Archon Manager
Master Archon MCP for strategic project management and knowledge operations.
Archon is the command center for AI coding assistants, providing:
- Strategic Layer: Project management, task tracking, priority-based workflow (WHAT/WHEN)
- Knowledge Layer: RAG queries, web crawling, document processing, code examples
- Integration Layer: Connects Claude Code, Cursor, Windsurf with unified context
The Two-Layer Architecture:
- Archon (this skill) = Strategic (WHAT to build, WHEN)
- Skills = Tactical (HOW to build well)
- Together = Optimal outcomes
When to Use This Skill
- Project Setup: Creating hierarchical projects with features and tasks
- Task Management: Priority-based workflow (P0/P1/P2), status tracking
- Knowledge Queries: RAG searches across documentation, code examples, PDFs
- Strategic Planning: Using Archon to decide WHAT to build next
- Two-Layer Workflow: Implementing Archon (strategic) + Skills (tactical) pattern
- Context Preservation: Maintaining project knowledge across sessions
- AI Coordination: Synchronizing multiple AI assistants on same project
Core Concepts
1. The Two-Layer Architecture
┌──────────────────────────────────────────────────┐
│ ARCHON MCP SERVER │
│ (Strategic Layer) │
│ │
│ • Project management & task tracking │
│ • Priority-based workflow (P0/P1/P2) │
│ • Knowledge queries (RAG) │
│ • Code example search │
│ • Progress tracking & metrics │
│ • Context preservation │
└─────────────────┬────────────────────────────────┘
│
│ invokes when needed
↓
┌──────────────────────────────────────────────────┐
│ AI-DEV-STANDARDS SKILLS │
│ (Tactical Layer) │
│ │
│ • Domain-specific expertise │
│ • Implementation patterns │
│ • Quality standards │
│ • Best practices │
└──────────────────────────────────────────────────┘Key Insight: Archon manages WHAT to build and WHEN, Skills guide HOW to build it well.
2. Hierarchical Project Structure
Project
├── Feature 1
│ ├── Task 1.1 (P0)
│ ├── Task 1.2 (P1)
│ └── Task 1.3 (P2)
├── Feature 2
│ ├── Task 2.1 (P0)
│ └── Task 2.2 (P1)
└── Knowledge Base
├── Web pages
├── PDFs
├── Code examples
└── Documentation3. Priority-Based Workflow
- P0 (Critical): Must have for core value prop, blocks everything
- P1 (High Value): Important but can wait, high impact
- P2 (Nice to Have): Enhancement, low priority
4. Knowledge Management
- Web Crawling: Automatic sitemap detection, intelligent scraping
- Document Processing: PDFs with intelligent chunking
- Code Examples: Extract from documentation
- Semantic Search: Vector-based RAG with embeddings
- Source Organization: Tags, categories, versions
6-Phase Archon Implementation
Phase 1: Setup & Configuration
Goal: Install and configure Archon for your project
1.1 Prerequisites
# Required
- Docker Desktop (running)
- Node.js 18+
- Supabase account (cloud: https://supabase.com or local)
- LLM API key (OpenAI, Gemini, or Ollama)1.2 Installation
# Clone Archon
git clone https://github.com/coleam00/Archon.git
cd Archon
# Create .env file
cp .env.example .env
# Edit .env with your credentials
SUPABASE_URL=your-supabase-url
SUPABASE_ANON_KEY=your-anon-key
SUPABASE_SERVICE_ROLE_KEY=your-service-role-key
OPENAI_API_KEY=your-openai-key # or GEMINI_API_KEY, OLLAMA_URL
# Set up database (run SQL migrations in Supabase SQL Editor)
# See: database/migrations/
# Start Archon
docker compose up --build -d
# Verify running
# Frontend: http://localhost:3737
# API: http://localhost:8181
# MCP: http://localhost:80511.3 Connect to AI Assistant
For Claude Code (.claude/mcp-settings.json):
{
"mcpServers": {
"archon": {
"command": "node",
"args": ["/path/to/archon/mcp-server/dist/index.js"],
"env": {
"ARCHON_API_URL": "http://localhost:8181"
}
}
}
}For Cursor/Windsurf: Similar MCP configuration in settings
1.4 Verify Connection
// Test Archon connection
archon: list_projects()
// Should return empty list or existing projectsPhase 2: Project Creation
Goal: Set up hierarchical project structure
2.1 Create Project
// Using Archon MCP tool
archon: create_project({
name: 'My Application',
description: 'Full-stack web application for task management',
status: 'active',
metadata: {
tech_stack: ['Next.js', 'Supabase', 'TypeScript'],
team_size: 1,
target_launch: '2025-12-01'
}
})
// Returns: { project_id: "uuid", name: "My Application", ... }2.2 Add Features
// Create major features
archon: create_feature({
project_id: 'uuid',
name: 'User Authentication',
description: 'Complete auth system with email/OAuth',
priority: 'P0',
estimated_effort: '2 days'
})
archon: create_feature({
project_id: 'uuid',
name: 'Task Management',
description: 'CRUD operations for tasks',
priority: 'P0',
estimated_effort: '3 days'
})
archon: create_feature({
project_id: 'uuid',
name: 'Team Collaboration',
description: 'Share tasks with team members',
priority: 'P1',
estimated_effort: '4 days'
})2.3 Break Down Features into Tasks
// Use AI-assisted task generation
archon: generate_tasks({
feature_id: 'auth-feature-uuid',
instructions: 'Break down authentication into implementation tasks',
use_ai: true
})
// Or create manually
archon: create_task({
feature_id: 'auth-feature-uuid',
title: 'Implement email/password signup',
description: 'Create signup form, API endpoint, database schema',
priority: 'P0',
status: 'todo',
estimated_hours: 4,
skills_to_use: ['api-designer', 'security-engineer', 'frontend-builder']
})Phase 3: Knowledge Base Setup
Goal: Build comprehensive knowledge base for AI queries
3.1 Add Web Documentation
// Crawl entire documentation site
archon: crawl_website({
url: 'https://nextjs.org/docs',
max_depth: 3,
follow_sitemap: true,
tags: ['nextjs', 'documentation']
})
// Archon automatically:
// - Detects sitemap
// - Crawls pages
// - Extracts text
// - Chunks intelligently
// - Generates embeddings
// - Stores in vector database3.2 Add PDF Documents
// Upload PDFs (design docs, specs, research papers)
archon: add_document({
file_path: '/path/to/architecture-spec.pdf',
type: 'pdf',
tags: ['architecture', 'design'],
project_id: 'uuid'
})
// Archon automatically:
// - Extracts text from PDF
// - Chunks by sections/pages
// - Generates embeddings
// - Indexes for search3.3 Add Code Examples
// Extract code examples from repos or docs
archon: extract_code_examples({
source_url: 'https://github.com/vercel/next.js/tree/canary/examples',
tags: ['nextjs', 'examples'],
language_filter: ['typescript', 'javascript']
})3.4 Organize Knowledge
// Tag and categorize
archon: update_source({
source_id: 'uuid',
tags: ['authentication', 'security', 'best-practices'],
category: 'implementation-guides',
version: '1.0'
})Phase 4: The Archon+Skills Workflow
Goal: Use two-layer architecture for optimal development
4.1 Phase 1: Strategic Planning (Archon)
Goal: Understand WHAT to build and WHY
// 1. Get next priority task
const task = archon:get_next_task({
project_id: "uuid",
filter_by: "status",
filter_value: "todo",
sort_by: "priority" // P0 first
})
// → { id: "P0-3", title: "Implement User Authentication", priority: "P0" }
// 2. Get task details
const details = archon:get_task({
task_id: "P0-3"
})
// → Complete task info: description, requirements, dependencies
// 3. Research the domain (RAG query)
const research = archon:perform_rag_query({
query: "JWT authentication Next.js best practices",
project_id: "uuid",
match_count: 5
})
// → Top 5 most relevant knowledge base entries with context
// 4. Find code examples
const examples = archon:search_code_examples({
query: "auth middleware Next.js TypeScript",
match_count: 3
})
// → Relevant code examples from knowledge base
// 5. Mark as doing
archon:update_task({
task_id: "P0-3",
updates: { status: "doing" }
})4.2 Phase 2: Tactical Execution (Skills)
Goal: Implement with domain expertise and best practices
// 6. Identify required skills (from task.skills_to_use)
// → ["api-designer", "security-engineer", "frontend-builder"]
// 7. Invoke skills for guidance
// (AI assistant automatically invokes these skills based on context)
// 8. Implement following both:
// - Archon research (RAG results + code examples)
// - Skill guidance (best practices + patterns)
// 9. Build the feature4.3 Phase 3: Quality Validation
Goal: Ensure quality before marking complete
// 10. Apply quality checks
// - Invoke testing-strategist skill
// - Invoke security-engineer skill
// - Run tests, validate security
// 11. Update task to review
archon:update_task({
task_id: "P0-3",
updates: { status: "review" }
})
// 12. After user validation → mark done
archon:update_task({
task_id: "P0-3",
updates: { status: "done" }
})
// 13. Get next task
const nextTask = archon:get_next_task({
project_id: "uuid",
filter_by: "status",
filter_value: "todo"
})
// → Repeat cyclePhase 5: Advanced Knowledge Operations
Goal: Leverage Archon's RAG capabilities
5.1 Semantic Search Patterns
// Broad research query
archon: perform_rag_query({
query: 'How to implement real-time features in Next.js',
match_count: 10,
similarity_threshold: 0.7
})
// Specific technical query
archon: perform_rag_query({
query: 'Next.js middleware authentication example code',
match_count: 3,
filter_tags: ['nextjs', 'authentication', 'code-example']
})
// Architecture decision query
archon: perform_rag_query({
query: 'PostgreSQL vs MongoDB for user data',
match_count: 5,
filter_tags: ['database', 'architecture']
})5.2 Context-Aware Queries
// Query within project context
archon: perform_rag_query({
query: 'How should we structure our authentication?',
project_id: 'uuid', // Uses project's knowledge base
match_count: 5
})
// Query specific feature context
archon: perform_rag_query({
query: 'Best practices for this feature',
feature_id: 'auth-feature-uuid',
match_count: 3
})5.3 Knowledge Base Versioning
// Version project documentation
archon: create_doc_version({
project_id: 'uuid',
document_name: 'Architecture Decision Record',
content: '...',
version: '1.0.0',
tags: ['architecture', 'decisions']
})
// Query historical context
archon: get_doc_history({
project_id: 'uuid',
document_name: 'Architecture Decision Record'
})Phase 6: Progress Tracking & Metrics
Goal: Monitor project health and velocity
6.1 Project Metrics
// Get project overview
archon: get_project_metrics({
project_id: 'uuid'
})
// Returns:
// {
// total_features: 5,
// total_tasks: 23,
// p0_tasks: 8,
// p1_tasks: 10,
// p2_tasks: 5,
// tasks_by_status: {
// todo: 15,
// doing: 3,
// review: 2,
// done: 3
// },
// completion_percentage: 13,
// knowledge_base_entries: 147
// }6.2 Velocity Tracking
// Tasks completed per week
archon: get_velocity({
project_id: 'uuid',
time_period: 'week'
})
// Burndown chart data
archon: get_burndown({
project_id: 'uuid',
sprint_id: 'sprint-1'
})6.3 Real-Time Updates
Archon uses Socket.IO for real-time progress updates:
- Task status changes
- Knowledge base additions
- Project metrics updates
- Team collaboration events
Integration Patterns
Pattern 1: Solo Developer Workflow
// Morning routine
const nextTask = archon:get_next_task({ project_id: "uuid" })
const research = archon:perform_rag_query({
query: nextTask.title + " implementation",
match_count: 5
})
// Work on task (using Skills for implementation)
// ...
// End of day
archon:update_task({ task_id: nextTask.id, status: "review" })Pattern 2: Team Collaboration
// Team lead creates project structure
archon: create_project({ name: 'Team Project' })
archon: create_feature({ name: 'Backend API', assigned_to: 'developer-1' })
archon: create_feature({ name: 'Frontend UI', assigned_to: 'developer-2' })
// Each team member queries same knowledge base
archon: perform_rag_query({ query: '...', project_id: 'uuid' })
// Real-time sync of task status across teamPattern 3: Multi-AI-Assistant Setup
// Claude Code for implementation
// Cursor for refactoring
// Windsurf for testing
// All connected to same Archon instance
// Same project, same tasks, same knowledge base
// Coordinated through Archon's unified contextArchon MCP Tools Reference
Project Management
// Projects
archon: create_project({ name, description, status, metadata })
archon: list_projects()
archon: get_project({ project_id })
archon: update_project({ project_id, updates })
archon: delete_project({ project_id })
// Features
archon: create_feature({ project_id, name, description, priority })
archon: list_features({ project_id })
archon: update_feature({ feature_id, updates })
// Tasks
archon: create_task({ feature_id, title, description, priority, status })
archon: get_next_task({ project_id, filter_by, sort_by })
archon: get_task({ task_id })
archon: update_task({ task_id, updates })
archon: list_tasks({ project_id, filter_by, filter_value })
archon: generate_tasks({ feature_id, instructions, use_ai })Knowledge Management
// RAG Queries
archon:perform_rag_query({query, project_id?, match_count, similarity_threshold, filter_tags?})
archon:search_code_examples({query, match_count, language_filter?})
// Content Ingestion
archon:crawl_website({url, max_depth, follow_sitemap, tags})
archon:add_document({file_path, type, tags, project_id})
archon:extract_code_examples({source_url, tags, language_filter})
// Knowledge Base Management
archon:list_sources({project_id, filter_tags?})
archon:update_source({source_id, tags, category, version})
archon:delete_source({source_id})Metrics & Analytics
archon: get_project_metrics({ project_id })
archon: get_velocity({ project_id, time_period })
archon: get_burndown({ project_id, sprint_id })Best Practices
1. Start with Clear Project Structure
// Good: Hierarchical and organized
Project: "E-commerce Platform"
├── Feature: "User Authentication" (P0)
│ ├── Task: "Implement signup" (P0)
│ ├── Task: "Implement login" (P0)
│ └── Task: "Password reset" (P1)
├── Feature: "Product Catalog" (P0)
│ ├── Task: "Product CRUD API" (P0)
│ ├── Task: "Product listing UI" (P0)
│ └── Task: "Search functionality" (P1)
// Bad: Flat, disorganized
- Task: "Build everything"
- Task: "Make it work"
- Task: "Deploy"2. Use Priority Effectively
P0 (Critical): Must have for MVP, blocks everything
- Core value proposition features
- Critical bugs
- Security vulnerabilities
P1 (High Value): Important, high impact
- Significant enhancements
- Important optimizations
- Major integrations
P2 (Nice to Have): Can wait
- Polish and refinement
- Minor features
- Non-critical improvements
3. Build Comprehensive Knowledge Base
// Add diverse sources
archon: crawl_website({ url: 'https://docs.framework.com' })
archon: add_document({ file_path: 'architecture-spec.pdf' })
archon: extract_code_examples({ source_url: 'https://github.com/...' })
// Tag consistently
tags: ['category', 'technology', 'type']
// e.g., ["authentication", "nextjs", "tutorial"]4. Use RAG Queries Strategically
// Before starting task
const research = archon:perform_rag_query({
query: "How to implement " + task.title,
match_count: 5
})
// During implementation (specific questions)
const answer = archon:perform_rag_query({
query: "How to handle edge case X",
match_count: 3
})
// Architecture decisions
const guidance = archon:perform_rag_query({
query: "Should I use pattern A or pattern B for ...",
match_count: 5
})5. Maintain Task Status Discipline
todo → doing → review → done- todo: Not started
- doing: Currently working on (limit to 1-3 tasks)
- review: Ready for validation
- done: Completed and validated
Update status immediately when changing.
Troubleshooting
Issue: Archon Not Connecting
Symptoms: MCP tools not available in AI assistant
Solutions:
- Check Docker containers running:
docker ps - Verify ports not blocked: 3737, 8181, 8051, 8052
- Check MCP configuration in
.claude/mcp-settings.json - Restart AI assistant after config changes
- Check Archon logs:
docker logs archon-api
Issue: RAG Queries Return No Results
Symptoms: Empty results from perform_rag_query
Solutions:
- Verify knowledge base has content:
archon:list_sources() - Check embeddings generated (wait for processing)
- Lower
similarity_threshold(default 0.7, try 0.5) - Broader query terms
- Check source tags match
filter_tags
Issue: Tasks Not Appearing
Symptoms: get_next_task returns empty
Solutions:
- Verify project exists:
archon:list_projects() - Check task status filters
- Ensure tasks created for correct feature/project
- Verify tasks not all marked "done"
Issue: Slow Performance
Symptoms: RAG queries or task operations slow
Solutions:
- Check Docker resource allocation
- Optimize knowledge base (remove duplicates)
- Use
match_countappropriately (5-10, not 100) - Consider upgrading Supabase plan (if cloud)
- Use local Ollama instead of API calls
Integration with ai-dev-standards
Using Archon + Skills Together
Archon provides (Strategic):
- WHAT to build (task from priority queue)
- WHEN to build it (P0/P1/P2 ordering)
- Context (RAG queries, project knowledge)
Skills provide (Tactical):
- HOW to build well (best practices, patterns)
- Domain expertise (security, performance, etc.)
- Quality standards (testing, validation)
Example Workflow:
// 1. Strategic (Archon)
const task = archon:get_next_task({project_id: "uuid"})
const research = archon:perform_rag_query({query: task.title})
archon:update_task({task_id: task.id, status: "doing"})
// 2. Tactical (Skills)
// AI automatically invokes: api-designer, security-engineer, frontend-builder
// Based on task.skills_to_use
// 3. Implementation
// Build following: Archon research + Skill guidance
// 4. Quality (Skills)
// AI invokes: testing-strategist, security-engineer
// 5. Complete (Archon)
archon:update_task({task_id: task.id, status: "done"})Task-to-Skill Mapping
When creating tasks in Archon, specify skills_to_use:
archon: create_task({
title: 'Implement authentication API',
skills_to_use: ['api-designer', 'security-engineer']
// ...
})This tells AI assistants which skills to invoke during implementation.
Success Metrics
You're using Archon effectively when:
- ✅ Clear project structure: Hierarchical, organized, prioritized
- ✅ Always know what's next:
get_next_taskguides daily work - ✅ Rich knowledge base: RAG queries return relevant, useful results
- ✅ Visible progress: Metrics show steady completion
- ✅ Status discipline: Tasks flow smoothly through workflow
- ✅ Context preserved: Can resume project after breaks without loss
- ✅ Quality maintained: Two-layer workflow (Archon + Skills) produces excellent results
Quick Reference
Common Commands
// Daily workflow
archon: get_next_task({ project_id })
archon: perform_rag_query({ query, match_count: 5 })
archon: update_task({ task_id, status: 'doing' })
// ... work ...
archon: update_task({ task_id, status: 'done' })
// Project setup
archon: create_project({ name, description })
archon: create_feature({ project_id, name, priority: 'P0' })
archon: create_task({ feature_id, title, priority: 'P0' })
// Knowledge building
archon: crawl_website({ url, tags })
archon: add_document({ file_path, tags })
// Progress tracking
archon: get_project_metrics({ project_id })
archon: list_tasks({ project_id, filter_by: 'status', filter_value: 'done' })Priority Guidelines
- P0: Core features, critical bugs, security issues
- P1: Important enhancements, optimizations, integrations
- P2: Polish, minor features, nice-to-haves
Status Flow
todo → doing (working) → review (validate) → done (complete)Summary
Archon is the strategic command center that:
- Manages WHAT to build and WHEN (priority queue)
- Provides context through RAG (knowledge base)
- Tracks progress and metrics
- Coordinates AI assistants
Combined with Skills (HOW to build well), Archon enables:
- Strategic coherence (all work aligned with goals)
- Tactical excellence (domain expertise + best practices)
- Context preservation (no lost knowledge)
- Quality assurance (multi-layer validation)
Key Takeaway: Use Archon for strategic planning (WHAT/WHEN), invoke Skills for implementation guidance (HOW). Together they create optimal outcomes.
For detailed integration patterns: See DOCS/ARCHON-INTEGRATION.md For complete example: See EXAMPLES/archon-workflow-example.md