- Update test to expect terminal button in overflow menu on mobile
- Update test to expect usage button inline on all screen sizes
- Clarify pause controls remain inline on mobile
- Fix test descriptions to match actual UI behavior
- Fix infinite loop in useMultiAgentLogs hook when handling task updates
- Correct cleanup logic to only close connections for removed tasks
- Add global EventSource mock infrastructure to vitest.setup.ts
- Refactor useMultiAgentLogs tests to use shared MockEventSource pattern
- Fix EventSource mock cleanup to prevent test pollution between cases
- Improve test isolation for SSE event handling
- Refactor test setup and teardown for better reliability
- Update UsageIndicator toggle active state to use var(--todo) and var(--bg)
- Add hover state styling for active toggle button
- Ensure consistent styling with Header view toggle across all themes
- Add changeset for patch release
- Extend mobile font size fix from textarea to all quick entry inputs (input, select)
- Update CSS selector to cover all quick entry form elements
- Add changeset for mobile zoom fix
- Add windowDurationMs (24h) to Gemini usage windows for daily rate limits
- Track resetMs in Gemini usage data for pace calculation
- Update applyPaceToWindow to apply pace to any window with timing data
- Fix pace indicators not showing for Gemini provider
- Update default VITEST_MAX_WORKERS from 8 to 16 in all packages
- Affects CLI, Core, Dashboard, and Engine vitest configs
- Improves test execution parallelism on multi-core machines
- Environment variable override still supported via VITEST_MAX_WORKERS
- Add Minimax provider backend with full test coverage
- Add Zai (Zhipu AI) provider backend for additional model options
- Implement pace calculation in UsageWindow with ahead/behind/on-pace detection
- Update UsageIndicator frontend component to consume backend pace data
- Fix pace indicator colors and icons to match design specification
- Resolve ListView QuickEntryBox type error for type safety
- Add verified POST /api/github/webhooks endpoint with signature verification
- Implement GitHub App installation token authentication for API calls
- Add webhook handlers for pull_request, issues, and issue_comment events
- Support multi-task resource matching by parsed badge URL
- Retire live GitHub polling service in favor of push-based updates
- Add comprehensive webhook integration and unit tests
- Keep 5-minute REST refresh endpoints as fallback
- Add kb task retry command to CLI with --force and --all flags
- Implement runTaskRetry() to re-run failed or completed tasks
- Add kb_task_retry extension tool for Pi agent integration
- Update CLI help text and command registration
- Add comprehensive tests for retry command and extension tool
- Include changeset for patch release
- Add `kb task delete` CLI command with required --force flag for safety
- Implement runTaskDelete function with task existence validation
- Register kb_task_delete tool in Pi extension for chat agent access
- Add comprehensive unit tests for runTaskDelete and extension registration
- Wire delete command in CLI bin.ts with proper argument parsing
- Add collapse state management with localStorage persistence per section
- Update section headers with chevron icon and click handler
- Conditionally render task rows based on collapsed state
- Add aria-expanded attribute for accessibility
- Add comprehensive tests for collapsible section behavior
- Auto-expand steps disclosure when task is in-progress
- Keep steps collapsed for todo/done/archived states
- Add comprehensive tests for auto-expand behavior in TaskCard
- Update TaskCard component with expanded state logic
- Update TaskDetailModal button text to 'Refine'
- Update test assertions in TaskDetailModal.test.tsx
- Add changeset for the UI text change
- Remove stale kb-100-header-test-branding changeset file
- Reorganize Header component button layout for mobile view
- Update Header tests with mobile layout coverage
- Improve responsive button arrangement in mobile breakpoint
- Replace native <select> with CustomModelDropdown for model selection
- Remove legacy option filtering and selection logic from SettingsModal
- Update SettingsModal tests to work with new dropdown component
- Add comprehensive test coverage for CustomModelDropdown interactions
- Add session activity tracking (lastActivityAt) to TerminalService
- Implement stale session eviction when approaching session limits
- Fix WebSocket close handler to properly kill PTY sessions on disconnect
- Add lastActivityAt to session listing endpoint for observability
- Add comprehensive tests for activity tracking, eviction, and terminal routes
- Add changeset for patch release
- Remove stale changeset files (mobile touch fixes, subtask breakdown)
- Fix type errors in QuickEntryBox and TaskCard components
- Remove obsolete test files (TaskCard.test.tsx, triage.test.ts)
- Fix import issues in planning.ts
- Clean up triage.ts and useAgentLogs.ts
- Implement touch gesture detection to distinguish tap from scroll gestures
- Add CSS touch-action property for proper mobile scroll handling
- Add comprehensive tests for touch gesture behavior in TaskCard
- Include changeset for patch release
- Implement touch gesture detection for swipe and tap interactions in TaskCard
- Add comprehensive tests for touch event handling (209 lines of test coverage)
- Include changeset for mobile touch sensitivity fix
- Enable smooth card interactions on mobile devices
- Trim whitespace from feedback before length validation
- Catch whitespace-only input at route level before store call
- Update test to verify route-level validation blocks empty feedback
- Use trimmed feedback for store operations and logging
- Fix SSE hook cleanup to prevent memory leaks and stale connections
- Cap agent log memory and optimize batch log processing
- Memoize Board, Column, and TaskCard with custom comparator to reduce re-renders
- Stabilize column task arrays and preserve pagination across live updates
- Add TaskCardBadge component for PR/issue state display
- Remove deprecated GitHub polling code and archive functionality from core store
- Replace fetch-based GitHub API calls with gh CLI in issues/fetch and issues/import routes
- Update tests to mock gh CLI instead of fetch with proper error handling
- Reduce code complexity by leveraging gh CLI's built-in JSON output
- Add changeset documenting the refactoring work
- Add ArchiveEntry type and archive.jsonl storage format for compact task archives
- Implement archiveTask() with optional cleanup and archiveTaskAndCleanup() convenience method
- Add restoreFromArchive() to reconstruct tasks from archive entries when directory is missing
- Implement cleanupArchivedTasks() for bulk cleanup of all archived task directories
- Add comprehensive tests for archive, cleanup, and restore operations (385+ lines)
- Add readArchiveLog(), findInArchive(), and unarchiveTask() API methods
- Document archive storage pattern and restoration behavior in AGENTS.md
- Include changeset for patch release noting archive cleanup feature
- Add fetchThrottled utility to GitHubClient with 429 retry handling and exponential backoff
- Implement POST /api/github/batch-import endpoint for bulk GitHub resource imports
- Add frontend API client function for batch import operations
- Add comprehensive tests for throttled requests and batch import endpoint
- Add retryAfter field to batch import result types for rate limit handling
- Clean up unfinished UsageIndicator component and worktree retry cleanup code
- Add changeset documenting the new batch GitHub import feature
- Add mergeRetries counter to track per-task retry attempts
- Auto-resolve lock files and generated files during retry attempts
- Clean up worktree when all 3 merge retry attempts are exhausted
- Add comprehensive tests for retry cleanup logic in executor
- Add changeset for the worktree retry cleanup fix
- Add pace calculation based on task velocity to UsageIndicator
- Display pace marker and progress bar with visual indicators
- Add view mode toggle for compact vs detailed display
- Include comprehensive tests for pace indicator behavior
- Add changeset for the new feature
- Fix double-render bug by using ref for hasAutoStarted instead of state\n- Enable auto-start planning tests in App.test.tsx and ListView.test.tsx\n- Add changeset for the planning modal autostart fix\n- Add comprehensive tests for planning mode auto-start behavior
- Replace inline QuickEntryBox with NewTaskModal trigger in list view
- Wire ListView to open NewTaskModal via onCreateTask prop
- Remove standalone task creation logic from QuickEntryBox
- Update App component to pass modal handlers to ListView
- Simplify QuickEntryBox to pure quick-entry without inline creation
- Update tests to verify unified modal-based creation flow
