- Add ChatManager class for managing chat sessions with AI agent integration
- Implement SSE streaming for real-time chat responses
- Add chat API routes for creating sessions, sending messages, and streaming responses
- Wire ChatStore and ChatManager into server initialization
- Add comprehensive test suite for all chat routes
- Add Budget Settings section to ConfigTab with token limit, usage threshold, budget period, and reset day inputs
- Add budget usage badge to DashboardTab showing token consumption and threshold warning
- Add budget API functions to api.ts (getBudgetSummary, updateBudgetConfig, deleteBudgetUsage)
- Add budget routes to routes.ts for GET/PATCH/DELETE agent budget endpoints
- Add CSS styles for budget progress bar, warning indicators, and section layout
- Add comprehensive tests for budget UI components and API routes
- Extract all color-theme CSS variable blocks from styles.css into a dedicated theme-data.css file
- Add dynamic loading of theme-data.css for non-default color themes via useTheme hook
- Preload theme-data.css for the default theme in index.html to avoid flash
- Update status-colors-theme tests for dynamic theme loading behavior
- Update memory and README with theme-data.css file location
- Add NodeProvider and useNodeContext to App shell for centralized node state
- Add node selector dropdown and status indicator to Header component
- Wire remote node data and event hooks into task-facing views
- Update App.test.tsx mocks for new node context providers
- Add project memory learnings for node context wiring patterns
- Update dashboard guide documentation for node selection
- Wrap AppInner with NodeProvider in App.tsx
- Read node context state (currentNode, currentNodeId, isRemote, setters)
- Add useRemoteNodeData and useRemoteNodeEvents hooks
- Sync selected node with useNodes() results
- Update Board/ListView/ExecutorStatusBar to use remote tasks when in remote mode
- Add mocks for NodeContext, useRemoteNodeData, useRemoteNodeEvents, useNodes in App.test.tsx
- Clear fusion-dashboard-current-node in test setup to avoid cross-test leakage
- Add search toggle button to header controls for non-mobile viewports
- Toggle persists search visibility state across board and list views
- Implement responsive visibility rules (show on non-mobile, hide on mobile)
- Add comprehensive tests for tablet header controls and Header component
- Update Header component with search toggle functionality
- Replace font-size/line-height with inline-flex layout for SVG compatibility
- Add explicit color: var(--text-muted) for consistent theming across light/dark modes
- Improve alignment with vertical-align: middle for scope indicator icons
- Create agentHealth.tsx utility with consistent health status logic
- Update AgentDetailView, AgentListModal, and AgentsView to use centralized utility
- Add comprehensive tests for health status edge cases
- Update agents.md documentation with health status reference
- Add changeset for @gsxdsm/fusion patch release
- Add comprehensive tests for Kimi API 404 retry behavior
- Expand retry logic to handle any 404 response from Kimi API
- Add test cases for various 404 scenarios including context errors and quota exhaustion
- Improve error handling resilience for Kimi model calls
- Reduce MobileNavBar height from 56px to 48px for a more compact mobile experience
- Update footer safe area padding to accommodate the new nav height
- Update board mobile layout tests to reflect the reduced height
- Update mobile nav bar CSS tests for the new height value
- Reorder Kimi API endpoint list to prioritize the underscore-variant endpoint
- Add tests to verify endpoint priority selection
- Ensure Codexbar-validated endpoint is tried first in the fallback chain
- Add explicit boundary comments to serve.ts and dashboard.ts clarifying semaphore lane usage
- Add regression tests for semaphore lane-vs-utility boundary in serve.test.ts and dashboard.test.ts
- Add changeset for @gsxdsm/fusion patch release
- Add ResponsiveLayout component with dynamic row/column splitting based on terminal size
- Create terminal utility for detecting terminal dimensions and layout preferences
- Add truncate utility for smart text truncation with min/max bounds
- Implement comprehensive responsive layout tests with all breakpoints
- Update TUI exports to expose new layout and utility components
- Enhance README documentation with responsive layout guide and examples
- Update ProviderIcon with Kimi's official crescent moon brand mark
- Add comprehensive tests for ProviderIcon component with Kimi logo variants
- Include changeset for @gsxdsm/fusion package
- Add agent picker dropdown to New Task modal for task assignment at creation time
- Include assignedAgentId in task creation payload when an agent is selected
- Add comprehensive regression tests for agent assignment functionality
- Update AGENTS.md documentation with New Task modal agent assignment details
- Create useGlobalShortcuts hook for centralized keyboard shortcut handling at app root
- Add FocusGuardRef module-level ref for tracking text input focus state
- Implement HelpOverlay component displaying available keyboard shortcuts
- Add Ctrl+C (emergency exit), q (quit), ?/h (help toggle), 1-5 (screen switch) shortcuts
- Focus guard prevents shortcuts when text input is focused (except Ctrl+C)
- Update ScreenRouter with controlled/uncontrolled mode support
- Wire global shortcuts into demo app with help overlay integration
- Add comprehensive tests for useGlobalShortcuts and HelpOverlay
- Update README with global shortcuts documentation and usage examples
- Add regression tests for SubtaskBreakdownModal
- Test send-to-background behavior preserves session state
- Test cancel behavior properly closes modal and cleans up
- Verify correct callback invocation for each action
- Update ProviderIcon.tsx to use Moonshot AI logo instead of Kimi branding
- Fix logo display in UsageIndicator component
- Add ProviderIcon unit tests covering all provider icons
- Add UsageIndicator tests for provider icon rendering
- Add CustomModelDropdown integration tests for model selection flow
- Add assigneeUserId field to Task and TaskDetail types
- Add reviewHandoffPolicy project setting (manual/auto/delegated modes)
- Implement handoff detection policy in executor with session state tracking
- Add dashboard API routes for user assignment and handoff endpoints
- Add frontend API functions for review handoff UI integration
- Add tests for review handoff feature endpoints
- Fix test duplicates and finalize documentation
- Add assigneeUserId field to Task type and SQLite schema for human assignment
- Add reviewHandoffPolicy setting to control automatic handoff behavior
- Implement handoff detection in executor: detect user assignment during review and auto-transition task
- Add dashboard API routes for user assignment, handoff queries, and completion
- Add frontend API functions: getHandoffTask, assignTaskToUser, completeHandoff
- Add comprehensive tests for store methods, API routes, and executor handoff logic
- Update memory documentation with review handoff pattern
- Add ScreenRouter component for multi-screen navigation in TUI
- Implement tab bar with icon labels and active state indicators
- Support keyboard navigation with arrow keys and Tab/Shift+Tab
- Add mouse click support for tab selection
- Export new screen router components from package index
- Add comprehensive unit tests for navigation behavior
- Update documentation with architecture and gap analysis
- Update TUI README with ScreenRouter usage examples
When a stuck task's session is disposed but an in-flight bash subprocess
keeps running, the executor's Promise never resolves, so the requeue
finally block never fires and the task stays stranded in-progress
indefinitely (observed with FN-1269).
Add a 60s safety-net timer in markStuckAborted: if the task is still in
executing after the grace period, force-requeue it directly by clearing
the worktree, setting status=stuck-killed, moving to todo, and removing
it from the executing set so the scheduler can re-dispatch normally.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Add abandon action to useBackgroundSessions hook for deterministic session cleanup
- Make background dismissal deterministic instead of relying on 7-day TTL expiration
- Update PlanningModeModal to include abandon button in header actions
- Add comprehensive tests for abandon behavior and session lifecycle
- Update dashboard guide with session lifecycle documentation
- Add explicit always-green test suite instructions to executor agent prompts
- Update executor to enforce test-suite validation before task completion
- Add tests for agent prompt generation and executor behavior
- Ensure test failures block merge-ready state rather than allowing broken builds
- Add Authentication section to Settings sidebar with globe icon
- Update SettingsModal to display Authentication link in sidebar
- Add corresponding tests for Authentication section visibility
- Update mobile settings tests to reflect new sidebar structure
- Update AGENTS.md with authoritative run lifecycle semantics documentation
- Add API-level regression test for repeated manual run prevention
- Update run lifecycle tests to match corrected behavior
- Refactor AgentStore to use structured run records as the authoritative source for run state
- Improve run status queries with better filtering and ordering
store.logEntry, assignTask, and checkoutTask now receive an extra
runContext argument. Updated test assertions in executor.test.ts,
agent-heartbeat.test.ts, and restart.integration.test.ts to match
the current call signatures using expect.objectContaining where
appropriate.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
In step-session mode each step is registered with StuckTaskDetector
under a compound key (e.g. "FN-1452-step-1") rather than the bare
task ID. When stuck detection fired, event.taskId was that compound
key, breaking three things simultaneously:
1. beforeRequeue("FN-1452-step-1") → store.getTask() threw (no such
task) → stuckKillCount never incremented on the real task.
2. markStuckAborted("FN-1452-step-1") → activeStepExecutors.get()
returned undefined (keyed by "FN-1452") → terminateAllSessions()
never called, other step sessions kept running.
3. stuckAborted.set("FN-1452-step-1") → executor checks
stuckAborted.has("FN-1452") → miss → stuckRequeue never set
→ task never moved to todo.
Fix: add a canonicalTaskId field to TrackedTask (defaults to the
tracking key for single-session mode where they are identical).
StepSessionExecutor now passes taskDetail.id as the third arg to
trackTask(). killAndRetry() uses entry.canonicalTaskId for all
external callbacks so they always resolve to the real task ID.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
