mirror of
https://github.com/stablyai/orca.git
synced 2026-11-01 17:47:40 +00:00
3d812eb9bf54c39e80098bb6ce55f8498a3d3c8c
9
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
  
|
bfa0dc01ba |
fix(mobile): preserve long terminal and chat pastes (#31473)
* fix(mobile): preserve long terminal and chat pastes Agents can keep only the last ~1 KiB read of unframed input (#13332). Frame long single-line chat messages like multi-line ones, and frame Live input appends over 1 KiB (an OS paste) with the terminal's paste rule, including held text flushed before a terminal click. Co-authored-by: Carlos Baraza Haro <carlos.baraza@ieee.org> * fix(mobile): end the live input session after a framed paste Agents such as Codex collapse a framed paste into one token, so later per-character erases from the field mirror desynced the line. A long append to a terminal in bracketed paste mode now commits the whole field in one frame and clears the field, recording it like other session ends so a late native echo cannot send the paste twice; without that mode the paste stays unframed and editable. Chat sends measure the sanitized bytes when deciding to frame. --------- Co-authored-by: Carlos Baraza Haro <carlos.baraza@ieee.org> Co-authored-by: Kelvin Amoaba <97001695+AmoabaKelvin@users.noreply.github.com> |
||
|
Neil
|
0f47bb16e8 |
refactor(mobile): organize native chat prompts and permissions (#34189)
* refactor(mobile): organize native chat prompts and permissions * fix(mobile): preserve extensionless test fixture imports |
||
 Brennan Benson
|
23798eff7c |
Let chats follow the agent permission setting (#32334)
* feat(agents): store each agent's permission itself instead of typing flags into Arguments Terminals and new chats read one stored choice per agent (Use the agent's own settings, Ask for approval, Accept edits, Approve for me, Full access). Orca adds the flag at launch; Arguments hold only the user's own text. A one-time move lifts Orca's flags out of saved Arguments. Older clients still see and write the flag as part of Arguments; the host translates. * fix(agents): say the all-agents switch resets agents set separately; keep primitives unstyled * test(agents): the one-time move returns every launch-profile field * Fix permission carrier ownership and surface-specific defaults * Preserve uncertain permission operands and name the option inventory * Check env entries without indexing typed agent maps * Combine Chat permission capability imports * Use folder execution-host platform for permission environment * Preserve launch map replacement and acknowledge permission saves * Keep permission feedback fallbacks statically discoverable * Provide preload platform to Agents settings render tests * Preserve Gemini approval arguments under Full access * fix(settings): retain complete environment edits through save outcomes * fix(settings): narrow local failure scope and preserve host messages * fix(settings): retire argument drafts after acknowledged saves * fix(settings): keep Goose environment editable after reset * fix(settings): show resolved arguments for legacy hosts * Preserve pending environment resets and bound launch settings work * Use agent permission choices for new chats * Let Claude resolve chat permissions from its own settings * Keep optional settings fallback in repository metadata reads * Consolidate permission report imports * Update permission choice contracts for own settings * Read repository visibility defaults through the raw settings fixtures * Match existing translated Agents settings labels * Update readback and legacy host test expectations * Align Chat behavior tests with host permission capability * Fix permission compatibility and host-owned descriptions * Prepare whole Environment edits and retire unchanged drafts * Handle null explicitly in permission report decoder * Keep Environment settings transitions in an acyclic module * Update raw settings reader test fixtures * Place test fixture safety notes on assertion starts * test: pin permission settings work and recovery contracts * test: type permission settings work fixtures * test: cover migrated permission inputs and captured CLI grammar * Publish permission reports and reuse host capability state * Combine mobile option state React imports * Honor mobile status feed capability for permission reports * test: preserve typed permissions in WSL Claude restarts * Correct permission report child and fixture type contracts * Delegate older-host chats to explicitly selected own settings * Avoid rebuilding live ACP option catalogs * test(chat): place cross-boundary permission contracts in integration project * Preserve existing chat capability gates and complete report test fixtures * fix(chat): preserve newer Claude startup permission reports * test(chat): clarify startup report fixture provenance * Show reported permission modes in chat menus * Keep reported permissions available after chat mode picks * Remove obsolete guard from concrete permission selection * Report the installed approval policy for ACP chat menus * test: keep accepted mobile mutation fixture on current fence * test: cover installed ACP approval report for own chats * test: align capability scaling contract with concrete permission menus * Ship required permission help in English runtime catalog * Adapt permission persistence test to shared menu |
||
  OrcaWinandOrcaWin
|
fdcf887bf0 |
chore(lint): fix suppressed lint violations instead of ignoring them (#33294)
* chore(lint): remove disable directives that suppress nothing
Each was reported unused by the oxlint config that owns its rule (root,
casting, react-doctor, anti-slop, native-plugins), including 17 file-level
react-doctor blanket disables whose rule no longer fires. An unused
react-hooks disable was masking a react/purity finding in
useAgentBucketCounts (compiler rules skip functions with hook
suppressions); suppress that one explicitly where the code explains it.
* test(tab-bar): use typed tab/file fixtures instead of @ts-expect-error partials
* chore(lint): fix misc suppressed lint violations instead of disabling
- consistent-type-imports: the "vi.importActual requires inline import()"
rationale was wrong; use typed vi.mock(import(...), importOriginal) or a
type-only namespace import.
- no-control-regex: source-control commit/push failure summaries, shell PATH
hydration, and ipynb ANSI parsing reuse the shared ansi-escape-sequences
stripper and control-character pattern instead of private copies.
- no-explicit-any / no-useless-spread / no-this-alias: typed `this` in
bridge test mocks, drop blanket file-level disables, drop a this-alias.
* refactor(main): replace avoidable type assertions with checked narrowing
17 production cast suppressions in main/wsl-guest removed: untrusted IPC
and RPC payloads (Jira issue update, WSL guest plugin install, handshake
type) now narrow through isRecord/type guards and build typed results;
redundant casts dropped; hand-rolled object checks and errno casts use the
existing isRecord / hasErrorCode helpers.
* refactor(renderer,mobile): replace avoidable type assertions with checked types
13 production cast suppressions removed: RPC failures read through the
`ok` discriminant instead of casting to RpcFailure; react-native-web host
refs narrowed with instanceof HTMLElement; redundant reply/diff casts
become checked annotations; the TUI agent param uses z.custom(isTuiAgent)
like its sibling schemas; the agent-status slice cast that hid missing
actions is gone so the creator's return type checks the whole slice.
* chore(lint): drop dead tiptap wrapper; scope anti-slop mocking rule to test support
- safe-react-node-view-renderer patched handleSelectionUpdate, which
@tiptap/react 3.31.3 no longer has, so it always returned early. Use the
stock ReactNodeViewRenderer.
- The anti-slop test override only matched *.test.{ts,tsx}, forcing
file-level no-module-mocking disables in test harnesses and fixtures.
Match *-test-harness, *-test-fixture(s), .test-fixture and *.test.mjs,
then drop those disables and the now-redundant per-file exemptions.
* test: replace repeated cast suppressions with typed fixtures (main/shared/e2e)
Net -119 cast directives. Typed fixtures replace recurring casts:
unusedSshConnection/unusedOrcadHostTransport placeholders, a typed
readPersistedState harness read (exposed a test treating optional
httpProxyUrl as always set), real Response objects in fetch spies, an
updater main-window fixture, a CDP webContents adapter, one shared
releaseModuleExport for cross-version tests, and requireOverlayDir in
place of `as string` on possibly-undefined overlay paths.
* fix(runtime): validate loaded device rows and host scope instead of casting
Two cast suppressions claimed checks that did not happen:
- device-registry-file-read: "normalizeLoadedDeviceRow re-validates every
field" was false; deviceId, token, name and timestamps were never
checked. Rows without an identity are now dropped and the file is set
aside through the existing corrupt-file path (registry reads as
doubtful, so no revocation verdicts); a non-string token is stripped
so the row stays listed but cannot authenticate.
- client-ui: "only optionality differs" was false; any string reached
persisted workspaceHostScope. Narrow it with isWorkspaceHostScope and
ignore unknown scopes (wire schema unchanged for mixed versions). The
remaining cast now states what it actually covers.
* test(runtime,ipc): drop redundant cast suppressions; typed runtime store fixture
-98 cast directives across runtime/ipc tests: 61 casts that typecheck
without the assertion; a typed vi.fn for scoped invalidation; a
partialRuntimeStore fixture in place of `store as never` for
OrcaRuntimeService; typed worker terminal/authority builders in the
incarnation-fallback suite (the old stub lacked runtimeId, ptyId,
worktreeId and launchTokenHash) plus removal of stacked duplicate
directives there.
* chore(lint): typed vi.mock imports and explicit listener snapshot in shared-bucket tests
Remaining files from the misc-rule pass that overlapped other buckets:
codex-accounts mocks use the typed vi.mock(import(...)) form instead of
a consistent-type-imports disable; the daemon late-exit fixture copies
its listener set with Array.from instead of a no-useless-spread disable.
* refactor(github,mobile): narrow routing settings type; drop remaining avoidable casts
- work-item-routing cast a one-field object to AppState['settings'] under
a claim that readers only consume activeRuntimeEnvironmentId. Return the
existing RuntimeFocusSettings pick instead and narrow the cache-key
chain's parameters, so the compiler enforces that claim.
- mobile file preview reads refusals through the RpcResponse `ok`
discriminant rather than a cast.
- terminal document host-notify accepts `unknown` errors, removing the
message-bridge cast.
- native chat readability drops a redundant repo-list cast.
* test(renderer): replace repeated cast suppressions with typed fixtures
-140 cast directives (and ~150 unsuppressed casts of the same shapes):
- mockPanePtyArgs owns the connectPanePty fixture casts, so a wrong
argument order or fixture is now a type error; drop binding casts for
members PanePtyBinding already declares.
- Seed AppState tab rows with the typed makeTerminalTab; this exposed a
split direction 'row' that is not a TerminalPaneSplitDirection.
- Partial GlobalSettings use getDefaultSettings overrides; Jira-only
stores use createTestStore; runtime environment rows/statuses come from
a typed paired-runtime-environment fixture.
* chore(lint): drop no-module-mocking disable now covered by the test-fixture glob
* test(e2e): fix all 221 grandfathered e2e type errors; baseline to zero
The e2e typecheck baseline grandfathered 221 errors in 119 files. About a
third were tests drifted from the current API, several of which would
fail or mislead at runtime:
- terminal-shortcuts called a nonexistent xterm buffer method;
- the daemon-lifecycle fake lacked terminateOwnedTree, hidden by importing
its type from a module that does not export it;
- arguments in the wrong position (listProcesses(hostId, opts), a dropped
pty.write inputKind, an extra toWebTerminalSurfaceTabId argument);
- removed enum values ('drawer', 'idle', 'editor',
'session-insecure-origin') and a launch_source the telemetry schema
rejects;
- missing required fields (turnScope, childWorkOpen, providerIdentity,
reviewLookupOutcome, runtimeServiceAccount behind an `as RelayConfig`);
- dead reads (tab.customTitle, bucket.targets, changedPixels).
Also lifts variables that narrowed to `never` after poll-callback
assignment, and adds a typed xterm render-dimensions accessor. Includes
the cross-version suites' shared releaseModuleExport adoption.
* chore(typecheck): gate the now-clean e2e typecheck in local lint
The e2e baseline is empty, so the ratchet now rejects any tests/ type
error. CI already runs it via run-ratchets; add it to pnpm lint's
explicit ratchet list and replace the stale 'not yet clean' note.
* test(mobile): typecheck every mobile test; baseline 105 files -> 0
mobile/tests-typecheck-baseline.txt grandfathered 105 test files (531
errors) whose types CI never checked. Contract changes had landed without
these tests following, and type pins in them silently checked nothing.
- Root cause: salvagedOptional (src/shared/zod-salvage.ts) cast away
zod's optional marker, so every reply type built on it claimed
salvaged fields were always present. Return a real ZodOptional;
runtime is unchanged (the outer optional short-circuits undefined).
- Production now takes the port it uses: RpcOperationSender instead of
RpcClient for send-only callers, a narrow fetch signature, and only the
request fields push-receive reads.
- findAllByHostType/findByHostType replace `'Text' as never` lookups;
stubRpcClient gives tests a typed RpcClient.
- Real test bugs: settings-host-client-lifecycle "relay" hosts were
direct hosts (relayHostId no longer exists); two toEqualTypeOf pins
could never hold; a recovery hook ran with an undefined scope key;
a create-PR test cast to a never-imported type; "no URL" assertions
passed on null; plus stale required props, args and enum values.
* test(e2e): parse workerStart result instead of casting it
* test(runtime): typed partial runtime stores with production-shaped fixture data
Route every partial store passed to OrcaRuntimeService through
partialRuntimeStore with typed row builders, removing 60 cast directives
that hid fixture data production never produces:
- paired-client-navigation-test-harness keyed setWorktreeMeta writes by
`[object Object]` (it took an id where the store takes a target), so
reads never saw them;
- invalid worktree lineage origin/capture and incomplete workspace lineage;
- getWorkspaceSession returning null where the store returns the default
session; getGitHubCache returning undefined;
- orcad migration states missing migrationId/manifest/receipt, with a
toEqual pinned to that incomplete shape;
- partial repos, metadata, folders, project groups and notification rows.
* refactor(renderer): resolve suppressed React hook lint instead of disabling it
109 of 187 react-hooks/react-doctor suppressions fixed:
- ~95 exhaustive-deps disables only hid stable refs and setters passed
through controller objects; list them (identity never changes, so
nothing re-runs), plus a few real stale closures (tabId/worktreeId).
- Effects keyed on one value that read the latest props now use
useEffectEvent (WebConnect, skill freshness, Bitbucket dialog seeding,
Landing preflight, automation catalog, structured-session recovery,
file explorer).
- The conditional layout-owner hook is selected once at module load from
the build constant instead of a rules-of-hooks disable that also hid
the whole controller from compiler and react-doctor rules.
- A hook disable was masking a react/purity Date.now() in the palette's
recent-tabs memo; sample it when the snapshot is captured.
- Three identical debounced-search effects share
useTaskSearchDebouncedApply; layout snapshot merge helpers move out to
stay under max-lines.
Remaining suppressions are mount-once resources, invalidation epochs,
deliberately keyed fields, and test harnesses that mock the dispatcher.
* chore(ratchet): exempt test harnesses and fixtures from the focus-read ratchet
The ratchet skipped *.test-fixture.ts but scanned *-test-harness,
*-test-fixture(s) and *.test-support files. Those seed settings with
getDefaultSettings (which the ratchet discovers as a focus reader) and
never route host work, so 17 baseline entries were false positives, and
replacing a test's `settings: {...} as never` with typed defaults tripped
it. Match the repo's test-support names and prune (which also drops one
stale entry), and rename the frame-ordering harness to *-test-harness like
its siblings. A blanket "-harness" match would also exempt
e2e-terminal-pty-harness, which production imports, so the names stay
explicit.
* test: merge duplicate shared/constants imports
* revert(file-explorer): restore original tree load effects pending SSH e2e investigation
* test(runtime): drop RuntimeStore cast now that the fixture is typed
---------
Co-authored-by: OrcaWin <293788423+OrcaWin@users.noreply.github.com>
|
||
 OrcaWinandm4air
|
88a82c0a7d |
refactor(providers): import provider contracts directly (main) (#27813)
Move the 52 non-runtime, non-ipc importers of src/main/providers/types.ts onto the per-domain contract files. The barrel itself stays until the runtime and ipc importers move (3.4b/3.4c). Co-authored-by: m4air <m4air@Mac.localdomain> |
||
|
OrcaWinandm4air
|
a4a96fee70 |
refactor(pty): delete the SSH PTY provider and its registry slot (#27499)
* refactor(pty): delete the SSH PTY provider and its registry slot Nothing in production registers an SSH PTY provider, so the registry's SSH map was always empty: every SSH lookup answered undefined or threw the 'No PTY provider for connection' error. Delete the provider stack, its consumer-session and source-recovery contracts, the registry maps and the getSshProvider dependency, keeping each caller's existing result for an absent provider. The registry now dispatches only 'local'; any other host throws PtyHostNotDispatchableError, which keeps the renderer-matched message for SSH ids and is never read as a gone or exited PTY. * test(pty): keep the no-provider SSH routing cases in local-only form * test: reword a comment naming the deleted SSH PTY provider * test: drop an import the rebase left unused --------- Co-authored-by: m4air <m4air@Mac.localdomain> |
||
|
OrcaWinandm4air
|
314993c200 |
refactor(terminals): terminal input and inspection route by the PTY's owner (#27279)
* refactor(terminals): terminal input and inspection route by the PTY's owner A remote PTY id already names the server that owns it. inspect, confirm, send, verified send and the data subscription took settings only to fall back to the focused server for an id without an owner segment, so an old ownerless id could reach a server that does not own it. They now route by the id alone: an ownerless remote id is never dialed and never confirmed locally. Dropping the parameter removes the settings plumbing the native chat, agent draft paste and follow-up paths carried only for this, including the raw focused-settings default in sendBracketedPasteToAgent, and deletes getSettingsForAgentTabRuntimeOwner. * test(e2e): native chat SSH delivery and remote completion follow the narrower terminal signatures --------- Co-authored-by: m4air <m4air@Mac.localdomain> |
||
|
Brennan Benson
|
35e173a5b7 |
fix(native-chat): alert when a structured chat asks for approval or input mid-turn (#25766)
* fix(native-chat): alert when a structured chat asks for approval or input mid-turn A structured Claude or Codex chat that stopped mid-turn to ask for approval or a question raised no OS notification, no phone push and no unread dot: the host's attention feed only fired when a turn settled. The host now announces each newly pending prompt once, from the same feed, and owns the phone push for its own structured sessions; desktops keep presentation only. - host feed: a `prompt` edge per pending approval/question, opt-in on the existing stream; a clean settle that only restates an announced prompt is not re-announced to prompt-aware clients; failures always are - host push: an in-process subscriber pushes the host's paired phones with labels from in-memory metadata, and withdraws a prompt's alert once answered - one attention identity for desktop banner, phone push, dedupe and retirement; delivery dedupes on it instead of the per-workspace burst window - agentSession.acknowledgeAttention routes reading a chat to its owning host, which retires what it pushed for that session via its dismissal record * fix(native-chat): keep prompt alerts under their retirement identity * Fix structured attention read and delivery lifecycle * Allow later reads to retry failed attention retirement * Prevent delayed prompt alerts after accepted reads * Keep attention integration tests across process boundaries * Check structured capability before cursorless attention reads * Keep host retirement fixture in the Node typecheck boundary * Keep attention test seams in the mock boundary * Keep host notification formatting usable without Electron * Declare agents in restored-host attention fixture * Group session attention capability declarations * Retire structured alerts on every read and when a remote prompt ends - Mark read and Mark all read on a chat that was never opened, or is hidden, now cover what lit its row: the read boundary is the later of the transcript cursor and the newest edge the tab was handed, captured at the click. The one boundary closes the banners, retires relayed phone alerts and drives the host acknowledgement; an edge after the click stays live. - An alert from a host older than journal cursors has no position, so a pane read retires it again, banner and relayed phone push, as before this change. - A desktop-relayed prompt alert for a remote chat is withdrawn once that host's live status stops reporting attention (answered elsewhere, cancelled, host restart); cached status after loss of contact never settles it. - The attention acknowledgement is agent-session.attention-ack.v1 with a required journal cursor and strict params; the cursorless no-op is gone. - The dismissal store keeps a record whose origin a newer build wrote, dropping only that origin. - Host reconciliation runs on restore and when a prompt leaves the pending set, not on every journal commit. - Remove unused session-wide retirement (retire, retireMatching, retireMobileNotificationsMatching) and test-only store lookups. * Settle relayed prompts only from host-dated status; reads default to view - A remote chat's status and its prompt edges arrive on separate sockets, so a status row queued before the prompt could land after it and withdraw a relayed alert that was still pending. Settlement now needs a live, non-attention row whose host updatedAt (the journal's latest row time, never decreasing) is later than the prompt's host raisedAt. The owed prompt is cleared only once the settle call succeeds, so a failed call retries on the next qualifying row. - An acknowledgement without captured reads now reads as a view. Mark read, Mark all read, the hover-card jump and both dashboard acks pass 'explicit'; Activity's automatic per-turn read stays a view. * Re-judge relayed prompt settlement after each call; dashboard watching is a view - Settlement is one check of the current mirror, run on each status row, after each settle call returns, and when a prompt edge arrives. A row that lands while a call is out is no longer dropped, and an edge that arrives after its own resolution settles at once. A rejected call is retried only once the mirror holds a different row, so a failing call cannot loop. - The Agent Dashboard's card click acknowledges as explicit; watching the open dialog acknowledges as a view, through both the in-window drawer and the pop-out relay, and no longer re-acks the state the click just read. * Keep a reused read owner findable after its holders release it A chat pane memoizes its read owner. When every holder let go in one commit and the same owner was picked up again (React StrictMode's effect re-run in development, or a pane remounted in one commit, as on a tab move), the release dropped it from the registry and nothing put it back. The attention bridge then found no owner, so a view read carried no cursor: the unread dot cleared but the banner and phone alerts stayed. An owner now names its key again whenever it gains an activation or a subscriber and the key is free, never over a different live owner. * Type the ack test's execution host ids * Keep the turn-completion types in main's turn-completion wire module Main moved these types into agent-session-turn-completion-wire.ts while this branch moved them into agent-session-attention-wire.ts. They now live at main's path with this branch's additions (journal cursor, prompt attention, prompt arm); the unused completion key is dropped. |
||
|
Brennan Benson
|
468e4e1167 |
fix(native-chat): a prompt card owns the chat input until its answer lands (terminal-backed chat, desktop and phone) (#25761)
* fix(native-chat): an answerable prompt card owns the chat input until its answer lands * fix(mobile): a terminal chat's composer waits while its prompt card is up * test(native-chat): type the prompt card fixtures without casts * fix(native-chat): scope replies to acknowledged prompt occurrences * fix(native-chat): preserve answer ordering and verified delivery * test(native-chat): keep mock RPC client inside test boundary * test(native-chat): place mock fixtures in the test-only scope * Keep runtime comments within the module size limit * test: preserve prompt delivery coverage in desktop CI * Treat an older host's accepted write as delivered A newer desktop or phone talking to a host that predates the write settlement field read every accepted reply as "unconfirmed". Prompt cards never dismissed, the phone showed "Response unconfirmed" on every tap and ordinary chat messages were held as "Delivery unconfirmed". The reader now uses writeSettlement when present and otherwise keeps the host's whole-write accepted/refused verdict, exactly as before this branch. Only prompt answers ask for provider settlement; ordinary callers (follow-up delivery, paste drafts, option commands, composer sends) are back on the original contract, so the legacy-handoff error class, its flag, the sequence-only send helper and the mobile handoff hook are gone. * Keep terminal-pane Escape on the plain accepted write Every pane's Escape/Ctrl+C goes through pty:writeAccepted. This branch had switched that IPC to wait for provider settlement, which dropped the "remount this pane" signal for a daemon session awaiting recovery and could stall later keystrokes behind a slow daemon acknowledgment. pty:writeAccepted is back to its original local-only, synchronous write. Prompt answers opt into settlement with requireWriteSettlement on the same channel, and a settled refusal while the daemon recovers now sends the same remount signal. Ordinary verified sends regain their original fallback write. * Report a partly accepted local paste as unconfirmed A settled local write split into chunks returned plain false when a later chunk was refused after earlier ones were accepted. Callers read false as "nothing was written", so chat showed "Message not sent" with a prefix already in the agent's input. It now reports the write as unconfirmed, the same verdict the paired host gives for a partial write. * Hide the chat composer under a prompt card instead of unmounting it When an approval or question card took the input region, the composer unmounted. A message still waiting for its Enter was cancelled and its bubble deleted after the draft had already been cleared, so the message vanished without a notice; composer history was also wiped each time. The composer now stays mounted but hidden while a card owns input, so its state survives. A send that has not submitted yet is still stopped (its Enter would answer the card), but its bubble stays with "Message not sent" so the text is not lost. The composer ref is detached while hidden, so root typing, paste and reveal focus never reach it. * Keep an answered prompt hidden after the chat view remounts The "answered" dismissal lived in component state. Toggling chat to terminal and back, a PTY reconnect, or leaving the phone session and coming back while the approved tool was still running brought the answered approval back, and it then took over the input again. Desktop now keeps the answered occurrence per pane outside the view; phone keeps it per chat tab outside the controller. Both still retire it when the pane observes the prompt clear or change, desktop also when the tab retires, and both maps are size-bounded. * Update the prompt-reply reliability gate for the review fixes Older hosts' accepted answers now dismiss like acknowledged ones, the composer stays mounted under a card, and answered prompts survive a view remount. The gate's invariant, oracle, assertion list, new test files and the two latest evidence runs now describe that contract. * Let users hide a prompt card, keep Escape from denying, and gate only Send on the phone The chat input could stay locked behind a card the host never closes (for example after a Deny typed in the terminal), and Escape on a focused approval card denied the tool even when the user meant to close a picker. - A Hide control (chevron) on terminal approval and question cards, desktop and phone, hides that prompt occurrence and gives the input back. It writes nothing to the agent and uses the same per-occurrence dismissal as an acknowledged answer, so a new occurrence shows the card again. - On desktop, Escape on a card now does the same Hide instead of Deny, and a card that appears while the user is typing no longer takes focus. - On the phone, a card blocks only Send: typing, dictation and image attach keep working on the draft. The placeholder is back to the normal one. * Fix two comments that still called older-host replies unconfirmed Since an older host's accepted write now counts as delivered, the requireWriteSettlement comment and the reliability gate's oracle said the opposite of the code. Both now describe the current rule. * Collapse prompt cards to a strip instead of hiding them, and close the round-2 gaps Hide removed a card completely, so nothing on screen said a prompt was still waiting, and several edges let the chat type into a live prompt. - Collapse (the header chevron, or Escape on desktop) folds the card to a one-line strip above the composer; the strip's chevron expands it back. Collapsing writes nothing, frees the composer, and is disabled while an answer is still being written. Each pane or tab keeps the occurrence as answered or collapsed, so a remount restores the same view. - Questions now carry the host wait's start like approvals, so an identical question in a new wait shows again (desktop and phone). A transcript-only prompt, which has no wait start, is dropped when the view stops observing it, and a transcript still loading no longer clears a dismissal. - Desktop: while a card owns the input, the hidden composer cannot send or interrupt even if it still has keyboard focus, and the card takes focus in the same commit. A send the card retires no longer types Ctrl+U under it. - Phone: an Ask hides the heuristic card read from the same waiting status, and the dismissal store is scoped by host, worktree and tab. * Keep a collapsed card's partial answer, and scope its focus to its own pane Collapsing a question card unmounted it, so expanding it again lost the chosen step, selections and typed "Other" text; Escape typed in that text field collapsed the card. A card arriving while the user typed in another surface (sidebar, notes, a browser URL bar) also took the keyboard. - The collapsed card now stays mounted but hidden (and inert on desktop) under its strip, on desktop and phone, so a partial answer survives collapse and expand. Escape inside the card's text field no longer collapses it. The question card shows the same focus ring as the approval card. - A card takes focus only from inside its own pane (its hidden composer) or from the page body, never from a text field elsewhere. - Desktop and phone share one dismissal store in src/shared, bounded by the existing scope-cache helper, which moves to src/shared with it. - The card send imports the verified helper from its own module, and the phone files are split so each name matches its contents (header action, strip, lane selector). * Return focus to the composer after a prompt card collapses Since a collapsed card stays mounted, Escape or the chevron left keyboard focus inside the now hidden, inert card. The composer's reveal-focus took that as focus already in the pane and stood down, then the browser dropped focus to the page body, so typed keys went nowhere. Reveal-focus now treats focus inside a hidden or inert subtree as not in the pane and focuses the composer. On the phone, collapsing a card also dismisses the keyboard so a hidden reply field does not keep it. * Keep the question card's collapse chevron beside its Cancel button The question card header spread its three items with justify-between, which put the new chevron in the middle of the header. The title now takes the free space, as in the approval card, so the chevron sits next to Cancel at the right edge. * Run the prompt tests on the merged main Main now runs Vitest under Bun, which resolves a long data: URL import as a package name, so the SSH delivery test loads its bundled mobile module from a temp file instead. The phone prompt harnesses mock the live line that main's view now renders, and add Platform, which main's text-selection helper reads, the same way main's own view tests do. |