Add Model Context Protocol (MCP) server integration to extend AI agent
capabilities with external tools. This allows users to configure MCP
servers (stdio, SSE, HTTP) in global settings and have agents use them.
Note: MCP servers are currently configured globally. Per-project MCP
server configuration is planned for a future update.
Features:
- New MCP Servers settings section with full CRUD operations
- Import/Export JSON configs (Claude Code format compatible)
- Configurable permission settings:
- Auto-approve MCP tools (bypass permission prompts)
- Unrestricted tools (allow all tools when MCP enabled)
- Refresh button to reload from settings file
Implementation:
- Added MCPServerConfig and MCPToolInfo types
- Added store actions for MCP server management
- Updated claude-provider to use configurable MCP permissions
- Updated sdk-options factory functions for MCP support
- Added settings helpers for loading MCP configs
The sandbox configuration was set in createChatOptions() and
createAutoModeOptions(), but was never passed to the ClaudeProvider.
This caused the sandbox to never actually be enabled.
Changes:
- Add sandbox field to ExecuteOptions interface
- Pass sandbox config from AgentService to provider
- Pass sandbox config from AutoModeService to provider
- Forward sandbox config in ClaudeProvider to SDK options
Now the sandbox configuration from settings is properly used.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
- Added concurrency limiting using p-limit to prevent ENFILE/EMFILE errors.
- Introduced retry logic with exponential backoff for transient file descriptor errors.
- Enhanced secure-fs with new functions for configuring throttling and monitoring active/pending operations.
- Added unit tests for throttling and retry logic to ensure reliability.
- Added rehype-sanitize as a dependency to sanitize Markdown content.
- Updated the Markdown component to include rehype-sanitize in the rehypePlugins for improved security against XSS attacks.
Add a new interactive graph view alongside the kanban board for visualizing
task dependencies. The graph view uses React Flow with dagre auto-layout to
display tasks as nodes connected by dependency edges.
Key features:
- Toggle between kanban and graph view via new control buttons
- Custom TaskNode component matching existing card styling/themes
- Animated edges that flow when tasks are in progress
- Status-aware node colors (backlog, in-progress, waiting, verified)
- Blocked tasks show lock icon with dependency count tooltip
- MiniMap for navigation in large graphs
- Zoom, pan, fit-view, and lock controls
- Horizontal/vertical layout options via dagre
- Click node to view details, double-click to edit
- Respects all 32 themes via CSS variables
- Reduced motion support for animations
New dependencies: @xyflow/react, dagre
- Changed the resolved URL for the @electron/node-gyp module from SSH to HTTPS for improved accessibility and compatibility.
This update ensures that the package can be fetched using a more universally supported URL format.
- Introduced .prettierrc for consistent code formatting with specified rules.
- Added .prettierignore to exclude unnecessary files from formatting.
- Updated package.json to include Prettier and lint-staged as devDependencies.
- Implemented GitHub Actions workflow for format checking on pull requests and pushes.
- Created a Husky pre-commit hook to run lint-staged for automatic formatting.
These changes enhance code quality and maintainability by enforcing consistent formatting across the codebase.
- Refactored session limit checks in terminal settings to use constants for minimum and maximum session values.
- Enhanced terminal layout saving mechanism with debouncing to prevent excessive writes during rapid changes.
- Updated error messages to reflect new session limit constants.
- Added new endpoints for terminal settings: GET and PUT /settings to retrieve and update terminal configurations.
- Implemented session limit checks during session creation, returning a 429 status when the limit is reached.
- Introduced a new TerminalSection in settings view for customizing terminal appearance and behavior, including font family, default font size, line height, and screen reader mode.
- Added support for new terminal features such as search functionality and improved error handling with a TerminalErrorBoundary component.
- Updated terminal layout persistence to include session IDs for reconnection and enhanced terminal state management.
- Introduced new keyboard shortcuts for terminal actions, including creating new terminal tabs.
- Enhanced UI with scrollbar theming for terminal components.
Changes:
- Added a new package, @automaker/prompts, containing AI prompt templates for enhancing user-written task descriptions.
- Implemented four enhancement modes: improve, technical, simplify, and acceptance, each with corresponding system prompts and examples.
- Updated relevant packages to utilize the new prompts package, ensuring backward compatibility with existing imports.
- Enhanced documentation to include usage examples and integration details for the new prompts.
Benefits:
✅ Streamlined AI prompt management across the codebase
✅ Improved clarity and usability for AI-powered features
✅ Comprehensive documentation for developers
All tests passing.
This commit addresses all "Should Fix" items from the PR review:
1. Security Documentation (platform package)
- Added comprehensive inline documentation in security.ts explaining
why path validation is disabled
- Added Security Model section to platform README.md
- Documented rationale, implications, and future re-enabling steps
2. Model Resolver Tests
- Created comprehensive test suite (34 tests, 100% coverage)
- Added vitest configuration with strict coverage thresholds
- Tests cover: alias resolution, full model strings, priority handling,
edge cases, and integration scenarios
- Updated package.json with test scripts and vitest dependency
3. Feature Loader Logging Migration
- Replaced all console.log/warn/error calls with @automaker/utils logger
- Consistent with rest of codebase logging pattern
- Updated corresponding tests to match new logger format
4. Module Format Consistency
- Verified all packages use consistent module formats (ESM)
- No changes needed
All tests passing (632 tests across 31 test files).
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
- Standardize vitest to v4.0.16 across all packages
- Clean up type imports in events.ts (remove verbose inline casting)
- Expand skipDirs to support Python, Rust, Go, PHP, Gradle projects
- Document circular dependency prevention in @automaker/types
- Add comprehensive error handling documentation to @automaker/git-utils
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
Changes:
- Move subprocess-manager tests to @automaker/platform package
- Tests need to be co-located with source for proper mocking
- Add vitest configuration to platform package
- 17/17 platform tests pass
- Update server vitest.config.ts to alias @automaker/* packages
- Resolve to source files for proper mocking in tests
- Enables vi.mock() and vi.spyOn() to work correctly
- Fix security.test.ts imports
- Update dynamic imports from @/lib/security.js to @automaker/platform
- Module was moved to shared package
- Rewrite prompt-builder.test.ts
- Use fs/promises mock instead of trying to spy on internal calls
- 10/10 tests pass
Test Results:
✅ Server: 536/536 tests pass
✅ Platform: 17/17 tests pass
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
CRITICAL FIXES:
- Fix dependency-resolver ES module failure by reverting to CommonJS
- Removed "type": "module" from package.json
- Changed tsconfig.json module from "ESNext" to "commonjs"
- Added exports field for better module resolution
- Package now works correctly at runtime
- Fix Feature type incompatibility between server and UI
- Added FeatureImagePath interface to @automaker/types
- Made imagePaths property accept multiple formats
- Added index signature for backward compatibility
HIGH PRIORITY FIXES:
- Remove duplicate model-resolver.ts from apps/server/src/lib/
- Update sdk-options.ts to import from @automaker/model-resolver
- Use @automaker/types for CLAUDE_MODEL_MAP and DEFAULT_MODELS
- Remove duplicate session types from apps/ui/src/types/
- Deleted identical session.ts file
- Use @automaker/types for session type definitions
- Update source file Feature imports
- Fix create.ts and update.ts to import Feature from @automaker/types
- Separate Feature type import from FeatureLoader class import
MEDIUM PRIORITY FIXES:
- Remove unused imports
- Remove unused AbortError from agent-service.ts
- Remove unused MessageSquare icon from kanban-card.tsx
- Consolidate duplicate React imports in hotkey-button.tsx
- Update test file imports to use @automaker/* packages
- Update 12 test files to import from @automaker/utils
- Update 2 test files to import from @automaker/platform
- Update 1 test file to import from @automaker/model-resolver
- Update dependency-resolver.test.ts imports
- Update providers/types imports to @automaker/types
VERIFICATION:
- Server builds successfully ✓
- All 6 shared packages build correctly ✓
- Test imports updated and verified ✓
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
- Updated 150+ files to import from @automaker/* packages
- Server imports now use @automaker/utils, @automaker/platform, @automaker/types, @automaker/model-resolver, @automaker/dependency-resolver, @automaker/git-utils
- UI imports now use @automaker/dependency-resolver and @automaker/types
- Deleted duplicate dependency-resolver files (222 lines eliminated)
- Updated dependency-resolver to use ES modules for Vite compatibility
- Added type annotation fix in auto-mode-service.ts
- Updated feature-loader to re-export Feature type from @automaker/types
- Both server and UI builds successfully verified
Phase 1 of server refactoring complete.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
Update package-lock.json to recognize all 6 new workspace packages:
- @automaker/types
- @automaker/utils
- @automaker/platform
- @automaker/model-resolver
- @automaker/dependency-resolver
- @automaker/git-utils
All packages are now part of the npm workspace.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
- Merge PR #162: cross-platform dev script (init.mjs)
- Updated init.mjs to reference apps/ui instead of apps/app
- Updated root package.json scripts to use apps/ui workspace
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Merge latest features from main including:
- PR #161 (worktree-confusion): Clarified branch handling in dialogs
- PR #160 (speckits-rebase): Planning mode functionality
Resolved conflicts:
- add-feature-dialog.tsx: Combined TanStack Router navigation with branch selection state
- worktree-integration.spec.ts: Updated tests for new worktree behavior (created at execution time)
- package-lock.json: Regenerated after merge
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Replace Unix-only init.sh with cross-platform init.mjs Node.js script.
Changes:
- Add init.mjs: Cross-platform Node.js implementation of init.sh
- Update package.json: Change dev script from ./init.sh to node init.mjs
- Add tree-kill dependency for reliable cross-platform process termination
Key features of init.mjs:
- Cross-platform port detection (netstat on Windows, lsof on Unix)
- Cross-platform process killing using tree-kill package
- Uses cross-spawn for reliable npm/npx command execution on Windows
- Interactive prompts via Node.js readline module
- Colored terminal output (works on modern Windows terminals)
- Proper cleanup handlers for Ctrl+C/SIGTERM
Bug fix:
- Fixed Playwright browser check to run from apps/app directory where
@playwright/test is actually installed (was silently failing before)
The original init.sh is preserved for backward compatibility.
- Added new RadioGroup and Switch components for better UI interaction.
- Introduced BranchSelector for improved branch selection in feature dialogs.
- Updated Autocomplete and BranchAutocomplete components to handle error states.
- Refactored feature management to archive verified features instead of deleting them.
- Enhanced worktree handling by removing worktreePath from features, relying on branchName instead.
- Improved auto mode functionality by integrating branch management and worktree updates.
- Cleaned up unused code and optimized existing logic for better performance.
- Added a new PlanningMode feature to manage default planning strategies for features.
- Updated the FeatureDefaultsSection to include a dropdown for selecting the default planning mode.
- Enhanced AddFeatureDialog and EditFeatureDialog to support planning mode selection and state management.
- Introduced PlanningModeSelector component for better user interaction with planning modes.
- Updated app state management to include default planning mode and related specifications.
- Refactored various UI components to ensure compatibility with new planning mode features.
