A mixed fan-out used to collapse the pill to its single most urgent
state, so a red dot beside "1 failed" over a child that was still
working read as the whole fan-out having stopped. Every state present
now gets its own mark and count, in the sidebar's status-group order.
Steer is the better default for Enter while an agent is running, but
flipping DEFAULT_CLIENT_SETTINGS alone would only reach new installs.
Client defaults are materialized into storage — the first load writes the
whole default blob and every save re-stamps it — so every existing install
already has "interrupt" on disk, and a stored "interrupt" cannot be told
apart from one the user actually chose. A one-time migration flips them
all; picking "interrupt" afterwards sticks.
The applied-migration marker sits in its own storage key rather than in the
settings blob: StoredAppSettingsSchema is strict and readValidatedJson
clears the value when parsing fails, so an unknown field there would wipe
every setting for anyone who downgrades.
migrateAppSettings owns both of its writes so the settings blob always
lands before the marker. A failed marker write leaves the migration to
re-run harmlessly; a failed settings write must leave the marker unwritten,
or the migration is lost and the user keeps "interrupt" for good.
* fix(app): preserve IME composition in form inputs
Form model publication was replaying changing default text into the native or DOM editing surface. Keep the editing surface authoritative, defer web publication until composition commits, and reserve resetKey changes for explicit text replacement.
* refactor(app): centralize text editing authority
Route every app text field, including the composer and plugin runtime, through one platform-owned editing primitive. Its type and runtime reject controlled value replay, its command handle owns programmatic replacement, and lint prevents raw React Native TextInput imports outside the boundary.
* fix(app): restore composer placeholder styling
* fix(app): defer composer updates during IME composition
* fix(app): read live DOM value from web composer getText
---------
Co-authored-by: Mohamed Boudra <boudra.moha@gmail.com>
Bulk archive keeps the panel mounted with zero rows while it is archiving or
after a failure — that is the state it lands in once it has archived every
finished subagent. The divider under the action then had nothing beneath it and
read as a hairline stranded at the bottom of the panel.
The archive-finished action sat at the foot of the track panel with nothing
between it and the subagent rows, so it read as one more subagent.
It now sits above the list behind a divider. The pointer enters the panel from
the pill below it, so the top edge is the far end of the trip: reaching a bulk
action costs a pass over every row it affects, and it can't land under a thumb
that was aiming for the nearest subagent.
It stays a row rather than becoming a Button. The panel is a menu surface, and
a bordered control inside a list of menu items reads as a foreign object — the
divider does the separating, not the chrome.
Track pills became menu surfaces when the trackers moved out of stacked
cards, but ComposerTrackRow kept the raw Pressable it had as an inline
card row. Nothing in that path reaches selectItem, so the popover or
sheet stayed parked over the transcript after opening a subagent, and on
iOS the action fired while UIKit was still tearing the sheet down. Rows
now hand the press to the engine and take closeOnSelect, as MenuItem
does; bulk archive opts out because its progress lands in the panel.
The running mark is the shared StatusRing rather than a still dot the
pill drew for itself, and the remaining dots read their colour from the
one bucket-to-colour map. The slot is sized for the ring in every state
so the label does not step sideways when the last child finishes.
Mounting the real menu engine in a test needed harness work: stubs for
two packages that ship untranspiled Flow, reanimated left out of Vite's
prebundle so its .web files win, and the gaps filled in the unistyles
test theme.
Layout measurement excluded pill margin on some renderers, collapsing the intended gap above visible tracks. Keep the two-pill rail on one line and derive its complete tail clearance from shared geometry so chat switching and layout callbacks cannot change the result.
Keep the roomier 16px separation on compact layouts while using 10px on wide layouts. Apply the same breakpoint contract to track and import-session pills so their spacing stays consistent.
Treat the appearance setting as the rendered base size so the type ramp scales predictably. Native starts one pixel larger while web and desktop preserve the authored 14px base, with legacy values migrated on load.
Keep agent and task pills as a transparent transcript overlay while measuring their rendered height for scroll-tail clearance. Share geometry and spacing with the import-session pill so borders, type, corners, and composer separation match across platforms.
The subagents track and the task list were stacked cards above the
composer. Each one pushed the composer further down the pane and painted
an opaque band the timeline stopped at, and both spent a full row on a
count that fits in a pill.
Both are now pills that open a panel — a popover on wide screens, a sheet
on compact ones, from the existing menu engine rather than a fourth
floating-panel implementation. The row frame is shared, so a subagent row
and a task row sit on the same rail whatever their contents.
The bar mounts inside the transcript's animated container and pins itself
to the bottom edge, so scrolled content passes under the pills instead of
stopping above them. It stays a child of that container rather than a
sibling floating over the composer: on Android a view outside its parent's
bounds paints but takes no touches, and every pill is pressable. The
container carries the same keyboard transform the composer does, so the
pills track the composer's top edge when the keyboard moves.
The tracker state moves out of the composer with it. None of it was
composer state; it lived there because that is where it used to render.
* feat(app): add persistent workspace side panes
Keep the common review and file-browsing workflow beside the active agent without tearing down pane contents. Workspace tabs now share one compact chip language, assistant file links route into the retained right pane, and diff/file trees share stable geometry.
* fix(app): localize pull request empty state
* test(app): update explorer pane coverage
* fix(app): preserve explorer pane workflows
* test(app): make explorer tab selection idempotent
* fix(app): persist workspace review pins
* fix(app): preserve sidebar scrim drag hook
* fix(app): restore companion pane interactions
* test(app): follow selected sidebar surface
Compact bottom sheets were still using the legacy independent-scroll gesture path, which intercepted Android sheet pans and row taps. Keep that path only for the desktop split viewport and let compact model browsers participate in sheet gesture handling.
* perf(app): keep composer typing responsive
The controlled web textarea pushed every keystroke through the draft store and broad composer tree before paint. Keep live text in the focused DOM input and publish draft and selection state after paint so persistence and reactive features remain intact without blocking the typed character.
* fix(app): preserve live composer state at actions
* fix(app): use live input for autocomplete clicks
* fix(app): keep submit activity atomic
Use the shared interface text tier across the workspace header, conversation, and composer so the working surface no longer shifts between adjacent font sizes.
* feat(agents): add active-turn steering
Let users choose whether an immediate prompt interrupts or joins the active turn. The daemon owns exact-turn admission, provider delivery, safe fallback, and canonical timeline identity while queueing remains client-side.
* fix(agents): stabilize steering history and replay QA
* fix(agents): preserve steering timeline order
* fix(agents): reconcile resumed provider history
* fix(agents): serialize steering with cancellation
* test(agents): prove Claude multi-tool steering
* test(agents): force explicit history hydration
* fix(agents): flush output before steer admission
* perf(app): reduce warm workspace switch work
Retained workspaces avoided mounts but pane activity still rerendered the full composer, message input, and agent-control presentation. Keep the activity contract intact while moving focus-sensitive behavior behind small context consumers, and add a Playwright benchmark that measures rapid retained workspace switching on one browser clock.
* perf(app): accelerate retained workspace switching
Avoid repeated retained-stream layout reconciliation and unrelated workspace structure recomputation. Use the lightweight route transition while the host is already focused, while preserving targeted root POP_TO for returns from app-wide routes where it prevents duplicate hidden decks.
* perf(app): defer composer focus during workspace switches
* perf(app): retain recently used workspaces longer
Keep up to ten workspace screens warm so normal multi-workspace navigation does not repeatedly rebuild recently used timelines. Inactive screens expire after ten minutes, while the active screen remains protected and capacity eviction stays least-recently-used.
* fix(app): reconcile retained timeline geometry
* fix(app): refresh retained reading position
* feat(plugins): add local plugin surfaces and RPCs
Keep plugin execution scoped to its daemon while allowing one local TypeScript entry point to contribute validated RPCs and native client surfaces. Paseo owns host selection, navigation chrome, and per-installation query caches.
* feat(plugins): add Linear attachment source
* fix(nix): refresh npm dependency hash
* test(server): exclude internal sessions from Hub counts
* test(server): distinguish Hub execution sessions
* feat(plugins): make local plugins explicitly opt-in
Keep plugin loading dormant until configuration enables it, and collect unsupported examples under one directory.
* refactor(plugins): establish host boundaries
* fix(nix): refresh plugin dependency hash
* fix(plugins): complete packaging and startup rollback
* feat(plugins): add managed local lifecycle
Make trusted local plugins recoverable and authorable through daemon-owned lifecycle operations, host settings, and a capability-gated CLI. Reloads fully stop the previous plugin before starting new code, while generated projects provide strict TSX checking without changing runtime dependency ownership.
* fix(nix): refresh plugin dependency hash
* fix(app): preserve native composer input semantics
Native text was round-tripped through application state on every keystroke, which could overwrite an input method's in-flight edits. Keep native text ownership inside the input and reserve application-driven replacement for explicit draft mutations.\n\nRoute native clipboard images through the existing persisted attachment pipeline so keyboard and system paste actions behave like other image sources.
* fix(nix): refresh dependency hash
The native composer dependency changes the fixed npm dependency closure used by desktop builds.
* fix(app): keep composer locked across image pastes
Count concurrent native image persistence operations so submit remains disabled until every pasted image is attached.
* fix(app): replace composer text through the DOM on web
Web TextInput refs are DOM elements, so application-driven draft replacement must update their value and selection without native setNativeProps.
* fix(app): preserve web IME composition
Carry the Chinese IME fix from #2990 into the web composer input boundary.
Co-authored-by: Jimmy Lee <jimersylee+github@gmail.com>
---------
Co-authored-by: Jimmy Lee <jimersylee+github@gmail.com>
* fix(app): keep profile modes scoped to providers
Applying a profile through sequential field setters allowed a provider change
to race mode persistence, poisoning modeless provider preferences with another
provider's mode. Apply profiles as one form transition and heal invalid
provider-scoped mode preferences when profiles are used.
* test(app): cover profile features across providers
* feat(agents): surface live task progress
Keep current task state above the composer and turn task snapshots into semantic timeline activity. Normalize provider task protocols while preserving genuine plan flows.
* fix(agents): admit tasks through stream sequencing
* test(agents): expect normalized task status
* Search models across all providers from the picker root
* Store agent profiles in daemon config and apply agent config in one RPC
An agent profile is a named bundle of provider, model, mode, thinking
option, feature values and notes, stored host-wide next to terminal
profiles. It carries no system prompt: AgentSessionConfig.systemPrompt is
creation-only, so a profile holding one would apply when starting an agent
and silently do nothing when applied to a running one.
Applying a profile through the four single-field RPCs can fail halfway and
leave an agent matching no profile, so agent.config.apply takes the whole
bundle. Model runs first because a thinking option only means anything
against the model that offers it.
Old daemons parse their persisted config strictly and would drop an
unknown agentProfiles key, making a save appear to succeed and vanish, so
both capabilities are advertised on server_info for the client to gate on.
list_profiles exposes the profiles to orchestrating agents, which read the
notes to pick one and copy its values into create_agent.
* Add the agent profiles settings section
Profiles live behind one module boundary so removing the feature is a
directory plus one mount, which is what favourites never was. The picker
and composer will ask for rows and an apply function rather than learning
what a profile contains.
The edit form follows the form-model pattern instead of the terminal
profile modal's per-field state, because provider selection cascades into
which models, modes, thinking levels and features exist below it.
The section is gated on server_info.features.agentProfiles: an older daemon
drops the key on write, and a save that appears to work and then vanishes
is worse than an unavailable state.
* Document agent profiles and name the four senses of profile
The glossary reserved Profile for HostProfile, terminal profiles owned the
user-facing word, and custom-providers.md used it for provider aliases.
Agent profile is now the term, with all four cross-referenced.
data-model.md never documented terminalProfiles either, so both profile
lists are backfilled together along with the whole-list patch semantics and
why absent and empty differ.
* Pin agent profiles to the model picker and delete favourites
Applying a profile writes its values into the agent controls and is then
forgotten, so rows carry no checkmark and no selected state: after applying,
nothing is selected. Chevron still means navigate deeper, so the absence of
one is what marks a row as an action.
A running agent is its provider's process and agent.config.apply carries no
provider, so the live picker only offers profiles matching the agent's
provider rather than showing rows that cannot do what they say. The draft
path has no such limit and switches provider freely.
Draft feature values go through preferences rather than the feature
callback alone, because the draft feature hook clears local values whenever
the provider changes and a provider-switching profile would lose them.
Favourites are deleted rather than migrated: they were device-local
AsyncStorage keyed by provider and model, while profiles are host-wide
daemon config. ProviderSelectionModelRow.favoriteKey stays as row identity
and moves to its real owner; renaming it is its own commit.
* Cover agent profiles and cross-provider search with E2E specs
Model rows expressed selection only as a checkmark glyph, which assistive
tech cannot see and a spec cannot assert. Rows that draw the indicator now
carry aria-selected; profile rows deliberately carry none, because a
profile is an action rather than a selection. That is what makes "nothing
is marked selected after applying a profile" a real assertion instead of a
tautology, with the materialized model one level down as the control.
The search fixture extends claude rather than mock: provider-config.ts
only allows the six shipped providers as extends targets, so a derived
provider cannot inherit the dev-only mock provider.
* State what applying a config bundle does on partial failure
applyBundle is not a transaction and the earlier commit message implied it
was. A step that rejects aborts the rest and the steps that already applied
stay applied, which a profile omitting modelId can reach: provider setters
validate against the agent's live model, so mode can land while a feature
that model does not support is rejected.
Nothing is hidden by that. Every setter emits agent state before returning,
so what applied is streamed to the client next to the rejected response.
Rolling back would re-invoke the setters that just failed, usually against
a session that is gone, and pre-flight validation would restate each
provider's model-dependent checks where they would drift from the setters
that own them.
What sending the bundle in one request removes is the multi-round-trip
failure class: interruption between hops, and interleaving with another
client. The test already pinned this behaviour; its name now says so.
* feat(agent-profiles): refine profile discovery and appearance
Make profiles discoverable from empty and provider-scoped model pickers, keep global search stable and virtualized, and move profile settings under Agents. Replace emoji identity with a shared Lucide icon and host-color picker, while requiring concrete provider selections so profiles materialize predictably.
* fix(agent-profiles): defer profile swatch styles
Resolving Unistyles proxies at module load can freeze the temporary startup color scheme before the persisted theme is restored. Keep only style-key names at module scope and resolve the selected swatch during render.
* fix(app): restore composer control tooltips
Composite composer triggers dropped tooltip interaction props or the anchor ref, so hover either never opened the tooltip or left it off-screen. Preserve pressable props and compose menu refs at the trigger boundaries.
* fix(app): align mode tooltip typography
* feat(app): launch a terminal from the New workspace screen
New workspace only ever started a chat agent. Empty workspaces existed but
were reachable only by submitting an empty composer, which nothing advertised,
and terminals could not be started from this screen at all. People concluded
Paseo did not support terminals.
Adds a launch chip to the meta row, shaped from the same ComboboxTrigger as
its neighbours, that chooses between Chat and a terminal. Terminal profiles
mark where a typed prompt goes with a {{{prompt}}} sentinel in their command
or args, substituted client-side before create_terminal_request, so the daemon
learns nothing new and the protocol is untouched. A profile without the
sentinel takes no input: the composer goes read-only and shows the exact
command, so there is nothing to type wrong.
Prompt forms come from each CLI's own --help. claude, codex, and pi take a
trailing positional; opencode's positional is the project directory, so it
uses --prompt=. Profiles predating the sentinel adopt the right form on read,
keyed on the command's base name rather than the profile id, because profiles
created in settings get generated ids.
* fix(app): stop existing launchers passing the prompt sentinel literally
Adoption made every caller of resolveTerminalProfiles see sentinel-bearing
args, but only New workspace substituted them. Pinned targets and the
workspace terminal menu forwarded profile.args straight through, so launching
Claude from an existing workspace ran `claude "{{{prompt}}}"` with that text as
the prompt.
Every launcher now goes through resolveTerminalProfileLaunch. Surfaces with
nowhere to type pass an empty prompt, which drops the prompt-only arg and
gives back the bare command they had before.
Also from the same review:
- Adoption skipped when args lead with a subcommand. `opencode run` takes a
positional message rather than --prompt, and `codex login` takes no prompt at
all. Only the first arg is checked, so `pi --thinking high` still adopts.
- Settings edits the persisted list again, not the adopted view. It writes the
whole list back on any change, so resolving first baked read-time adoption
into the user's config the moment they reordered a row.
- A hidden draft no longer names a workspace. Text typed for a prompt-taking
profile stayed in state after switching to a read-only one and still seeded
the generated title and branch.
- Windows .cmd escaping now has a round-trip test: a real cmd.exe runs a real
.cmd shim that echoes its argv, covering metacharacters, an injection
attempt, and newlines. The existing tests only asserted the command line we
generate.
- Fixed an ENOTEMPTY flake in the terminal suite. kill() only signals, so the
shell was still writing into $ZDOTDIR while its directory was being removed.
Cleanup waits for exit instead.
* fix(app): keep the host and isolation accessible names stable
Wiring the new tooltips also replaced these two accessible names with the
tooltip copy, which broke `sidebar-project-grouping` at the host picker: its
selector is `getByRole("button", { name: "Host", exact: true })` and the button
had become "Choose the host". It timed out on both attempts in CI.
An accessible name identifies a control; a tooltip explains it. Conflating them
renamed a control that other surfaces address by name. The names go back to
what they were and the tooltips keep the "Choose the ..." copy.
* test(server): stop temp-dir teardown failing unrelated terminal tests
The ENOTEMPTY flake survived both earlier attempts. Waiting for the session to
exit shrank the window; retrying the remove did not close it either. The writes
come from processes zsh forks on the way out, which outlive the pty and answer
to nothing the test holds.
Teardown is now best-effort and logs what it left behind. A temp directory we
cannot remove is not a defect in the terminal code under test, and failing on it
reports the race against whichever assertion ran last, which is exactly how this
surfaced: as a flake in an unrelated zsh title test. 10 consecutive runs green.
Also replaces a cast in the Windows .cmd helper with a real narrow, so a change
to ResolvedTerminalCommand fails loudly there. Test files are excluded from
typecheck (tsconfig.server.json), so a cast in one is never checked at all.
* fix(app): keep terminal launch policy in its owning workflows
Profile adoption now parses known global option values before deciding whether argv already contains a positional. Workspace naming is derived inside the terminal workflow so hidden launch-only drafts cannot affect branch or title generation.
* fix(server): preserve cmd shim arguments through both parses
npm command shims forward %*, so metacharacters lose one escape layer before the target process receives argv. Exercise the production node-pty string command line against a forwarding shim, and only tolerate the known ENOTEMPTY teardown race.
* fix(server): stabilize Windows subprocess tests
* test(server): isolate Windows argv recorder output
* test(server): allow native watcher integration to settle
Retained tabs stay mounted under display:none while inactive, which
reports a zero-width onLayout. ControlledAgentControls fed that
straight into its density resolver, so a hidden tab's toolbar
collapsed to its tightest density and painted one frame collapsed
before re-expanding when the tab became visible again.
Extract the zero-width guard that useContainerWidthBelow already had
into a shared readMeasuredWidth helper, and use it in the composer
toolbar's own layout handler so a display:none measurement can never
drive its density.
Also bump the chat outline rail's MIN_PANEL_WIDTH breakpoint (818 ->
918) so the rail disappears earlier under a narrower panel, and widen
the chat-outline e2e spec's viewport so its panel keeps healthy margin
over the new threshold.
Route composer attachments through the shared compact menu sheet so headers and rows use the same alignment rails. Base the final-row padding on the device safe area to keep actions clear of system navigation.
* fix(composer): keep background context out of loading state
Pull request link resolution can block a new workspace until its checkout target is known, but it is not submission work. Keep that safety gate while reserving the send-button spinner and drop lockout for actual submission or upload progress.
* test(composer): cover link resolution at browser boundary
Exercise the real new-workspace composer while pull request context resolves, including the submit presentation and file-drop path. Keep the two presentation booleans local to their only caller instead of exporting policy solely for an internal-state unit test.
* fix(composer): preserve background lookup ownership
Composer edits could release the resolving gate while a pull request lookup still owned the request, briefly allowing workspace creation from the wrong branch. Track logical candidates independently of unrelated draft changes and settle each reference independently so only invalid work releases early.
* fix(composer): preserve pasted PR selection order
* fix(composer): renew re-added PR lookups
* fix(composer): track pasted PR source order
* fix(composer): preserve resolved attachment order
* fix(composer): coordinate active lookup order
Native composers had no path from the device clipboard into image attachments. Route clipboard data URLs through the existing attachment persistence flow and expose the action only in the native attachment sheet.
* Reapply: Stop completed turns from appearing stuck (#2484)
* Reapply: Keep older chat history and image previews stable (#2490)
* Lock timeline layout stability with user-driven specs
Two reds reproduce the reported jank:
- new agent, first prompt: the row paints at top=100 then jumps to 144 through
authoritative hydration, because the draft tab and the agent panel each render
their own AgentStreamView for the same logical row.
- long loaded timeline, submit: scrollHeight collapses 10793 -> 9111 mid-turn,
because splitOrderedTail reclassifies measured rows into the virtualized segment
where the virtualizer falls back to estimateStreamItemHeight.
Also restores centered timeline rows: the pagination anchor wrapper introduced in
#2490 was a plain block div, which broke alignSelf: center.
* Measure the turn footer blink from when the footer appears
Bounding the window at the prompt row conflated two things: the footer arriving a
frame after the row, and the footer leaving the layout and coming back. Bounded from
the footer's first frame, the blink still reproduces, so the footer-ownership race is
real rather than a sampling artifact.
* Rename the shadowed prompt row binding in the frame recorder
* fix(desktop): allow concurrent worktree dev instances
Desktop terminals can inherit another instance's profile path and debugging flags. Scope the profile to the current dev root and select an available debugging port so instances do not share global runtime state.
* fix(desktop): expose development providers in dev builds
Desktop-managed daemons inherited no Paseo runtime mode, so dev builds constructed the production provider registry. Mark unpackaged internal launches as development while continuing to force packaged launches to production.
* fix(timeline): stabilize optimistic turn layout
Keep initial history loading dormant until user intent and preserve active-turn layout across delayed transport updates. Add painted-frame coverage for optimistic timeline transitions.
* fix(timeline): preserve optimistic prompts through hydration
Canonical replacement relied on a transport submission record that can settle before a provider echoes the user row. Keep presentation continuity owned by the unreconciled row while destructive history replacement remains governed by active transactions.
* fix(timeline): make submitted prompts canonical
Submission settlement previously depended on status updates that could arrive after the optimistic row. Persist every accepted prompt in the canonical timeline and keep the footer active until RPC and canonical acknowledgement converge.
* fix(timeline): stop older history pages repeating entries
Backward pagination counted source sequence rows, so page boundaries inside a merged projected item could return that item repeatedly. Count projected entries and advance from their display anchors instead, with server and browser regressions covering the contract.
* fix(timeline): resume agents from the latest history
Use a bounded tail probe on resume so long background runs appear atomically instead of replaying every missed page. Preserve unchanged views, reconcile local/live rows, and keep skipped history reachable through backward pagination.
* fix(timeline): reconcile turn state after missed updates
Turn completion events can be missed while an agent timeline is unsubscribed, leaving footer activity stale even after the directory reports idle. Reconcile the single-owned turn lifecycle from authoritative resume snapshots while preserving newer live evidence.
* test(timeline): cover running state across refresh
Protect the hydration boundary where directory state must restore consistent tab, composer, and footer activity. Expose the running indicator semantically so the regression checks user-visible state instead of spinner implementation details.
* fix(timeline): keep history loading geometry stable
Reserve a fixed history-start gutter regardless of pagination state so showing or hiding the loader cannot shift timeline content.
* fix(timeline): keep turn activity coherent
Derive footer, tab, and composer activity from one lifecycle and submission projection. Idle directory transitions now request an authoritative resume so missed terminal events converge without treating directory delivery order as turn authority.
* fix(tests): align checks with runtime boundaries
* fix(desktop): stop dev app surviving its terminal
Electron was detached from the desktop runner, allowing it to outlive a stopped workspace terminal with broken output pipes and retain the single-instance lock. Keep Electron owned by the runner and preserve explicit process-group cleanup for detached child trees.
* fix(timeline): preserve activity through turn handoff
Correlate lifecycle events and snapshots with daemon-owned turn identity, and keep optimistic submissions active until their canonical row arrives. This prevents stale updates and RPC timing from exposing a false idle state.
* refactor(timeline): remove submission timing state
* fix(timeline): restore prompt identity on resume
* fix(timeline): exclude attachments from canonical prompt text
Structured provider prompts can include hidden history attachments before the user's text. Persist only plain prompt blocks so authoritative user messages match what the composer displayed while provider input remains unchanged.
* refactor(agent): remove unused pending-run message identity
* fix(desktop): let Chromium allocate the debugging port
* refactor(agent): remove impossible finalized-turn tracking
* fix(timeline): load history after keyboard scrolling
Treat observed upward displacement as user intent so keyboard and scrollbar navigation can wake dormant pagination.
* test(server): preserve accepted turn stream order
The manager-owned accepted turn boundary intentionally precedes coalesced output. Assert that lifecycle event explicitly while retaining the timeline-before-terminal ordering contract.
* test(app): scope sidebar fixtures by identity
Shared E2E daemon state may contain unrelated projects. Keep stale-route coverage tied to its named missing project and drag only the two rows seeded by the reorder spec.
* fix(timeline): gate prompt revisions by client capability
* fix(timeline): finalize canonical prompt delivery
Treat provider identity as a single stored-row transition so legacy clients receive one deliverable canonical prompt. Release identity-pending rows before later foreground output or the current terminal while preserving stale-turn isolation.
* fix(server): preserve legacy prompt pagination
* fix(timeline): keep assistant block identity stable
Assign canonical block IDs during live promotion so terminal finalization cannot remount image-bearing rows.
* fix(timeline): apply covered revisions during resume append
Run the exact canonical revision pass before overlapping resume pages append newer rows, while leaving cursor advancement to the incremental path.
* fix(server): preserve queued agent transitions
Project each queued live snapshot before awaiting stored metadata so later lifecycle state cannot overwrite an earlier transition.
* fix(server): snapshot queued agent updates
Capture each live agent payload before it enters the per-agent queue so shared nested state cannot change while an earlier storage read is pending.
* docs(app): date agent turn compatibility cleanup
* refactor(timeline): mint submitted prompt identity at acceptance
Canonical prompt rows now keep one immutable wire identity while provider identity remains internal for rewind translation. This removes the unreleased same-sequence revision and legacy projection contract.
* refactor(app): remove prompt revision reconciliation
Submitted prompt rows are immutable on the wire, so resume and live reducers no longer need to apply already-covered same-sequence updates.
* test(app): cover submitted prompt rewind roundtrip
* docs(timeline): describe immutable submitted prompt identity
Provider echoes now enrich daemon-internal identity only; wire rows keep their acceptance-time identity and rewind performs the provider translation.
* test(server): validate mock rewind targets
Reject unknown provider message IDs so app rewind coverage exercises wire-to-provider identity translation.
* fix(server): validate rewind before interrupting
* fix(server): omit rewind identity for daemon prompts
* test(app): guard rewind eligibility and rejection
* fix(server): preserve prompt identity across hydration
* refactor(timeline): remove stale revision remnants
* Revert "fix(server): preserve prompt identity across hydration"
This reverts commit 6adcf7e5b0bc8601761af59b461a746cb5e395fa.
* fix(timeline): settle missed submission acknowledgements
Dictation auto-send bypassed the controlled composer update, so preserved draft surfaces could lock while showing an empty prompt. Publish the transcript before invoking the existing submit lifecycle so pending and failed submissions retain the user's spoken input.
* Switch reasoning, mode, plan, and fast from the Command Center
Extend the Command Center (Cmd-K) with selectors for a focused agent's
Thinking (reasoning effort), access Mode, Plan mode, and Fast mode,
mirroring the model switcher. Typing surfaces tiered "<Setting> › <Value>"
breadcrumb rows with icons and a check on the current value, for both a
running agent and a new draft tab. Rows only appear while typing, so the
default palette is unchanged.
- Thinking and Mode are multi-option selects; Plan mode and Fast mode are
On/Off toggles.
- Plan mode is dual-natured: a permission mode for Claude/Copilot/OpenCode
(excluded from the Mode list, given its own toggle) and the plan_mode
feature for Codex. Fast mode drives the fast_mode feature (model-gated).
- All logic lives in command-center/agent-setting-contributions.ts as pure,
unit-tested builders, with a small icon helper and a hook that registers
the running-agent selectors; the draft registers from its composer tab.
No changes to the command palette component and no protocol changes.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
* Unify Command Center agent controls
* refactor: clarify agent control ownership
* refactor(app): tighten command center agent controls
---------
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Co-authored-by: Mohamed Boudra <boudra.moha@gmail.com>
* feat(settings): choose which orchestration skills to install
The skills row installed the whole bundle or nothing. A settings control
next to it now opens a sheet where you keep every bundled skill or pick a
subset, and saving converges ~/.agents, ~/.claude, and ~/.codex to it.
"All skills" reads the bundle directory instead of a hardcoded catalog, so
a skill added in a later release installs itself without another code
change. Retired bundle names stay in a separate cleanup list so old copies
on disk are still removed but never become selectable.
The selection persists under Electron user data, separate from install
state, so uninstalling does not erase the choice — a later install brings
back exactly the chosen set. Installs with no saved selection read as
"all", which is what they already had.
* Restore skill selection after failed save
* fix(settings): don't save a skill selection that never reached disk
A failed save wrote the preference, converged, then wrote the old preference
back. That leaves a window where a selection the user was told did not save
is the one on disk, and a crash inside it makes the window permanent.
Converge first and persist only on success — nothing to roll back, and a
failed save now leaves disk and preference exactly as Cancel would.
Resolving the bundle path when the command handlers are wired also broke
daemon-manager's tests: the packaged branch reads process.resourcesPath,
which is undefined outside Electron. Resolve targets per command instead —
registering handlers should not touch the environment.
* fix(settings): make skills convergence single-writer and confirm removals
Startup auto-update and the settings commands were separate writers over the
same skill directories. Either could apply a plan computed from a selection the
other had already replaced — narrow the selection while startup is repairing
drift and the repair puts the removed skill back. There is now one controller
that owns scan, apply, and persist, and every entry point queues through it, so
concurrent callers see a consistent result instead of an interleaved one.
Save also had no recovery once it had converged: if the selection could not be
committed afterwards, disk was changed while the preference still said the old
thing, and the next startup silently undid the change. Save now puts the
managed directories back to what the committed selection describes before it
reports the failure, so a failed Save lands where Cancel would.
Deselecting a skill deletes its directory in all three agent homes, including
files the user put inside it. That is now confirmed like every other
destructive action, naming the skills and the directories. Adding skills or
saving an unchanged selection still saves without a prompt, and a skill the
host has not installed yet deletes nothing so it does not prompt either.
* fix(settings): confirm removal for skills installed in only some agent homes
The removal confirmation read an `add` op as "not on disk yet". An `add` op
means "missing from at least one target", so a skill sitting in ~/.claude but
not the other two looked absent, and deselecting it deleted that directory —
with anything the user had put inside it — without asking.
The status snapshot now carries the installed skill names directly, taken from
union presence across the three agent homes, and the confirmation reads that.
`ops` keeps its one meaning.
A host that predates the field cannot say what is on disk, so the app bridge
assumes the saved selection is installed. Asking about a directory that turns
out to be absent is harmless; the reverse is not.
* fix(settings): warn about every skill directory a save deletes
The confirmation asked "was this selected before and is it selected now",
which is a different question from the one the host answers. The host
converges to the desired set and deletes every managed directory outside it,
so a skill on disk that the saved preference never mentioned — reinstalled by
hand, restored by a sync, left behind by drift — was deleted with no warning,
user files included.
The predicate now mirrors the host rule: everything installed that the draft
will not keep, regardless of what was saved before. It moved out of the sheet
into its own module so the rule has one owner and can be tested directly.
Saving with nothing to delete still does not prompt.
* fix(settings): a failed save keeps files the user added inside a skill
Save converges before it commits, and convergence deletes whole skill
directories. When the commit then failed, the recovery re-installed the
committed selection — which rebuilds the bundled files and nothing else, so
notes, hooks, and anything else the user had put inside a deselected skill
were gone for good. Cancel does not do that, and a failed save is supposed to
land where Cancel would.
Save now runs as a transaction. Before converging it stages a copy of every
managed skill directory across the three agent homes, next to the tree so the
copy stays on one filesystem. Committing drops the staging; failing anywhere
puts each directory back exactly as it was, deletes the ones the candidate had
created, and then drops the staging. Staging left by a process that died
mid-save is cleaned up by the next one.
The controller only sees begin/commit/rollback; the filesystem work lives in
skills-transaction.ts.
* fix(settings): make skill removal confirmation authoritative
* fix(settings): verify skill convergence before commit
* fix(settings): preserve concurrent skill edits on rollback
* fix(settings): preserve staged skill metadata
* fix(settings): stage skill deletions atomically
* fix(settings): stage removals per filesystem
* fix(settings): restore staged skill symlinks
* fix(settings): canonicalize skill transaction roots
* fix(settings): keep automatic skill updates non-destructive
* fix(settings): report committed skill saves accurately
* fix(settings): keep skill updates non-destructive
* fix(settings): route skill removals through selection
* fix(settings): close skill selection edge cases
* fix(settings): preserve skill data during maintenance
* fix(settings): hide unavailable skill install
* fix(desktop): keep install policy at command boundary
* fix(desktop): restore staged skill directories atomically
* test(app): retry transient skills sandbox cleanup
* fix(desktop): recover externally deleted skill stages
* fix(desktop): preserve conflicting skill rollback data
* fix(desktop): generalize skill rollback conflicts
* fix(desktop): preserve skill rollback collisions
* fix(desktop): recover quarantined skill backups
* fix(app): inset adaptive sheet content on compact
The compact Choose skills sheet rendered edge to edge: the All skills card
touched both screen edges, the section label was glued to it, and the skill
list ran to the edges too. Desktop was correct.
AdaptiveModalSheet expressed its content inset through
BottomSheetScrollView.contentContainerStyle. That is a library prop, not the
style prop Unistyles registers, and gorhom's scroller was never rewritten by
the Babel plugin, so on web the emitted class was never applied and the
padding and gap resolved to nothing. The desktop card uses a real ScrollView,
which Unistyles does track, which is why only compact was affected.
The inset now lives on a View the sheet owns, rendered by every presentation,
so one definition covers compact and desktop and a scroller only ever gets
theme-free layout. The caller-facing prop is contentStyle, since it no longer
lands on a scroll content container.
Choose skills drops its ad-hoc label for SettingsSection, so the label carries
the app's section rhythm instead of sitting equidistant between two cards.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
* fix(desktop): prevent skill rollback data loss
Preserve externally replaced directory trees during recovery and reject managed child symlinks before synchronization can write outside a skill directory.
---------
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
* Add npm release track invariant and beta dist-tag procedure
* Revert "Keep older chat history and image previews stable (#2490)"
The timeline optimistic/pagination rework needs more work than main releases can wait for. Reverted together with #2484; both are reapplied on integration/timeline-optimistic-rework.
* Revert "Stop completed turns from appearing stuck (#2484)"
See the #2490 revert. #2490 built on this submission-authority change, so the two revert and reapply as a pair.
* fix(app): stop completed turns from appearing stuck
Turn activity was inferred from a user-message row flag, so a stale duplicate row could keep the working footer active after the turn ended. Track submission lifecycle separately and route every send path through the shared submission flow.
* fix(app): guard pending message identity
* test(app): measure submission layout independent of scroll
* fix(app): keep submitted messages consistent through reconnects
Track each in-flight send until its own RPC establishes acceptance, and keep canonical timeline placement authoritative. Restore legacy cached rewind IDs during cache deserialization.
* fix(app): close canonical submission races
* test(app): cover canonical submission races in browser
* fix(app): keep replacement submissions authoritative
Keep current pending rows out of legacy cache migration and leave ambiguous terminal lifecycle events to the daemon snapshot. This prevents fabricated rewind identities and stale completion events from marking replacement turns idle.
* fix(app): keep submitted messages stable across sync
Give submission transport and canonical timeline ingestion separate authority. Preserve every unresolved local send across replacement, reconcile provider identity once, and bridge RPC acceptance to authoritative running state without deriving lifecycle from timeline rows.
* fix(app): settle submissions in either acknowledgement order
Complete submission transactions when RPC and provider acknowledgement arrive in either order. Cache only transaction-owned local rows as transient data, preserve canonical ID-less prompts, and invalidate ambiguous legacy display caches instead of inventing provider identity.
* fix(app): keep agent visible during history handoff
Running and terminal updates could clear create continuity before the initial authoritative timeline arrived, leaving a streaming agent behind the loading screen. End the handoff only when authoritative history is applied.
* fix(app): settle attachment-only submissions
Canonical providers can acknowledge image-only prompts with empty text. Reconcile those events by client identity without rendering a blank canonical row.
* fix(agent): settle out-of-band message submissions
Accepted commands that do not allocate a foreground turn previously had no canonical user acknowledgement. Record the command before its handler runs so submission state and reconnect history converge through the normal timeline producer.
* fix(app): settle out-of-band submissions compatibly
* fix(app): preserve canonical prompt order
* test(app): keep workspace status check timing-independent
The workspace-status scenario asserted footer settlement before initial agent creation had necessarily completed. The dedicated draft-handoff coverage owns that lifecycle contract.
* fix(chat): load older timeline history consistently
Pagination depended on scroll events, so short or compacted initial pages could never request older history. Reevaluate edge visibility as layout and history change, and standardize loading indicators on the canonical app spinner.
* fix(chat): keep older history loading reliably
Use authoritative timeline cursor progress so merged rows cannot stall pagination. Wait for bottom anchoring before evaluating the history edge, and re-arm retries on a fresh upward gesture without automatic failure loops.
* fix(app): preserve image attachment MIME types
Image intake discarded clipboard and desktop path metadata, then guessed JPEG when a later file object had no type. Resolve MIME once at the platform boundary and require it downstream.
* fix(app): honor native picker MIME metadata
Render optimistic turn feedback before host acknowledgement and roll it back on rejection. Reconcile iOS keyboard offsets from the native transition end event so JS contention cannot leave the composer displaced.
* fix(app): refine compact composer controls and native scrolling
Consolidate responsive agent controls and model browsing inside the compact sheet. Preserve manual native scrolling by suspending sticky bottom maintenance while the user owns the viewport.
* fix(app): suppress recoverable resume refresh errors
Resume revalidation can race host reconnection and reject while the host is offline. Treat the refresh as deferred so development LogBox does not surface a recoverable disconnect.
* fix(app): stabilize compact model sheet loading
Let the model list consume the sheet space above persistent controls as the sheet expands. Keep unresolved defaults in a loading state so Select model only represents a genuinely empty selection.
* fix(app): preserve route anchors during native scroll
* fix(app): preserve native scroll intent and sheet dismissal
* fix(app): keep compact sheet styles composable
Open the complete working comparison in a desktop workspace tab while preserving inline sidebar diffs when the tab is closed.
Working and commit comparisons share one diff panel, and workspace layout is the sole owner of live tab persistence and identity.
Refs #1520
Mohamed Boudra