touch-action: manipulation is needed on the textarea so iOS registers
first-tap focus reliably, but it also allows pan-y — which let the
user drag the input box up off-screen with the keyboard up.
Cancelling touchmove blocks the drag without affecting tap (a tap
fires touchstart + touchend with no touchmove in between), so the
composer stays locked AND first-tap focus works.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds unassigned branch filter options to the dashboard header, allowing users to filter tasks by unassigned status. Updates the Header component with new filter controls and includes corresponding tests for the App and Header components.
Fusion-Task-Id: FN-3424
Two fixes for residual mobile chat issues:
1) Textarea touch-action: none was preventing iOS from registering a
clean tap-to-focus, causing the keyboard to flash up via the
programmatic focus() in onTouchStart and then auto-dismiss because
iOS never saw the gesture complete. Switching to manipulation
allows tap while still blocking pan/zoom — the composer stays
anchored thanks to overscroll-behavior: contain on its container.
2) On switch-away-and-back the visualViewport metrics could get
stuck in a half-state (composer pushed up, or blank pane covering
it). Adding a visibilitychange / pageshow handler on ChatView that
force-blurs and re-focuses the active textarea makes iOS resync
the keyboard / vv metrics cleanly.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
When ChatView remounted (e.g. tab switch with keyboard still up), the
hook started with keyboardOpen=false and corrected itself only after
the effect ran. That single stale-state render briefly unhid the
executor status bar, which appeared as a blank pane covering half the
input box before the next state update settled it.
useState initializers now call getKeyboardMetrics() lazily on first
render so the very first paint already reflects the live keyboard
state.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Tab/page switches with the keyboard up sometimes still left the
chat-thread anchored wrong because the timed re-reads (50/200/500ms)
missed the actual settlement frame on slower devices.
updateWithTail now also kicks off a rAF poll that re-reads metrics
each frame until offsetTop is stable for 2 consecutive frames or 1.5s
elapses. Combined with the extended timed tail (now adds 1000/1500ms),
this should catch both fast and slow iOS settlement paths without
indefinite polling.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Completes Step 3 of FN-3618 by surfacing AI provider icons throughout the Settings modal and model dropdowns, updating both the component styles and adding integration and unit tests to cover the new icon display behavior.
Fusion-Task-Id: FN-3618
Switching away from the chat view and back left the hook with stale
metrics — keyboardOpen was false at remount even though the system
keyboard was still visible, so the .chat-thread transform never
re-applied and the message list anchored too high. focusin and
synchronous reads also captured offsetTop mid-transition.
Adds a tail of delayed re-reads (50ms, 200ms, 500ms) on focusin,
visibilitychange, and pageshow so iOS has time to settle before the
final snapshot is taken.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Implements a distributed task-ID allocator with mesh networking endpoints for multi-project coordination (FN-3449), alongside substantial ChatView and GitManager mobile polish including improved iOS keyboard handling, rubber-band containment, scroll-drift fixes, and visual viewport stabilization. In
Fusion-Task-Id: FN-3449
A pan-y gesture on the focused textarea still triggered iOS auto-
scroll-into-view, lifting the composer up off-screen even with body
locked. touch-action: none on .chat-input-textarea blocks gesture-
driven panning entirely (tap-to-focus, typing, caret moves still
work) so the composer stays anchored. Trade-off: loses gesture-
scroll within the textarea once content exceeds max-height — minor
since textareas rarely exceed 6 lines in this composer.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
iOS fires visualViewport scroll events at 60fps during a pan with the
keyboard up. Routing those through React state and into the .chat-thread
translateY(--vv-offset-top) transform amplified the pan into a visible
judder + ~300px shift + body background exposure.
useMobileKeyboard now uses two listeners: a full update (resize +
focusin/focusout) that re-snapshots all metrics including offsetTop, and
a scroll-only update that updates only height/keyboardOpen. offsetTop
is therefore frozen between keyboard open/close events — the transform
correctly compensates for iOS's initial visualViewport shift on focus
without following pan-time movement.
Restores the translateY anchor (so the thread isn't off-screen on
first focus) while keeping the swipe-jitter fix.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The transform amplified jitter during iOS textarea swipes: visualViewport
scroll events updated --vv-offset-top rapidly, the transform followed,
and the thread juddered then shifted ~300px exposing the body background.
Removing the transform leaves the thread where iOS places it; the
height-shrink to var(--vv-height) still constrains it to the visible
area. Trade-off: on initial focus the thread anchor relies on iOS's
own viewport adjustment instead of explicit compensation, but that
beats the swipe-time jitter.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Three scoped overscroll fixes for the ChatView mobile keyboard-up case:
- .chat-messages: overscroll-behavior: contain so edge-swipes on the
messages list don't bounce the parent / document.
- .chat-input-area: overscroll-behavior: contain so swipes on the
composer footer don't escape upward.
- .chat-input-textarea: touch-action: pan-y + overscroll-behavior:
contain so a vertical swipe started on the textarea stays inside
the textarea (multi-line scroll still works) instead of
rubber-banding the page and exposing the body background.
User-reported symptom: keyboard up + swipe up on the textarea edge,
the composer shakes and a gray box (body background) partially
covers it.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This merge delivers five major themes: a large mobile chat UX pass (viewport-aware GitManager CSS, compact tool-call layout, keyboard-safe modals, and iterative fixes for the send-button double-fire behavior), a complete refactor of the TaskDetailModal test suite from a 6745-line monolith into eight
Fusion-Task-Id: FN-3614
After dismissing and re-bringing up the mobile keyboard, iOS could
leave window.scrollY > 0 and visualViewport.offsetTop > 0. With
useMobileScrollLock then pinning body{position:fixed} relative to that
drifted scroll, the message thread anchored above the visible viewport
and a large blank area appeared below it.
handleInputFocus now resets window scroll to (0,0) on mobile in a
zero-delay timeout — late enough that iOS finishes its own
scroll-into-view first, but before useMobileScrollLock observes the
drifted state.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Two mobile chat regressions addressed together:
1) ChatView send button: preventDefault now fires on pointerdown for
touch pointers, before iOS blurs the textarea. The previous
onMouseDown.preventDefault was too late on iOS (mousedown is
synthesized after touchend, by which point the keyboard has
already started dismissing). Click still runs the action so quick
taps remain reliable.
2) ExecutorStatusBar: hidden on mobile while keyboard is open,
mirroring MobileNavBar. The bar is position:fixed against the
layout viewport, which iOS leaves anchored below the keyboard;
during a swipe/pan it would slide over the messages list.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The imperative useLayoutEffect approach made mobile worse — first tap
flickered and didn't bring up the keyboard, while the original
swipe-overlap symptom remained. Restoring the previous React-state
flow until a better fix is identified. Removes the changeset that
shipped with the failed attempt.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The mobile composer/footer slid over the message list when the user
swiped with the keyboard up. Cause: --vv-height / --vv-offset-top were
routed through React state via useMobileKeyboard, so on iOS — which
fires visualViewport scroll/resize on the same frame as its keyboard
animation — the .chat-thread translation lagged by one paint, visible
as the composer momentarily floating over messages.
Now those two vars are written imperatively in a useLayoutEffect
directly to the .chat-thread DOM node on every visualViewport event,
mirroring the working pattern at QuickChatFAB.tsx:1032-1052 (which
already works correctly on mobile). Only --keyboard-overlap (a
structural open/close signal, not per-frame) still flows through
React state.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Re-add preventDefault on mousedown so the textarea doesn't blur when
the user taps send — keyboard stays up, no viewport reflow jumping
the input to the top of the screen. The action still runs on click
(which fires reliably from the iOS touch sequence even for quick
taps), so this preserves the previous fix's quick-tap reliability.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The mobile send button used pointerdown + touchstart with preventDefault
and a focus-preservation dance to keep the keyboard up while sending.
That path silently failed on quick taps on iOS — only a long press
registered. Switching to plain onClick (with touch-action: manipulation
to skip the click delay) fires reliably on tap. The soft keyboard may
dismiss on send now, which is a minor regression vs. the previous
intent but vastly preferable to silent failure.
QuickChat is unchanged because it already works on mobile.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The previous fix introduced a regression: handledMobileSendRef /
handledMobileActionRef was set in onPointerDown/onTouchStart but only
cleared in onClick. preventDefault() on pointer/touch events suppresses
the synthesized click on iOS Safari, so the ref stayed true forever
after the first tap — every subsequent tap (quick OR long press) hit
the new dedupe guard and silently bailed.
Now both handlers schedule a 500ms setTimeout to self-clear the ref
alongside their action. That covers the full pointerdown/touchstart/
click burst from one tap while still letting the next user tap go
through.
Applied to: ChatView send button, QuickChat send button, QuickChat
stop button.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The ChatView and QuickChat send/stop buttons each had both an
onPointerDown and an onTouchStart handler invoking the action. On a
quick mobile tap both fire, so handleSend / handleSendMessage /
stopStreaming ran twice in rapid succession. The second invocation
closed the first's SSE stream (streamRef.current.close()), the server
treated that as a cancel via beginGeneration, and the chat ended with
no output — exactly matching the reported "tap silently fails, long
press works" symptom (long press happened to suppress one of the two
events).
Both handlers now early-return when the existing handledMobile*Ref
flag is already set, so only the first event for a given tap fires the
action. The send button additionally gets touch-action: manipulation
(removes the click delay that lets the textarea blur win the race) and
an expanded invisible hit area via ::before so slightly-off taps don't
land on the surrounding textarea and dismiss the keyboard without
sending.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- Add built-in plugin catalog metadata and constants for shipped installable plugins
- Render a Built-in Plugins section with install, setup-required, and manage states
- Fetch and display plugin setup status, including setup install flow and status badges
- Add PluginManager CSS and tests covering built-in plugin install/setup behaviors
Fusion-Task-Id: FN-3574
Merges the evaluator follow-up suggestion system (FN-3392, Steps 1–5), which adds a normalized contract, provenance tracking, and lifecycle documentation for AI-generated follow-up tasks, alongside chat UI improvements including unread indicators in header and mobile nav, corrected message routing,
Fusion-Task-Id: FN-3392
- Document that agent-backed chat sessions expose fn_send_message and fn_read_messages when MessageStore is available
- Clarify that model-only chat sessions continue to work without mailbox tools
- Complete the Step 4 dashboard chat documentation update
Fusion-Task-Id: FN-3608
The merge lands three commits for FN-3622's canonical task timing calculations, adding `taskTiming.ts` logic and stats timing semantics to `TaskDetailModal` and `TaskTokenStatsPanel`, backed by regression tests across those panels and the root `test-changed.mjs` script. The remaining commits introdu
Fusion-Task-Id: FN-3622
- Create a disposable HOME/USERPROFILE env for scripts/test-changed.mjs execution
- Run isolation guard checks with the same isolated env, including cache-hit no-op paths
- Clean up temp HOME after test execution to avoid residue
- Update test coverage and contributing docs for the shared isolation behavior
Fusion-Task-Id: FN-3610
Merger receives substantial hardening: autostash race-rescue with de-duplication, advisory observer for destructive operations, and subject-line preference for step headlines. The TUI gains a narrow log-split mode on System panel with proper back-navigation to the main pane. Chat header and mobile n
Fusion-Task-Id: FN-3606
The merge completes Step 2 of FN-3605 by refactoring the ChatView mobile header to preserve its visual identity while converting the toggle control to an inline layout, accompanied by corresponding tests.
Fusion-Task-Id: FN-3605
Implements a chat unread indicator across the header and mobile nav bar, driven by a new unread-response tracker in the app state, with tests and a docs update for the feature.
Fusion-Task-Id: FN-3604
Hardens async assertions in the task detail modal's pause/unpause action tests, improving test reliability for that interaction flow.
Fusion-Task-Id: FN-3600
Documentation delivery alignment completing Step 5 of FN-3369, updating the CLI reference, research docs, and research hardening preflight guide with consistent documentation delivery guidance across all three files.
Fusion-Task-Id: FN-3599
The merge introduces a major TaskDetailModal test refactor (splitting a 6.7K-line monolith into five focused suites), significant merger improvements including autostash race-rescue, deduplication, and advisory logging for destructive operations, a new TUI narrow-mode log-split feature for the dashb
Fusion-Task-Id: FN-3596
Previously ← in the narrow log strip cycled the section (matching n/p),
which was disorienting once the user had drilled into the strip — they
expected ← to back out, not flip them to a different panel. Now ← from
a focused log strip clears the narrow-split sub-focus and returns to
the main pane (System / Stats / Utilities / Settings) where the user
can drag-select tokens or interact with the panel as usual. p/P still
cycle sections in all states; ↑ at the top of the strip and Esc
continue to work as before.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Bring back the bottom log strip on the System panel and make sure
nothing that lives in System is dropped when the strip steals rows:
- Render the [Enter] / [c] / drag-to-select hint unconditionally so
shortcuts stay discoverable when the strip has sub-focus and System
isn't the keyboard target. The split-aware focus border still tracks
which pane the keyboard is wired to.
- Move URL and Token to their own full-width rows under the status-chip
wrap row. Long token values now wrap onto multiple lines instead of
being truncated or pushed off-panel — the auth token is always
visible (and selectable) in full.
- Keep mouse reporting OFF whenever System is the active section, even
when the log strip below has sub-focus, so click-drag selection of
the token text continues to work.
- Update estimateSystemContentRows to mirror the new layout: greedy
pack of small status chips, full-width URL / Token rows that wrap by
ceil(width / inner), and an always-counted hint row. The dynamic
split sizer now reserves enough rows for the panel in every state.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
System has interactive UI — Enter to open URL, [c] to copy auth token,
and click-drag for token selection — whose [Enter]/[c] hint row only
renders when the panel has focus. Auto-shifting focus into a bottom log
strip on ↓ hid that hint and removed a clear visual path back to the
panel for token operations. Restrict the auto-split to inert sections
(Stats / Utilities / Settings); System keeps the full single-pane view
it had before.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
When the dashboard TUI collapses to single-pane mode (narrow width or
short terminal), show a horizontal log strip at the bottom of the screen
while the active section renders on top. The split is content-driven:
the top pane gets exactly the rows it needs to render its section
without truncating (computed from SystemPanel's chip wrap at the current
width, or each panel's known row count for Stats/Utilities/Settings),
and the log strip absorbs every remaining row to maximize log
visibility. Down-arrow shifts sub-focus into the log strip with the
same key bindings as the dedicated Logs section (j/k, Home/G, Enter to
expand, w to wrap, c to copy, f to filter, mouse wheel). Up-arrow at
the top of the strip or Esc returns focus to the main pane.
Right/Left/Tab continue to cycle sections, so the dedicated full-screen
Logs view is unchanged. The split auto-disables if the log strip would
get fewer than 6 rows.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
P0 — parsePorcelainZ rename/copy handling
Git's -z porcelain emits `R <new>\0<old>\0` for renames (and
C for copies). The naive split-and-slice treated <old> as an
independent dirty path, which made runObservedDestructiveSyncOp
warn about phantom "cleared paths" whenever a rename was in
flight. Now we detect R/C status and skip the trailing entry.
P1 — race-rescue loop unstages between attempts
`git stash create` snapshots the index without clearing it, so
iteration 2's `git add -A` would re-stage atop iteration 1's
leftovers. Tree differences inside the loop then reflected stale
staging rather than genuine new writes. Added a `git reset` at
the top of each iteration so every attempt starts from a clean
index baseline.
P1 — writeActiveMergerStatus is now atomic
Switched from in-place writeFileSync to temp-file + renameSync.
POSIX guarantees rename atomicity on the same filesystem, so a
reader can no longer catch the file mid-flush and return a
false-negative "no merger active" advisory.
P2 — Step regex em-dash clarity
`[—\-:]` is functionally fine but obscures intent; switched to
`(?:—|-|:)` so the em-dash branch is obvious. Added a test case
for the em-dash separator.
New tests:
- parse-porcelain-z.test.ts (8 cases including renames + copies)
- em-dash case added to derive-subject-summary.test.ts
247/247 merger-suite tests pass.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The race-rescue loop was firing on every merger run because
`git add -A && git stash create` does not clean the working tree —
files stay dirty post-stash, so a subsequent `snapshotDirtyFiles` saw
the SAME paths the primary stash had just captured and stashed them
again, producing identical-tree race-rescue duplicates (visible in
git stash list as `fusion-merger-autostash:FN-XXXX:race-rescue-0`
sitting next to its identical `fusion-merger-autostash:FN-XXXX:`).
Fix: list the path set captured by the primary stash via
`git stash show --name-only`, and only rescue paths in the current
dirty snapshot that are NOT in that set — those are genuine
late-dirty writes from concurrent dev edits or interleaved ops.
Also drop any rescue whose tree-SHA exactly equals the primary,
as a defensive belt-and-braces.
Existing duplicate race-rescue stashes are harmless (identical
content to their primaries) and can be dropped manually.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- Document how branch field data flows through task lifecycle
- Clarify architecture-level plumbing responsibilities for branch metadata
- Capture Step 4 documentation updates in docs/architecture.md
Fusion-Task-Id: FN-3421
The merge brings TaskDetailModal test coverage in line with the project's testing standards by splitting the monolithic 6,745-line test file into five focused suites. It also delivers merger reliability improvements including headline-step subject preference, redundant-merge verification skipping, a
Fusion-Task-Id: FN-3594
- writeActiveMergerStatus: writes .git/.fusion-merger-active.json
(taskId, pid, hostname, startedAt) at merge entry, deleted in finally.
Not a lock — purely informational so dashboards / status lines /
pre-Edit hooks can warn devs that rootDir is volatile during the run.
readActiveMergerStatus(rootDir) is exported for consumers.
- runObservedDestructiveSyncOp: snapshot-before/after wrapper around
destructive rootDir ops that are *supposed* to preserve unrelated
working-tree edits. resetMergeWithWarn now uses it — any future
silent wipe of dirty paths surfaces as an actionable warning instead
of going unnoticed. Not applied to the autostash's own reset
--hard / clean -fd; those are intentionally destructive and already
protected by the race-rescue stash.
- Race-rescue stashes from stashUnrelatedRootDirChanges are now
attached to the AutostashHandle and surfaced via store.logEntry so
the recovery command lands on the task feed instead of only
mergerLog.warn.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
