Auto-Claude-Plus Auto-Claude-Plus is an autonomous multi-agent coding framework that plans, builds, and validates software with minimal human intervention. It combines Auto-Claude's spec pipeline and multi-agent architecture, Ralph Wiggum's iterative persistence loop for difficult tasks, and the frontend-design skill's philosophy for producing distinctive, polished UIs. The system features a three-tier memory system (project, global, infrastructure), a Re-Spec capability for refining project vision mid-stream, and intelligent error recovery that continues work on non-blocking tasks while flagging issues for review. Electron + React 18 + TypeScript Tailwind CSS + Custom Design System Zustand (lightweight, scalable) Radix UI primitives (accessible, unstyled) Vite Python 3.12+ Claude Agent SDK SQLite (portable, Docker-friendly) SQLAlchemy FastAPI (for internal IPC and potential future REST API) IPC between Electron and Python backend WebSocket for agent terminal streaming Docker + Docker Compose Windows, macOS, Linux - Python 3.12+ with uv package manager - Node.js 20+ with npm - Git - Docker (for containerized deployment) - Claude Pro/Max subscription - Claude Code CLI: npm install -g @anthropic-ai/claude-code 165 - Full access to all features - Manage global and infrastructure memory - Configure deployment targets - Access all projects - All routes accessible Local only (no auth required for single-user internal use) none N/A - Infrastructure memory contains server access info (stored locally only) - Git push operations require confirmation - Delete project requires confirmation - Create new project with spec interview - Open/browse existing projects - Adopt existing codebase (import and run through spec interview) - Project dashboard with status overview - Project settings and configuration - Delete project with confirmation - Spec interview/creation wizard - Re-Spec feature (re-interview with existing context) - Automatic milestone detection for large projects - Re-Spec offered at milestones (MVP, phase completion) - Spec validation and completeness checking - Spec history and version tracking - Existing work re-evaluation against new spec - Visual task board with columns (Backlog, In Progress, QA, Done, Blocked) - Task cards with status, priority, agent assignment - Drag-and-drop task movement - Task filtering by status, agent, phase - Task details panel with full context - Phase grouping and visualization - Dependency visualization - Planner Agent (creates implementation plan) - Coder Agent (implements subtasks) - QA Reviewer Agent (validates acceptance criteria) - QA Fixer Agent (resolves issues) - Parallel agent execution (up to 12 terminals) - Agent terminals UI with live output streaming - Agent status indicators (idle, working, stuck, complete) - Subagent spawning for complex subtasks - Single-pass validation (default mode) - Auto-escalation for failed tasks - Auto-escalation for detected complexity - Manual retry trigger - Ralph Wiggum mode (persistent iteration until completion) - Completion promise detection - Max iterations safety limit - Task flagging for multi-iteration mode - Session insights (what worked, what failed) - Codebase map (what files do what) - Patterns discovered (coding conventions) - Gotchas (pitfalls to avoid) - Build progress history - User preferences (coding style, UI tastes) - Cross-project patterns - Common solutions ("last time I built auth...") - Technology preferences learned over time - Dev server locations and access methods - Docker server configurations - Deployment targets (local, cloud) - Database server connections - API key storage locations - Environment-specific configurations - Memory items tagged by project or global - Memory search and retrieval - Memory editing and cleanup - Memory import/export - Design philosophy applied to all frontend projects - MVP mode (functional drafts acceptable) - Finished mode (frontend-design criteria required) - Opt-out via spec configuration - Design thinking framework (purpose, tone, constraints) - Typography, color, motion, spatial composition guidelines - Anti-pattern detection (generic AI slop) - Unit test generation for each component - E2E test generation at phase completion - Test execution against specified infrastructure - Test reporting with pass/fail details - Coverage tracking - QA loop (review -> fix -> re-review) - Infrastructure-aware testing (dev servers, Docker) - Git worktree isolation for safe development - Auto-commit per subtask completion - Auto-push on project/phase completion - Branch management (feature branches per spec) - Commit message generation - Git history visualization - Merge conflict detection - Real-time notifications (task complete, fail, blocked) - Estimated time remaining per task/phase - Changelog generation from completed tasks - Build history and statistics - Agent activity timeline - Phase progress indicators - Overall project completion percentage - Autonomous error recovery attempts - Task flagging for human review - Non-blocking continuation (work on other tasks) - Error categorization (recoverable, blocking, needs review) - Error history and patterns - Re-Spec suggestion for persistent issues - Docker containerization - Docker Compose configuration - Multi-machine access support - Infrastructure memory integration - Deployment pipeline generation - Environment configuration management - Polished, distinctive interface (non-generic) - Dark/light theme support - Responsive layout (desktop-first, but usable on tablet) - Keyboard shortcuts for power users - Settings/preferences panel - Onboarding wizard for first-time setup - id, name, description, path, status, created_at, updated_at - complexity (simple, medium, complex) - current_phase, total_phases - spec_version, last_respec_at - id, project_id, version, content, created_at - is_active, parent_spec_id (for Re-Spec tracking) - id, project_id, spec_id, phase_id, name, description - status (backlog, in_progress, qa, done, blocked) - priority, estimated_time, actual_time - agent_type, iteration_count, flagged_for_review - depends_on (JSON array of task IDs) - id, project_id, name, order, status - depends_on (JSON array of phase IDs) - milestone_type (none, mvp, release) - id, project_id, task_id, agent_type - started_at, ended_at, status, output_log - iterations, completion_promise - id, project_id, type (insight, pattern, gotcha, codebase_map) - content, metadata, created_at - id, type (preference, pattern, solution) - content, metadata, created_at, usage_count - id, name, type (server, docker, database, deployment) - config (JSON), projects (JSON array of project IDs or null for all) - created_at, updated_at - id, project_id, type, message, read, created_at - id, key, value, updated_at - GET /api/projects - List all projects - POST /api/projects - Create new project - GET /api/projects/:id - Get project details - PUT /api/projects/:id - Update project - DELETE /api/projects/:id - Delete project - POST /api/projects/:id/adopt - Adopt existing codebase - GET /api/projects/:id/spec - Get current spec - POST /api/projects/:id/spec - Create/update spec - POST /api/projects/:id/respec - Trigger Re-Spec - GET /api/projects/:id/spec/history - Get spec versions - GET /api/projects/:id/tasks - List tasks (with filters) - POST /api/projects/:id/tasks - Create task - PUT /api/tasks/:id - Update task - PUT /api/tasks/:id/status - Update task status - POST /api/tasks/:id/retry - Retry task - POST /api/tasks/:id/flag - Flag for review - POST /api/agents/start - Start agent session - POST /api/agents/:id/stop - Stop agent session - GET /api/agents/active - List active agents - WS /api/agents/:id/stream - Stream agent output - GET /api/memory/project/:id - Get project memory - POST /api/memory/project/:id - Add project memory - GET /api/memory/global - Get global memory - POST /api/memory/global - Add global memory - GET /api/memory/infrastructure - Get infrastructure memory - POST /api/memory/infrastructure - Add infrastructure config - PUT /api/memory/infrastructure/:id - Update infrastructure config - GET /api/notifications - Get notifications - PUT /api/notifications/:id/read - Mark as read - DELETE /api/notifications/:id - Delete notification - GET /api/settings - Get all settings - PUT /api/settings - Update settings - Left sidebar: Project list, navigation, quick actions - Main content area: Context-dependent (kanban, terminals, settings, etc.) - Right panel (collapsible): Task details, agent output, memory viewer - Top bar: Project name, phase indicator, notifications, settings - Bottom bar: Agent status summary, progress indicator - Dashboard: Project overview, recent activity, quick stats - Kanban Board: Visual task management - Agent Terminals: Live agent output with multiple terminals - Spec Editor: View/edit current spec, trigger Re-Spec - Memory Manager: Browse/edit all memory types - Infrastructure Config: Manage servers, Docker, deployment targets - Settings: App preferences, theme, shortcuts - Project Creation Wizard: New project setup flow - Re-Spec Interview: Guided re-interview with context - Distinctive, professional interface (not generic AI slop) - Clean but not sterile - subtle personality - Information-dense but not overwhelming - Emphasize clarity and readability for long coding sessions - Primary: Deep blue (#1a365d) - trust, focus - Accent: Vibrant teal (#0d9488) - action, success - Warning: Amber (#f59e0b) - attention - Error: Rose (#e11d48) - problems - Background: Slate gray tones (dark mode default) - Text: High contrast for readability - Headings: Inter or similar geometric sans-serif - Body: System font stack for performance - Code: JetBrains Mono or Fira Code (monospace with ligatures) - Subtle transitions (150-300ms) - Meaningful animations (task completion, agent status changes) - No gratuitous motion - respect user focus - Set up Electron + React + TypeScript project structure - Configure Vite bundler and Tailwind CSS - Set up Python backend with FastAPI - Configure SQLite database with SQLAlchemy - Implement IPC bridge between Electron and Python - Create Docker configuration - Implement database schema and migrations - Create Python models for all entities - Build CRUD API endpoints for projects, specs, tasks - Implement basic memory storage (all three types) - Set up WebSocket for real-time updates - Build spec interview wizard UI - Implement spec creation flow - Build Re-Spec interview with existing context - Implement spec versioning and history - Add milestone detection logic - Build spec validation - Integrate Claude Agent SDK - Implement Planner Agent - Implement Coder Agent - Implement QA Reviewer Agent - Implement QA Fixer Agent - Build agent session management - Implement parallel agent execution - Implement single-pass validation - Build auto-escalation detection - Implement Ralph Wiggum mode - Add completion promise detection - Build task flagging system - Implement retry logic - Build kanban board UI with drag-and-drop - Implement task cards with status - Add filtering and sorting - Build task details panel - Implement phase grouping - Add dependency visualization - Build terminal component with live streaming - Implement multi-terminal layout - Add agent status indicators - Build terminal history/scrollback - Implement terminal controls (pause, clear, copy) - Build memory manager UI - Implement project memory CRUD - Implement global memory with cross-project access - Build infrastructure memory management - Add memory search and tagging - Implement memory learning/inference - Implement git worktree management - Build auto-commit logic - Implement auto-push on completion - Add branch management UI - Build commit history visualization - Implement unit test generation - Build E2E test generation - Integrate with infrastructure memory for test targets - Build test execution and reporting - Implement QA loop (review -> fix -> re-review) - Integrate design philosophy guidelines - Build MVP vs Finished mode detection - Implement design validation checks - Add anti-pattern detection - Build notification system - Implement time estimation - Build changelog generation - Add progress indicators throughout UI - Build activity timeline - Implement autonomous recovery logic - Build task flagging for review - Implement non-blocking continuation - Add error categorization - Build Re-Spec suggestion for persistent issues - Apply design system throughout - Implement dark/light themes - Add keyboard shortcuts - Build onboarding wizard - Refine animations and transitions - Ensure responsive layout - Finalize Docker configuration - Build deployment pipeline support - Create user documentation - Add in-app help system - Final testing and bug fixes - Can create a new project through spec interview - Can adopt existing codebase and run through spec - Re-Spec works with full context awareness - Agents execute tasks autonomously - Iterative loop handles difficult tasks - Memory persists and informs across sessions - Git operations work correctly (worktree, commit, push) - QA loop validates and fixes issues - Kanban board is intuitive and responsive - Agent terminals provide clear visibility - Notifications keep user informed - Progress is always visible - Error states are clear and actionable - No mock data - all real persistence - Unit tests for core functionality - E2E tests for critical workflows - Clean error handling throughout - Performance acceptable with many tasks - UI follows design system consistently - Dark mode is fully implemented - No generic "AI slop" aesthetics - Animations are subtle and meaningful - Typography is readable for long sessions