ts. Group-scoped activation is a prerequisite for
rendering the floating workspace through the shared tab-group surface.
- setActiveFile and setActiveBrowserTab take an optional target workspace, mirroring
setActiveTabType: global selection moves only when the target IS the globally active
workspace; per-workspace state always updates. Omitted keeps legacy behavior.
- The shared tab-group activation/creation commands pass their workspace, which is a
no-op for the main surface (hidden surfaces are inert, so commands only fire on the
active workspace) and confines floating activation to floating state.
- Terminal setActiveTab was already owner-scoped; now pinned by test.
* refactor(tab-group): separate the workspace drag scope and group tree from main-window chrome
Preparation for hosting the floating workspace on the shared surface, whose shell owns its own
titlebar tab strip and cannot mount the main chrome:
- WorkspaceTabDragLayer: the one dnd-kit scope for a workspace's strips and tree, extracted from
TabGroupSplitLayout as a render-prop layer so a host can put shell-owned chrome (a titlebar
strip) inside the same drag scope as the tree.
- TabGroupSplitNodeTree: the chrome-free recursive group tree, with an explicit tabStrip
composition point on TabGroupPanel ('attached' keeps today's per-group strip; 'external' hosts
render no per-group strip at all — a hidden one would register duplicate drag sortables).
- WorkspacePaneOverlayLayers: the retained pane hosts extracted from WorktreeSplitSurface so any
tree host mounts the same pane lifecycle stack.
- TerminalPaneOverlayLayer gains ownsNativeChatToggleShortcut, following the existing rule that
only one live workspace registers the chord.
TabGroupSplitLayout and WorktreeSplitSurface recompose these parts; the main surface's DOM is
unchanged.
* feat(floating-workspace): decide the panel's emptiness and layout atomically
An empty floating workspace has no tab-group layout at all (the slice starts empty and
hydration skips tab-less workspaces), while the shared group tree requires one. This model
resolves the two together: the panel shows its empty state exactly until the first tab exists,
and mounts the tree only with a layout plus a live focused group (stored focus, else the group
with an active tab, else the first group — matching the shell chrome's focused-group policy).
* feat(floating-workspace): render the floating panel through the shared workspace surface
The panel previously projected exactly one tab group and hand-rendered its panes, so any tab
that left that group — 'Move Tab to Split' creates a second group — stayed alive in state but
vanished from the panel. The body now mounts the shared group tree and retained pane overlay
stack for the floating workspace id, so every group renders, splits work, and structured chat
finally has a floating surface.
- One drag scope: WorkspaceTabDragLayer wraps the titlebar strip AND the body tree, so exactly
one DndContext owns floating tab drag; strip drops reorder, body-edge drops split. The
floating-only FloatingWorkspaceTabDragContext is deleted.
- The titlebar keeps the panel's single TabBar, now fed by the shared per-group workspace model
of the focused group (tabStrip='external' keeps per-group strips out of the shell), with
workspace-scoped activation commands so global selection never moves.
- Panes render through WorkspacePaneOverlayLayers: terminals gate on the resolved cwd and a
settled viewport exactly as before; browser guests ride the shared retention layer; agent
session tabs appear in the strip and render through the structured overlay.
- The empty state renders exactly until the first resolvable tab exists (stale unified entries
still show it), per the atomic layout/emptiness model.
- Deleted: the single-group projection, the hand-written pane maps, FloatingBrowserSlot, and
the floating pane-handle registry.
Known behavior deltas, both narrowing floating onto main-workspace semantics: Cmd/Ctrl+W with
panel (not terminal) focus closes the active tab rather than one pane of a multi-pane tab, and
a pane-owned close of the last tab no longer arms the panel focus reclaim (strip and shortcut
closes still do).
* test: re-register moved pane-ownership sites with the extraction
The retained overlay stack moved into WorkspacePaneOverlayLayers and the floating chrome model
became a guest-paint consumer; the structured-pane layering pin and the browser-guest retention
census both name files, so they follow the seam.
* fix(floating-workspace): mount the shared group tree in a flex frame so panes own their height
The tree's nodes size themselves as flex items, and the floating shell mounted
them in a block container, so every group body — and every pane anchored to
one via anchor-size() — measured 0px tall. Wrap the tree and overlay layers in
the same 'absolute inset-0 flex' frame the worktree split surface uses, record
the host contract on the tree, pin the frame in the panel's element-tree test,
and add a real-renderer e2e spec that measures group-body and pane boxes for
the single-group and split layouts (jsdom computes no layout, which is how the
0px regression passed 50k tests).
* fix(runtime): import the floating worktree minting the structured create path calls
floatingWorkspaceToResolvedWorktree called floatingWorkspaceToWorktree without
importing it. The module is @ts-nocheck (mechanically split runtime), so tsc
never saw the missing import and it only failed at call time — the structured
createSupport probe for the floating workspace threw a ReferenceError. Add the
import and a test that runs the real method, which fails on the missing import
where the typechecker cannot.
* fix(browser-pane): resolve floating and folder workspace paths in notebook navigation
navigateBrowserPageToUrl looked the workspace up in allWorktrees(), which has
no row for folder or floating workspaces, so their notebook navigations ran
with an undefined workspace path. Use getKnownWorktreeById, the same resolver
every other consumer was moved to.
* fix(tabs): keep a workspace's tab selection out of the main window's global selection
* refactor(floating-workspace): launch and close panes like every other workspace
* refactor(tab-group): let the surface that mounts tab groups describe its own chrome
* refactor(floating-workspace): render the floating panel through the shared workspace surface
* fix(floating-workspace): resolve the floating directory once on the host, and keep the controls' inset
* test(floating-workspace): pin the unified floating panel against the shared surface
* fix(floating-workspace): read the floating directory's paths through the app environment
The runtime now resolves the floating directory itself, and the headless
runtime has no Electron app, so the Electron path lookup threw there.
* fix(floating-workspace): keep the main window's frame and shortcuts out of the floating panel
The shared layout drew the main window's 4px titlebar band and sidebar seam
inside the floating panel, and the chat-view toggle fired in every workspace
on screen, so one key press flipped the main window's tab and the panel's.
* fix(floating-workspace): close floating tabs without taking over the main window
An unsaved floating note's save prompt made the floating panel the main
window's active worktree, and never appeared on a fresh launch because the
workbench that owns the prompt was not mounted. The prompt now shows a
floating note in the panel, and the workbench mounts while the panel holds
tabs.
Every close path now reads the same captured emptied-workspace reaction, so a
floating pane closing itself, its shell exiting, a group close, and an unsaved
note close all keep the panel's keyboard ownership, as a tab close already did.
* test(tab-group): pin that a dirty editor close carries its emptied reaction
* refactor(tabs): one rule for when a selection moves the main window's selection
Opening a file kept a floating-only exception while every other writer checked
whether the tab's workspace is the active one. Every writer now asks the same
question, so opening a file in a background worktree no longer points the main
window's editor selection into that worktree.
* refactor(floating-workspace): drop what the shared surface left unused
The terminal pane's imperative close handle lost its last consumer when the
floating panel stopped keeping its own pane registry, and the floating and
folder workspaces built the same lineage-free resolved worktree twice.
* test(floating-workspace): type test fixtures without casts and drop the removed pane ref
* fix(floating-workspace): keep panel focus when a close lands after the panel emptied
Saving the last floating note closes it after an awaited write, so the panel
could render empty before the close armed its keyboard reclaim, leaving the
intent armed with nothing left to consume it. The panel now re-checks whenever
the intent arms as well as when its tab count changes.
A pinned tab's confirmation took focus before the close read panel ownership;
the dispatcher now captures it before any prompt.
* refactor(floating-workspace): finish what the shared surface left behind
- The floating panel no longer accepts AI vault session drops: a session
cannot resume into it, and its window-level drop handler fired beside the
main window's.
- The last global-selection writers that spelled the active-workspace check
inline now use the shared rule.
- Remove the save-dialog hook and item counter the old floating body used, and
correct comments that said the workbench only mounts once a workspace exists.
* fix(floating-workspace): keep a floating leaf detach from changing the main window's tab type
* fix(floating-workspace): give floating chats their workspace identity and directory
* fix(floating-workspace): read one visibility fact for chat attention and auto-ack
* docs(floating-workspace): correct comments the shared surface made stale
* fix(attention): one on-screen rule for every tab's unread marker, main window and floating panel
* fix(agent-session): pin the folder a session launched in; floating chats resume there or refuse
* fix(native-chat): resolve a floating chat's file links, images and skills in its pinned folder
* fix(agent-session): open a floating chat's terminal handoff in its pinned folder
* fix(agent-session): surface refusal text on chat open, require the launch-directory rule, and keep bookkeeping from gating a launch
* fix(floating-workspace): wait for a floating chat's pinned folder and read panel visibility from the store
* test(floating-workspace): assert the toggle event without a cast
* fix(native-chat): show why a chat could not be opened instead of dropping the host's refusal
* fix(floating-workspace): close chat tabs through shared commands
* fix(i18n): remove obsolete floating editor loading key
* fix(floating-workspace): preserve tab ownership on activation and browser close
* fix(floating-workspace): keep terminal activation and focus reclaim in sync
* fix(readme): point translations at tracked media
* fix(floating-workspace): retain split surface with an empty focused group
* docs(tabs): describe workspace-owned activation accurately
* fix(floating-workspace): finish the main merge and name the pin a launch directory
Port what the merge left on main's newer shapes:
- The floating chat's "folder is gone" refusal becomes a typed reason,
launchFolderMissing, worded like every other start refusal (copy, desktop
words, five translations). Older clients drop the unknown reason and show
the code's generic words.
- The PR's launch-folder tests run on main's journal-database record store;
the attach-level case is a row in the pre-spawn refusal table.
- Tests main added since the PR's base learn the PR's close-queue settling,
workspace-scoped notebook lookup and tab-stamped chat host.
- Move the store's visible-tab-index reads into the tab table, so the record
store stays under the line cap with the pin writer in it.
- Drop a duplicate Worktree import the merge left in a @ts-nocheck file.
Rename the persisted field, the status summary field and the renderer slice
from workspacePath to launchDirectory: it records the folder a session first
launched in, not its workspace's path, and it has not shipped yet.
* Fix floating chat visibility, closes, and directory pinning
* fix(i18n): drop the floating save dialog's strings with the dialog
The floating panel now closes through the shared unsaved-changes prompt, so
its own dialog's six strings had no caller and failed the runtime-catalog
check.
* fix(native-chat): pin floating founders and retain pinned editors
* test: align auto-ack fixtures with unified tab visibility
* test: give the forget-status attach fixture a location
Attach now reads params.location to decide whether a new record is a floating founder, so the partial fixture needs the location real attach params always carry.
* test: type the forget-status attach fixture instead of casting it
The cast tripped the changed-code type-assertion gate.
* fix(native-chat): read a floating chat's model defaults in the folder it was created in
The model picker read the floating setting's current folder, so a chat
pinned to another folder could show that folder's project default.
* test(cross-version): give the stub host the record store the catalog read now checks
* test(claude): give main's thinking-display launch stub the store shape the resolver takes
|
||
|
Brennan Benson
|
8d2e9264bc |
fix(native-chat): ignore the terminal launch command when starting a native chat (#25720)
A custom launch command in settings made structured native chat unavailable: the renderer route, the host launch-mode decision, and the host's create-support all treated it as terminal-only, so a user with the chat default on silently got a terminal. The command names the CLI binary and now applies to terminal launches only; native chat ignores it. A start directory outside the workspace still forces a terminal. The internal route field and blocker are renamed to that concept; the wire reason `tui_launch_command` is kept and its receipt text now names the start directory. |
||
|
Brennan Benson
|
761d63a4e5 |
feat(agent-launch): keep long prompts off the launch line and paste them after readiness (step 1 of 7) (#24257)
* feat(agent-launch): host-side prompt delivery for agent.launch The host's agent.launch typed any launch prompt into the shell as part of the launch command. A long or multi-line prompt then ran line by line in the shell, and an agent that never showed readiness or crashed at startup had nothing guarding where its text went. agent.launch now carries a prompt on the typed line only when the line stays one line, control-free and at most 512 bytes; otherwise the agent starts clean and the host pastes the prompt once the agent's own ready signal fires (bracketed paste plus its composer marker or a quiet render, read only after the shell's last hand-off, never while the pane's own shell is proven in front), with main's draft-paste bytes and an Enter 50 ms later. Orchestration worker starts wait on tui-idle as before. A replay-safe launch admits and claims its ledger row in one write, Qwen Code gets a second Enter, the desktop and phone share one launch-refusal classifier, and hosts advertise agent.launch.prompt-carry.v1. Split out of #23748, which moves the desktop source-control buttons onto this path. * fix(agent-launch): keep a short-lined multi-line prompt on a local zsh launch line, as main did #24257 moved every multi-line or over-512-byte prompt off the typed launch line and pasted it after readiness. The phone's AI buttons and review notes, whose multi-line prompts main typed whole into zsh, then reached Claude 0.5-3 s later and their RPC reply waited for the paste. The host now names the shell a local macOS or Linux line is typed into, the way the spawn picks it, and a multi-line prompt rides a zsh line when every line is at most 512 bytes and the whole line at most 8 KB. A real-zsh test types such a line through Orca's own ready barrier and startup write, including when a slow user config makes the write land early. Elsewhere the measured unsafe cases keep the paste: bash 3.2 runs multi-line lines piecemeal, fish drops an early multi-line write, and any shell loses a line over 1 KB written early. * test(agent-launch): keep the real-zsh launch-line test out of the Windows lane's gate scan The Windows lane registration check read `const ZSH_PATH = process.platform === 'win32'` (the head of a multi-line ternary) as a Windows-true flag, so `describe.skipIf(!ZSH_PATH)` looked like a Windows-only suite. The file is POSIX-only; the zsh lookup is now a function. * refactor(protocol): move the agent.launch capabilities into their own module Main's protocol-version.ts sits at the 300-line cap, so the prompt-carry capability pushed it over. The four agent.launch capabilities and their doc move to agent-launch-runtime-capability.ts, re-exported by name and spread into RUNTIME_CAPABILITIES at the same position; the advertised lists and every export are unchanged. * refactor(protocol): import the agent.launch capabilities from their own module `export *` from protocol-version left the four names undefined under the mobile recording loader, which resolves a relative import through a Proxy with no own keys, so 37 phone recordings lost agent.launchReplay. Importers now name agent-launch-runtime-capability directly; protocol-version only spreads its list. * refactor(agent-launch): drop the unshipped viewMode field and trusted local caller id Both were inert in step 1 and existed only for step 2. agent.launch will become a public plugin API, so every wire field is permanent once shipped; a top-level viewMode reads as "choose terminal vs chat", which the host decides. Step 2 introduces placement and view intent under a placement object instead. * fix(agent-launch): read Codex's provisional startup from the rule files' hold anchor Main (#24375) moved Codex's provisional-header check into codex.json's provisional_startup hold anchor and deleted codex-terminal-readiness.ts, so the launch readiness hold now asks showsHoldAnchor, as main's own settled check does. * fix(agent-launch): hold rule-file name titles to quiet for a launch, and census the zsh fixture Main (#24375) answers a name-only title from each agent's rule file ahead of the sustained-title lane, so gemini.json's name_title settled a launch readiness wait on the shell's auto-title while Gemini was still booting. A launch now asks quiet of every weak idle verdict, as that lane did. Main's readiness census requires a recorder for every runtime fixture; the zsh prompt recording is a non-agent control. Gemini's synthetic baseline is regenerated for this PR's stated change: a bare gemini title is no longer its rest mark, so name-only rows settle weak, and a fresh working or blocked status is no longer overridden. * fix(agent-launch): paste a launch prompt only when the launched agent is proven in front A launch pasted its prompt unless a shell was proven in the terminal's foreground, so any read that could not prove one let the prompt through. After an agent exited at startup, its shell turned bracketed paste on at the next prompt, readiness fired on it, and the prompt was typed into the shell: - macOS: a pane runs its shell under login, so the process-group fence's root was never the shell's group and never proved it; the cached foreground name could also still name the exited process. - Windows Git Bash and WSL: the shell-alone-in-its-job check never answers. Now one fresh read of the terminal's foreground decides: agent, shell or unknown. Only 'agent' lets a write through (paste, Enter, second Enter, reused panes too); 'shell' still drops a ready signal. A Windows host never proves the agent, so there the launch line carries the prompt at any size, as on main. * test(agent-launch): cover the Windows QA stub, a grok override that exits at once * fix(agent-launch): keep the local socket alive while a prompted launch waits for its agent A launch with a prompt now waits up to 60 s for the terminal agent to be ready before it writes the prompt, and reports not-delivered when the agent never is. The local runtime socket closes a connection idle for 30 s unless the request is a long poll, so a launch whose agent exited at startup lost its reply and the caller saw 'runtime closed the connection' instead of not-delivered. Classify a prompted agent.launch and agent.launchReplay as a long poll, as orchestration.workerStart already is for the same wait. * refactor(agent-launch): narrow the launch params by 'in' instead of a cast * fix(agent-launch): find a launched agent behind a wrapper that leads its process group A tcsh or nu launch line runs the agent from /bin/sh '<script>', and a wrapper script that does not exec its agent does the same: the wrapper leads the terminal's foreground process group and the agent is a member of it. The fresh foreground read names the group's leader, sh, so a prompted launch was refused or pasted late (M4Air tcsh: 2 of 4 not delivered, 2 pasted ~9 s late). Before that read, take the host's process-group observation as positive proof when it names the launched agent among the foreground group's members and is younger than a ready signal's quiet window. It never proves a shell. * fix(agent-launch): judge the foreground-group proof by when its capture began, not how long ps took The age the host stamps on a process-group observation runs from the start of its whole-machine ps, so on a loaded Mac a capture begun after the read was asked for still read as older than 1 s and the proof was dropped. Count an observation whose capture began after the read was asked for, less the window a shared capture is reused across. * test(agent-launch): keep the crash-guard live test out of the Windows lane's gate scan The Windows-lane registration scan read the const assigned from a platform check as a Windows-only gate, though the suite runs everywhere but Windows; find zsh in a function instead, as the real-zsh typed-line test does. Under load the fresh foreground scan can fail to answer, which lets the shell's prompt settle readiness (2 of 4 paired runs). The guard still refuses that write, so assert the refused write, the property that must always hold. * perf(agent-launch): read a local pane's foreground from its own terminal, not the whole process table The foreground read that gates every launch paste ran the daemon's inspectProcess capture and then a fresh scan, each a whole-machine ps; the fresh one also waits for any capture already running before it starts its own. Measured here at load 5: 1.2 s a read (M4Air QA: 3.4-5.0 s, and worker starts 17.6-32 s against main's 9-12 s at load 25-84). On a local macOS or Linux host, take the pane's root pid from the provider's session inventory and run one ps limited to that pane's terminal. Its foreground process group decides: the launched agent or any non-shell member is the agent (a wrapper that did not exec its agent leads the group), a group of shells alone is the shell. Same pane, same verdict: 2.7 ms a read. SSH hosts keep the relay's observation and name. * test(mobile): re-measure the web app's script sweep after agent.launch's capabilities moved out of protocol-version The mobile web bundle check failed at 124 assets against a ceiling of 123. Main already sat exactly on that ceiling: its sweep table read 69 scripts at 16 routes while the tree builds 73, the whole margin of 4. This branch imports the agent.launch capabilities from their own module, so protocol-version is no longer pulled into the root layout and four other routes. That moves which routes share which modules, and the Qoder capability module, imported by protocol-version and the AI-vault resume path, no longer shares an importer set with anything, so it gets a chunk of its own: 74 scripts. The fence says to re-derive the bound rather than raise it, so the sweep is re-measured on this head (every prefix of the sorted route list). The worst route now adds 10 scripts (session), not 9, which moves the pinned shell crossing from 32 to 30 routes; main re-measured on its own lands on the same crossing. * fix(agent-launch): a worker's brief needs its agent found in front, and Grok's start answers on its composer A paired-server worker start whose agent exited at startup typed its brief into the server's shell, which ran it: the idle wait can settle on a shell back at its prompt, and the brief was written with no foreground read. Both worker-start paths now check before each brief write, as a launch prompt is checked: on a host that can find the agent in front it must be there; on one that cannot (Windows) a shell proven in front still refuses, and anything else writes as before. A Grok worker start waited ~10 s more than main: its only rest signal is its bare name, which a launch holds to quiet output, and Grok animates its logo for ten seconds after its composer glyph. A worker start for an agent whose rest signal is its bare name and whose composer draws a marker (Grok, DSH, mimo-code) now also answers on that marker, whichever comes first. |
||
|
Brennan BensonandClaude
|
16784c1a67 |
fix(native-chat): name a chat write by its target, not the owner generation (#22812)
* refactor(native-chat): remove the unused terminal handoff No client ever called agentSession.requestHandoff or mounted the handoff chrome. Delete the handoff coordinator, the terminal-owner runtime, the proof write path and the unmounted UI. Keep agentSession.handoffStatus, which released desktop clients read for worktree activation, and let records an older build left mid handoff reconcile through the ordinary restart and recovery paths. * fix(native-chat): never let the pre-stop snapshot hold a chat's stop Eviction now drains delivered events before quit's resume-offer snapshot. An unbounded wait there sits ahead of the provider stop, so a sink whose journal write stalls kept the child running until the step deadline aborted the eviction. The offer is advisory: bound the drain and stop the child regardless. Co-Authored-By: Claude <noreply@anthropic.com> * refactor(native-chat): drop helpers only the terminal handoff called `claudeAuthEnvCarriedForward`, `isPathWithinDirectory` and `queryWindowsProcessRowsFresh` lost their last caller with the handoff. The fresh-scan tests now go through `queryWindowsProcessDescendants({ fresh: true })`, the teardown path that still depends on that contract. Co-Authored-By: Claude <noreply@anthropic.com> * docs(native-chat): stop citing the removed handoff in lifecycle comments Six comments still named the handoff coordinator, a handoff suspend, or a terminal-owned session as live participants in the flows they describe. Co-Authored-By: Claude <noreply@anthropic.com> * test(native-chat): type the stalled snapshot drain without a cast Co-Authored-By: Claude <noreply@anthropic.com> * test(native-chat): pin that a start dead before proving owes no settlement The removed restart handoff test pinned this branch; nothing else did. Co-Authored-By: Claude <noreply@anthropic.com> * fix(native-chat): keep the owner-status read behind an in-flight attach The handoff removal dropped the per-session queue from `handoffStatus`, so a read landing mid-start reported the reservation (no owner) instead of the settled chat owner, and shipped desktop clients blocked worktree activation on it. The read is queued again, as it was before the removal. Co-Authored-By: Claude <noreply@anthropic.com> * refactor(terminal): remove the agent-session PTY write gate The gate only refused a write when a PTY had been bound to a chat session, and the only code that ever bound one was the terminal handoff this branch removes. With it gone, every admit/readmit returned "admitted" unconditionally, so the checks on the renderer write path, the runtime controller backstop, terminal.send, agent prompts, preview input and orchestration pointers, the refusal fields on terminal.send and worker-start receipts, the plugin and CLI refusal copy, and the adopted-pane orchestration routing could no longer run. Ordinary writes take the same path in the same order as before. Co-Authored-By: Claude <noreply@anthropic.com> * refactor(native-chat): drop the transcript helpers only the handoff called appendLegacyTranscriptMessages fed the terminal transcript catch-up and proveClaudeTranscriptBranch backed the terminal owner's exit proof. Both lost their last caller with the handoff. Their tests now go through the live entry points instead: the roster bounds through the legacy import, the pinned-read and growth tests through the ancestry replay the history window uses, and the marker rules through the string proof in their own file rather than the session-file resolver's. Co-Authored-By: Claude <noreply@anthropic.com> * fix(native-chat): stop calling a starting chat "mid-handoff" A send refused because the chat's owner is not settled showed "The session is mid-handoff (<stage>)." in the composer. With the handoff gone, the stages that reach it are a chat that is still starting, or one whose previous agent process has not yet been confirmed stopped. The message now says which of the two it is. The refusal code is unchanged. Co-Authored-By: Claude <noreply@anthropic.com> * test(native-chat): type the stand-in roster decoder without a cast Co-Authored-By: Claude <noreply@anthropic.com> * refactor(codex): name the pinned rollout lookup for what it does With the terminal handoff gone, the module named codex-tui-rollout-proof holds only the pinned rollout lookup that structured Codex launches use to resume a thread, so the name described code that no longer exists. Rename the module and its options type. Also drop a mobile allowlist assertion that pinned the removed agentSession.requestHandoff method, which no longer exists to allow. * refactor(native-chat): type the owner-status reply as the host sends it The handoffStatus reply type still listed the terminal handoff's fields and states (terminal placement, host label, proof retry, queued and waiting phases, the to-terminal direction). No host writes them any more and the only client reader parses the reply as unknown, so they described nothing. The reply on the wire is unchanged. * refactor(native-chat): normalize terminal-handoff lease values once at decode Nothing in this build writes a terminal owner (`runtimeKind: 'tui'`) or the handoff's `preparing` / `old-owner-stopped` stages, but the in-memory types still admitted them, so readers across the host kept branches for values no path produces and the compiler could not point at them. The store now validates the on-disk shape, which still accepts those values so an older record is not quarantined, and maps them once while parsing: - `preparing` and `old-owner-stopped` become `recovering` - a `tui` lease becomes `native`; when it records a process it also becomes `conflicted`, the claim every build probes but never stops. A plain native owner would be stopped by restart recovery, here and in older builds. Revisions are taken over the normalized state on both sides of every compare, and the mapped record reaches disk with the store's first transaction, the same way the tab-id backfill does. The in-memory types narrow to what this build writes, and the branches that existed only for the removed values go. Structured-worker identity keeps its verdict for a former terminal owner by refusing a conflicted claim rather than a non-native kind. * refactor(native-chat): stop threading the owner kind through a reservation A reservation only ever names a native owner now, so the request no longer carries a kind and the reserved lease records `native` directly. The attach params keep `runtimeKind`: agentSession.ensure and create accept it, and the operation fingerprint stored in the ledger covers it. * test(native-chat): pin the legacy-lease rewrite with a transaction that changes nothing else Hiding a tab also committed the visibility index, so the no-op transaction wrote the file even when its open-time revision was wrong. Committing the index first leaves the pending rewrite as the only reason to write. * fix(native-chat): name a chat write by its target, not the owner generation A write carried the fence of the last frame the pane read, and the host refused it unless that fence was still current. An idle release and the restart after it each move the fence, and the release publishes nothing, so a send after a release was refused "Expected runtime fence 1; the session is at 3", and a Stop queued behind a cold start was refused as stale. Every write already names what it acts on: a send its conversation, a cancel its turn, a prompt answer its item revision, a rewind its epoch; an option is last-writer-wins. So admission stops comparing the client's fence, and the rebase that papered over one restart (admitAtResumedFence, resumedFromFence) goes with it. The writer-lease check stays, and so does the attach's compare-and-swap. Frames now stamp the fence read when each frame is sent instead of a copy each subscriber kept, which went stale on the same release. * docs(native-chat): say mutation admission checks only the writer lease * docs(native-chat): drop the send rebase from comments that still described it * fix(native-chat): keep each pane's own fence on frames so a failed restart is not resent * docs(native-chat): drop the fence from the admission the send effects run behind * docs(native-chat): give the fence move on release the reason that still holds * docs(native-chat): stop citing a write fence check in launch and mailbox comments Three places still gave the removed fence check as a reason: the launch replay said admission puts the ledger ahead of the fence, the launch surface said a send must name the lease it was admitted against, and the direct-mailbox path said the lease fence decides whether delivery is safe. Admission now checks only the writer lease. --------- Co-authored-by: Claude <noreply@anthropic.com> |
||
|
Brennan BensonandClaude
|
1069bb053f |
fix(agent-launch): a cwd at the workspace root no longer forces a terminal (#22729)
"Continue in New Session…" always names a cwd, and both the renderer route input and the host launch-mode decision read any cwd as a custom start directory, so the continuation opened a terminal agent even when chat was the user's default. Both now share one rule: only a cwd outside the workspace root (after normalising slashes, Windows case, WSL aliases and the distro's Linux spelling) requires a terminal. A subdirectory still does, because a structured session cannot start there. Co-authored-by: Claude <noreply@anthropic.com> |
||
|
Brennan Benson
|
069dc8a1d8 |
feat(agent-launch): let a caller reserve the chat session, and start terminal launches with the session picks (#22523)
* feat(agent-launch): let a caller reserve the chat session and carry session picks to a terminal launch * fix(agent-launch): keep a caller-minted session id named for its agent, and mint the fallback the same way * test(mobile): model the older host from the launch fields, not the refined schema * docs(agent-launch): describe the reserved session id as conversation identity, not placement The caller mints the session id so it knows which conversation it started; tab placement is not keyed on it. Also puts the terminal surface's doc comment back on createTerminalSurface. * docs(agent-launch): say a terminal launch reads the session picks on the wire contract The `sessionOptions` field doc still said a terminal launch ignores them, which this branch changed. * fix(agent-launch): check a reserved session id's token after the agent name, not the whole id A hyphenated agent name failed the one-token check, so any session id for such an agent was refused at the wire, while every other agent without a chat has its id ignored on the terminal. |
||
|
Brennan Benson
|
d60043787b |
feat(agent-launch): carry the launch inputs the host cannot derive (#22037)
* feat(agent-launch): carry the launch inputs the host cannot derive Desktop's launch call sites cannot move onto `agent.launch` while the wire drops inputs they depend on. This adds the three the host genuinely cannot work out for itself, and deliberately adds nothing the host can. - `agentArgs` — the host read only `settings.agentDefaultArgs`, so a saved launch recipe's arguments had no way across. Tri-state is preserved: `null` is "no arguments", absent is "use the settings default". - `cwd` — `TerminalCreateOptions.cwd` already reached the spawn, but nothing on the wire filled it. It also decides the route: only a terminal can start somewhere other than its workspace, so the host now feeds it to `requiresTuiLaunchCommand` and downgrades with `tui_launch_command` rather than running a structured session in the wrong directory. - `launchSource` — telemetry, and the only member of the `agent_started` triple the host cannot derive; `agent_kind` and `request_kind` are computed host-side. Typed `z.string()`, not the closed enum: params are validated by the HOST, so a closed arm set would let an older host refuse a newer client's launch over a label. Attribution must not gate a user action. Not added, because the host already derives them: `launchPlatform` (`getAgentLaunchPlatformForWorkspace`, from the same connectionId/path/ projectRuntime the renderer uses) and `startupCommandDelivery` (a pure function of the agent inside `buildAgentStartupPlan`). Fingerprint: `agentArgs` and `cwd` are in — they change what the call does, so a retry carrying different ones must conflict rather than replay. `launchSource` is out — two buttons producing the same launch are one operation, and folding it in would refuse an honest re-attributed retry. A caller sending none of the new fields digests exactly as before, because the canonicalizer drops undefined keys, so launches admitted by an older build still replay across the upgrade. Arguments reaching a structured route are ignored by an existing deliberate decision (the Agent SDK and app-server version their option sets separately from the interactive CLI), so the host reports it in `warning` instead of overriding the user's preference on the strength of a field that is not evidence about the surface. * fix(agent-launch): forward create-target launch inputs |
||
|
Brennan Benson
|
663d670878 |
feat(agent-launch): deliver a launch prompt to a terminal agent (#21891)
`agent.launch` could hand its initial text to a structured session but not to
a terminal. The contract already anticipated the terminal half — the
`handed-to-terminal` arm has been declared in agent-launch-intent.ts since the
receipt was written and had zero producers — and the executor's own docstring
recorded the assumption behind the gap: that a terminal's paste belongs to the
pane owner. That assumption is what this overturns. The host owns the PTY, so
it can write into one whether or not any window is open on it, which is why
mobile and the CLI got an agent and no prompt.
A terminal takes its prompt one of two ways, and which one is not a
preference. `argv` exists so multi-line and special-character text reaches a
CLI as one argument rather than keystrokes, and it has no readiness race
because the text is in the process's arguments at exec time. So an agent whose
CLI accepts a prompt argument gets it on the launch command, and only a
`stdin-after-start` agent — plus any reused terminal, whose process started
before the launch existed — is written to as a bracketed paste.
That fork is asked once. `agentPromptRidesLaunchCommand` is derived from the
same injection table `buildAgentStartupPlan` branches on, and
tui-agent-prompt-transport.test.ts pins the two against each other for all 37
agents, so adding an agent or changing its mode fails loudly instead of
silently dropping that agent's prompt.
Reused rather than rebuilt: `sendTerminalAgentPrompt` is the runtime's one
agent-prompt writer (bracketed paste, per-PTY serialization, lifecycle
generation pinning, per-agent submit timing, and local/WSL/SSH routing), gated
by `waitForTerminal('tui-idle')` — the same pair orchestration's worker
dispatch already delivers a preamble through. The agent-first create path
needed no new mechanism at all: `startupPrompt` already flows to
`buildWorktreeStartupForAgent`, and the launch had simply been stripping it as
a reserved field without re-supplying its own.
Receipts stay consequences of the act they name. `handed-to-terminal` is
reported only from a launch command that carried the text or a PTY write that
returned; everything unproven under-claims as `not-delivered`. No fourth arm.
The one inversion is a stalled submission, which the verifier raises after the
write: that is reported as delivered, because a resend would paste the whole
prompt a second time into an agent already working on it.
A prompt the launch command cannot carry is refused at the terminal-create
resolver rather than dropped, since that path returns options and has no PTY
to fall back to.
`delivery: 'draft'` remains `not-delivered` for both surfaces. The host could
paste a terminal draft without submitting it, but it cannot observe that the
composer accepted it, so a receipt claiming delivery would be a guess.
No call site is migrated, nothing is added to the wire, and placement and tab
creation are untouched.
|
||
|
Brennan Benson
|
754134fd67 |
feat(agent-launch): deliver a launch prompt from the host (#21155)
* feat(agent-launch): deliver a launch prompt from the host `agent.launch` created the surface and then reported the caller's text as `not-delivered`, always: delivery lived in the renderer, so mobile and any other caller got an agent and no prompt. The host now commits a `submit` prompt to the structured session it just created, through the same send path `agentSession.send` runs, and reports `journaled` with the transcript row's id. Nothing is queued — the durable record that the text is owed is the journal's own submission row, which the send appends before dispatching, so a host-side copy could only disagree with it. The outbox's entry and envelope builders are reused so this send is shaped exactly like a client's, fingerprint included. Everything else under-claims as `not-delivered`: a terminal's paste is observed by whoever owns the pane, a `draft` has no host-side home, and a refused or thrown send commits nothing. There is no fourth "maybe" arm — a caller holding one could neither resend nor drop the text — and dispatch doubt stays on the submission row where it already lives. * fix(agent-launch): recover committed prompt after send errors |
||
|
Brennan Benson
|
b66ef2e8a8 |
fix(agent-launch): resolve a launch scope, not a git worktree record (#21193)
* fix(agent-launch): resolve a launch scope, not a git worktree record `agent.launch` asked the runtime for a managed worktree record and then read exactly one field off it, `.id`. That record does not exist for every workspace a launch can run in, so the request refused launches the method could otherwise run: the floating workspace resolves to a scope with an id and a path but no worktree row, and `showManagedTerminalWorkspace` throws `selector_not_found` rather than hand back the id it had already resolved. A folder workspace survived that only because the resolver fabricates a worktree row for it. The scope is the answer that is real for all three kinds, so the launch asks for that instead. `showManagedTerminalWorkspace` is unchanged - callers that genuinely need the git record still get it, and still get the refusal. With floating now reaching the mode decision, the host must know which kind of workspace it resolved. The kind is derived from the id it resolved itself, never accepted from a caller, and the route module's existing `floating` blocker does the rest: a workspace with nowhere to keep a session runs a terminal agent. Behaviour change, deliberate: a floating-workspace `agent.launch` used to fail with `selector_not_found` and now succeeds as a terminal agent. That is what lets the floating titlebar agent button move onto the shared launch command instead of driving tab startup itself. No wire change: `AgentLaunchTarget` is untouched. * test(agent-launch): cover floating RPC workspace resolution |
||
|
Brennan Benson
|
4b87bc718e |
refactor(agent-launch): redefine the agent.launch contract (#20999)
* refactor(agent-launch): redefine the agent.launch contract
`agent.launch` has no clients yet, so the contract is redefined in place
rather than versioned.
- params require `operation.id`, pinned to the shipped operation-id mint so
the host can read the embedded timestamp back. No caller-supplied
fingerprint: the host derives its own.
- the result carries `disposition` ('created' | 'replayed', the same
vocabulary `RuntimeCreateAgentSessionResult` already uses) and a single
top-level `warning` instead of one on the terminal arm only.
- the prompt receipt becomes an outcome enum, so a receipt can under-claim
instead of reporting a bare `delivered: false`.
- the dead `customization` field is deleted, and the mode-reason union and
receipt are declared once in shared with main re-exporting.
- `clientMutationId` joins the reserved create fields, with a test pinning
the list to the create schema in both directions.
Contract only; no behaviour change and no ledger wiring.
* docs(agent-launch): stop calling the stripped set "agent fields"
`clientMutationId` joined AGENT_LAUNCH_RESERVED_CREATE_FIELDS, so three
comments describing the stripped set as agent fields now teach the wrong
model — including a SAFETY rationale, where a reader is trusting it most.
The rationale's claim is unchanged and still sound: deleting keys from a
parsed object leaves the rest the parsed shape.
* refactor(agent-launch): make the attempt id the launch's only idempotency key
Review follow-ups on the contract redefinition.
`operation: { id }` becomes a flat `clientOperationId`, spelled the way
`terminal.createAgentSession` and the structured mutation envelope already
spell the same concept, and admitted by the shipped
`parseAgentSessionOperationTimestamp` rather than a second copy of its
pattern — so `agent-session-host-authority` keeps the regex private.
The handler now dedupes on that id instead of the create payload's
`clientMutationId`. That field is optional, so keying on it left any launch
that omitted one with no idempotency at all, while the required attempt id
did nothing. Reserving `clientMutationId` is still right, but for the reason
the comments now give: `createManagedWorktree` never reads it, so a copy left
in the forwarded payload is inert while still reading as a guarantee. The
previous rationale — that it was a second live dedupe key — was not true.
`messageId` moves onto the prompt receipt's `journaled` arm so a producer
cannot report the text as committed without saying where, and `rpcCallerKey`
picks up the `terminal.create` call site it was lifted from instead of
shipping with no callers.
* docs(agent-launch): record why disposition is two-valued only for now
The ledger admits attempts whose outcome was never recorded, and neither
`created` nor `replayed` can say "I cannot tell you" — a caller handed
`created` for an unresolved attempt starts a second agent. Noted at the type
rather than in review, so whoever wires the ledger reads it where they edit.
* fix(agent-launch): keep contract within implemented guarantees
|
||
 Brennan BensonandMerge Sim
|
6da72383df |
feat(agent-launch): one executor for agent launches, exposed as agent.launch (#19849)
* refactor(agent-launch): make the launch-mode decision surface-neutral
`decideWorkerStartMode` was the only shared answer to "structured chat session
or terminal agent?", but it lived in an orchestration-named module and spoke
orchestration's vocabulary, so the other launch surfaces could not call it.
Move the decision to `main/agent-launch/agent-launch-mode` unchanged and leave
`orchestration-worker-start-mode` as the adapter that supplies the noun.
A worker is not a special kind of launch; it is the same launch with a dispatch
attached. Naming the receipt's subject is the only thing orchestration actually
contributed, so that is the only thing the adapter keeps: "worker" in both
sentences, plus the `--terminal` wording, which reads as nonsense anywhere a
`--terminal` flag does not exist. Both are pinned, because they are asserted.
No behavior change. The receipts are byte-identical for every reachable case,
proven by running the new pin against both implementations.
Also pins the wording, which nothing was holding. The existing suites assert
`toContain` fragments ('terminal agent', 'cannot create') and the CLI suite
asserts a receipt handed to it by a mock rather than one this code produced;
all six files stayed green against a deliberately corrupted vocabulary. A
dispatch receipt is the only place a structured-to-terminal downgrade explains
itself, so the whole sentence is the contract, not a fragment of it.
* feat(agent-launch): add the launch intent and the one executor that runs it
The sequencing around the launch decision was duplicated per surface, and the
duplicate is where the bug lives. A new worktree was created agent-first, so
its startup terminal WAS the agent and the structured branch below it could
never be reached — every new-worktree launch was a PTY regardless of the user's
default. Orchestration fixed that for itself in #19431; mobile and the CLI
still have it.
`executeAgentLaunch` inverts the order once, for everyone. When the preference
is structured the worktree is created with NO startup agent, the executing host
is then asked whether it can host a session for the workspace that now exists,
and only then is a surface created. The host verdict cannot be hoisted above
creation: `agentSession.createSupport` only answers for a workspace it can
resolve, which is why the decision stays in two halves.
Agent-first creation is deliberately preserved for PTY launches — it is what
sequences the agent's startup command behind the setup runner, so wait-for-setup
comes for free there.
What actually differs per surface is only how a surface is built (an
orchestration worker's session takes a dispatch hold and a mailbox a plain
launch must not take), so that is injected as a factory rather than branched on.
The intent also strips the reserved agent fields from a migrated create payload:
a caller moving off `worktree.create` passes its existing params, and a stale
`startupAgent` in there would re-create the very path this replaces.
Tests assert order and arguments, not just the resulting mode. Reintroducing
agent-first creation reddens 4 of 11.
* feat(agent-launch): expose the launch executor as the agent.launch RPC
Adds `agent.launch` — one host-side method that decides structured-vs-terminal and
creates the surface — wired to the real runtime factories: `createManagedWorktree`
for the workspace, forking on `startupAgent` exactly as the orchestration worker
path does; `createStructuredAgentSessionForWorktree` for a chat session; and
`createTerminal` for a PTY agent. Allowlisted for mobile, which is the surface the
routing gap was reported on.
`worktree.create` is untouched. Its `startupAgent` keeps meaning "spawn a PTY agent"
verbatim, because it answers with `agentTerminalHandle` only on that path: a host
that quietly routed it to a structured session would hand every older client a
response with no handle and no error. All new behaviour sits behind
`agent.launch.v1`, which the host now advertises and a remote client must negotiate,
so a client that does not gets today's behaviour unchanged.
* fix(agent-launch): drop the deleted draft-prompt blocker from the reason map
main removed the draft-prompt blocker in #19681 (a structured session now holds
an unsent draft), so the exhaustive Record no longer typechecks.
* chore(agent-launch): carry a SAFETY rationale on the agent placement cast
The type-assertion gate landed after this branch's base, so the new file's
copy of the worker-start cast is now a changed-code finding.
* chore(agent-launch): carry agent.launch through main's RPC typing and casting gates
The typed-method contract, the generated params catalog and the
`assertionStyle: never` casting scan all landed after this branch's base.
- AGENT_LAUNCH_METHODS kept an `RpcMethod[]` annotation, which widened its
method name to `string` and broke assignability; every sibling infers instead.
- `agent.launch` binds a schema under src/main, so it joins the catalog's
RPC_METHODS_WITHOUT_SHARED_PARAMS and the parity gate's hand-listed twin.
- The now-typed methods make most test casts unnecessary; the few that remain
carry the line-specific SAFETY rationale the casting gate requires.
* docs(agent-launch): stop the receipt-wording comment claiming a migration
The decision was never moved out of orchestration-worker-start-mode; this PR
adds a second copy beside it. Say so, and name the unenforced agreement.
* docs(agent-launch): stop the executor comment claiming a migration that has not happened
The header asserted two things the tree does not support: that every launch
surface routes through the executor, and that the mode decision "already lived"
in `agent-launch-mode`. `agent.launch` is the executor's only consumer, and
`orchestration-worker-start-mode.ts` is byte-identical (blob
|