- Add ModelOnboardingModal component with provider selection, API key input, and model verification flow
- Add API key authentication support to SettingsModal with clear-key functionality
- Trigger onboarding from App.tsx when modelOnboardingComplete setting is false
- Add comprehensive tests for ModelOnboardingModal and SettingsModal authentication
- Document model onboarding flow and API key auth in AGENTS.md and README
- Add CSS styles for onboarding modal, provider cards, and auth section
- Add generateSubtasksFromPlanning function to generate subtasks from planning session key deliverables
- Add POST /planning/start-breakdown and POST /planning/create-tasks API routes
- Add startPlanningBreakdown and createTasksFromPlanning frontend API functions
- Extend PlanningModeModal with Break into Tasks UI and subtask editing (drag-and-drop, dependency validation)
- Add comprehensive tests for generateSubtasksFromPlanning with edge cases
- Document planning mode multi-task creation feature in AGENTS.md
The guard `if (!window.resetText)` prevented overwriting the raw CLI
text (e.g. "Resets Apr 9 at 8pm") with a relative duration because
resetText was already populated from the parsed CLI line. Remove the
guard so the parsed date is always converted to "resets in 3d 8h".
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add activeTab and connectionStatus to the subscription effect dependency array in TerminalModal
- Ensures WebSocket reconnects properly when terminal tab is reactivated or connection status changes
Vite's dev server was returning HTML error pages instead of JS for
dynamically imported @xterm/* modules, causing 'text/html is not a
valid JavaScript MIME type' failures during terminal initialization.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add @mariozechner/pi-ai mock to restart.integration.test.ts and
project-runtime.test.ts to fix module linking errors. Fix call count
assertions in executor and restart tests to account for the
retry-without-task_done behavior.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Introduces ai_sessions table (schema v9), AiSessionStore, and background
session support for mission interviews, planning, and subtask breakdown.
Adds BackgroundTasksIndicator component, SSE-based progress streaming,
and dashboard API routes for session management.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add MemoryInsightsEngine class for extracting insights from task memory
- Implement pattern, metric, trend, anomaly, recommendation, and summary extractors
- Add insight extraction settings fields (enabled, maxInsights, types, schedule) to ProjectSettings
- Export MemoryInsightsEngine and related types from @fusion/core
- Add comprehensive test suite covering all extractor types and edge cases (538 lines)
- Fix backend regex in usage.ts to correctly parse CLI date-based reset text format
- Add frontend fallback in UsageIndicator to display relative reset time when absolute date is unavailable
- Add tests for backend regex parsing (usage.test.ts)
- Add tests for frontend UsageIndicator component (UsageIndicator.test.tsx)
- AgentsView: add missing fetchModels/fetchAgentStats mocks, update tests
for new multi-step NewAgentDialog and ActiveAgentsPanel dual rendering
- routes-diff: rewrite tests to match new resolveDiffBase() two-dot syntax
- routes.test: fix spec/revise test — done→triage is valid per VALID_TRANSITIONS
- activity-feed CSS: replace hardcoded rgba(88,166,255) with color-mix token,
fix var(--error) → var(--color-error)
- server/webhook/websocket tests: add getDatabase() mock for AiSessionStore
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add a conversational AI interview flow to the missions screen that
interviews users to break down missions into milestones, slices, and
features with verification criteria at every level.
Backend: Replace stubbed interview logic with real AI agent integration
using createKbAgent, SSE streaming, and robust JSON parsing. Wire up
all 5 placeholder mission interview route endpoints (start, respond,
cancel, stream, create-mission).
Frontend: New MissionInterviewModal component with conversational UI,
thinking output display, question renderers, and hierarchical plan
review/edit. Integrated into MissionManager with "Plan with AI" button.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add getFusionDir() method to TaskStore for resolving agent database path
- Update in-process-runtime to use getFusionDir() instead of hardcoded path
- Update dashboard routes AgentStore initialization to use getFusionDir()
- Add changeset for published package patch bump
When the stuck task detector killed a task and immediately called
moveTask("todo"), the scheduler could re-dispatch the task before the
old execution's finally block cleared this.executing. The new execute()
call hit the guard and silently returned, stranding the task in
"in-progress" with no active session or worktree (seen on FN-810/FN-912).
Move the requeue responsibility from StuckTaskDetector.killAndRetry to
the executor's finally block, which runs after this.executing.delete().
The beforeRequeue budget check now runs before session.dispose() and its
result is passed via StuckTaskEvent.shouldRequeue → markStuckAborted().
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Move initialBufferRef reset from the beginning of connect() to inside
ws.onopen to ensure the buffer is fresh only when the new connection is
established. This prevents a race condition where the buffer could be
cleared before the WebSocket is ready, causing scrollback data to be
lost when terminal tabs are switched.
- Close the refine model dropdown as soon as a selection is made instead of waiting for the next render cycle
- Add test verifying the dropdown closes immediately upon refine selection in QuickEntryBox
- Replace plain text input for model selection with CustomModelDropdown component in NewAgentDialog
- Add provider/model selection with grouped dropdown UI and search filtering
- Support editing existing agents with pre-populated model values
- Add comprehensive test suite (374 lines) covering rendering, interaction, and edge cases
- Wire up agent creation and editing flows to use the new dropdown for model selection
- Unify stuck-task derivation using project settings (taskStuckTimeoutMs) instead of hardcoded values
- Thread taskStuckTimeoutMs through Board, Column, WorktreeGroup, and TaskCard components
- Add visual stuck indicator with pulsing animation on in-progress task cards
- Resolve merge conflicts with isSearchActive prop from main (both props coexist)
- Add taskStuck utility unit tests and update README with stuck indicator docs
- Update dashboard dev script to run vite build, tsc, and app typecheck before starting vite dev
- Ensures the project is fully built and type-safe before development server starts
- Fix resolveBaseBranch to use stored branch name and consistent fusion/ prefix
for both explicit deps and blockedBy paths (was using kb/ for blockedBy)
- Add main branch checkout verification in merger before squash merge to prevent
feature code from landing on wrong branch lineage
- Align all branch prefix references from stale kb/ to fusion/ across executor,
merger, store, and routes
- Fix executor test OOM by mocking merger fully, adding fake timers to retry
tests, and switching vitest pool to vmThreads
- Update all test assertions to use fusion/ branch prefix
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Rewrite TaskDetailModal tests to reflect the unified Logs tab (agent log + activity log merged)
- Update test assertions to match new tab structure and component behavior
- Update README to describe the unified Logs tab layout
- Add Usage Dialog section to README dashboard features documentation
- Enhance reset time visibility in the UsageIndicator component
- Add comprehensive tests for Claude 5h usage scenarios (155 lines)
- Update dashboard styles for usage indicator improvements
- Fix branch name assertion in executor test: use kb/ prefix instead of fusion/
- Add comprehensive tests for stale worktree and branch reference recovery (267 lines)
- Document stale worktree/branch recovery procedure in README
Instead of immediately failing when an agent finishes without calling
task_done, spawn a fresh session with a recovery prompt that asks the
agent to review the worktree state and complete the task. Only fail
if the retry also doesn't produce a task_done call.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Document nested model menu behavior in QuickEntryBox in AGENTS.md
- Add README section describing the model selection nested menu pattern
- Add regression tests for nested model menu submenu open/close behavior
- Server /tasks/:id/diff and /tasks/:id/file-diffs now both use
resolveDiffBase() for consistent file lists across card and modal
- Both endpoints handle done tasks via mergeDetails.commitSha,
returning structured { files, stats } from the merge commit
- TaskChangesTab uses fetchTaskDiff for all columns (no more
fetchCommitDiff/parsePatch client-side path for done tasks)
- useChangedFiles hook simplified — server handles done tasks
- Done task cards show "N files changed" button using mergeDetails
- Updated tests for new unified data flow
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Add Planning/Triage model resolution and display in AgentLogViewer header
- Wire planning model resolution into TaskDetailModal with model info tab
- Add comprehensive tests for AgentLogViewer and TaskDetailModal model display
- Document planning/triage model display feature in README
- Port model menu dropdown to a portal for proper z-index layering above overlays
- Stabilize filter input behavior inside the portaled menu
- Add CSS styles for portal-based menu positioning and layering
- Add comprehensive tests for the portaled QuickEntryBox model menu
- Add changeset for the published package fix
The merger was using `git diff HEAD..branch --stat` which includes
artifacts from other tasks when branches fork from older main commits.
Switch to `git diff $(merge-base)..branch --stat` so commit messages
only describe the branch's own changes.
Also surface the "files changed" button on done task cards using
mergeDetails, opening the same ChangedFilesModal with commit-backed
diffs (matching the Changes tab in the task modal).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Add CSS styles for changed-files list items with button reset, layout, hover/focus states, and theme-aware colors
- Add regression tests for ChangedFilesModal structure and styling (file type indicators, click-to-open, checkbox behavior)
- Update dashboard README with notes on changed-files list styling conventions
- Replace var(--error) with var(--color-error) in ActivityFeed TYPE_CONFIG for consistency with theme system
- Add unit tests for theme-aware color variables in TYPE_CONFIG
- Add integration test verifying ActivityFeed renders error colors correctly
- Document theme-aware styling note in dashboard README Activity Log section
- Add theme-aware CSS classes for changed-file status badges (added, modified, deleted, renamed)
- Update ChangedFilesModal, CommitDiffTab, and TaskChangesTab to use theme-aware status classes
- Add comprehensive tests for ChangedFilesModal and TaskChangesTab status styling
- Refactor stuck-task-detector to extract StuckTaskConfig and simplify detection logic
- Clean up 370+ lines of redundant test code in stuck-task-detector tests
- Update dashboard README with theme-aware styling notes
- Pass throughChangedFiles option in executor for better diff handling
- Add explicit no-progress loop detection signals to executor agent sessions
- Enhance stuck-task detector with configurable strategies (timeout, no-progress, combined)
- Add comprehensive test coverage for stuck-task detector (371 lines of tests)
- Remove modelFilter test utilities (198 lines of dead test code)
- Simplify modelFilter.ts by removing unused filtering logic
- Clean up dashboard server startup and minor README fix
- Rewrite filterModels utility with deterministic fuzzy matching (no probabilistic ranking)
- Extract model filtering logic from TaskDetailModal into dedicated modelFilter.ts utility
- Add comprehensive regression test suite for fuzzy matching edge cases
- Remove inline model filter tests from TaskDetailModal test file
- Clean up unused CSS styles related to old model filter UI
- Document fuzzy model search behavior in dashboard README
- Show Workflow tab for tasks with enabled workflow steps OR previous workflow results
- Add comprehensive tests for TaskDetailModal workflow tab visibility
- Add workflow tab styles including step result cards and status badges
- Clean up TerminalModal and useTerminalSessions (remove unused code and tests)
- Remove dashboard README section on terminal sessions
- Add bootstrap error state and retry logic to useTerminalSessions hook to prevent indefinite 'Starting terminal' hang
- Add error/retry UI states in TerminalModal component with CSS styling
- Add tests for useTerminalSessions bootstrap error and retry behavior
- Refactor WorkflowStepManager with template support and improved UX
- Extract CSS from inline styles to stylesheet for terminal and workflow components
- Document terminal startup error handling and non-hanging guarantee in dashboard README
- Introduce baseCommitSha on sessions for task-scoped file diffs independent of shared worktrees
- Add /api/sessions/:id/files and /api/sessions/:id/file-diffs routes using baseCommitSha
- Update TerminalModal to consume task-scoped file list and diff endpoints
- Add regression tests for shared worktree scenario (multiple tasks in same worktree)
- Expand route tests for file diffs and session files with edge cases
- Clean up stale TerminalModal test helpers and simplify diff display styles
- Update README and route documentation for task-scoped diffing
- Expose delete action on idle and terminated agents in AgentsView and AgentListModal
- Update AgentDetailView to support delete with confirmation
- Update agent-related documentation in README.md and dashboard README
- Add/update tests for AgentDetailView, AgentListModal, and AgentsView delete behavior
- Remove outdated TaskDetailModal test
- Fix comment display state not propagating correctly in TaskDetailModal
- Remove unused AgentListModal and AgentsView components and their tests
- Simplify GitManagerModal by removing unused code paths
- Remove dead CSS styles from dashboard stylesheet
- Update README docs to clarify comment display behavior
- Add tests for TaskDetailModal comment state handling
