Skill Flywheel

A unified MCP server with 839+ specialized skills for AI agent development, all running on a single FastAPI service.

🚀 Quick Start

# Install dependencies
pip install -r requirements.txt

# Start the unified server
python -m src.flywheel.server.unified_server

# Verify status
curl http://localhost:8000/health

📁 Repository Structure

• /src/flywheel/server/unified_server.py: Main entry point — FastAPI server with discovery, execution, and optimization
• /src/flywheel/core/: Components (SkillManager, Cache, Telemetry, ML, ResourceOptimizer)
• /src/flywheel/evolution/: Darwinian evolution system for skill genome optimization
• /src/flywheel/skills/: 839+ Python skill modules across 32 domains
• /domains/: SKILL.md specifications organized by domain — source material for skill generation
• /data/: SQLite registry (skill_registry.db), backlog tracking, and skill index
• /scripts/: scaffold_skill.py (generate skills), validate_skill.py (format checking), evolution_cli.py (skill evolution)
• /docs/: Architecture and developer documentation
• /tests/: Automated test suites (320 tests, all passing)

📁 Skill Domains (32)

Domain	Description
discovered	Auto-discovered skills
ML_AI	Machine learning and AI
AI_AGENT_DEVELOPMENT	AI agent frameworks and implementations
agentic_ai	Agentic AI patterns
APPLICATION_SECURITY	Security vulnerability assessment
DATA_ENGINEERING	Data pipelines and processing
orchestration	Workflow orchestration
mcp_tools	MCP tool integrations
WEB3	Decentralized finance and layer 2 scaling
SPECIFICATION_ENGINEERING	PRD generation, API design
GAME_DEV	Game development
DATABASE_ENGINEERING	Database architecture and SQL
model_orchestration	Model routing and selection
CLOUD_ENGINEERING	Distributed systems, Buildroot cross-compilation
skill_registry	Skill registry and auto-evolution
ARCHIVED	Archived skills
skill_validation	Format compliance and naming conventions
security_engineering	Security engineering
search_algorithms	A*, genetic algorithms, simulated annealing
probabilistic_models	Bayesian models and probabilistic programming
mobile_development	Cross-platform architecture
meta_agent_enhancement	Agent enhancement patterns
logic_programming	Logic programming
logic	Logic and reasoning
formal_methods	Formal verification methods
epistemology	Knowledge representation
agent_evolution	Agent evolution
FRONTEND	Frontend development
ALGO_PATTERNS	Algorithm design patterns
DEVOPS	CI/CD and infrastructure
META_SKILL_DISCOVERY	Skill discovery and indexing
QUANTUM_COMPUTING	Quantum algorithms

🔍 Key Features

• Unified FastAPI Server: Single port 8000 for discovery, execution, and metrics
• SQLite Registry: Skills registered in data/skill_registry.db
• Lazy Loading: Modules loaded on-demand, cached after 10+ hits
• ML Optimization: Predictive loading and resource adaptive scaling
• Prometheus Telemetry: Request counters, duration metrics
• Skill Scaffold: scripts/scaffold_skill.py generates properly formatted skills from CLI args or SKILL.md specs
• Format Validation: scripts/validate_skill.py checks skills match the required format

📖 API Endpoints

Discovery

Endpoint	Method	Description
/	GET	Server health
/health	GET	Detailed health with telemetry status
/skills	GET	List all skills (filter by domain, paginate)
/skills/search	GET	Search skills by name/description
/domains	GET	List domains and skill counts

Execution

Endpoint	Method	Description
/skills/execute	POST	Execute a skill by name

Optimization

Endpoint	Method	Description
/metrics	GET	System performance metrics
/skills/optimize	POST	Trigger ML optimization and predictive preloading
/skills/predict	GET	Predict usage probability for a skill
/skills/preload	GET	Get skills recommended for preloading

Evolution

Endpoint	Method	Description
/skills/evolve	POST	Trigger skill evolution for a group
/skills/evolve/groups	GET	List available evolvable skill groups

Orchestration

Endpoint	Method	Description
/skills/orchestrate	POST	Orchestrate multi-agent workflows (AutoGen, LangChain, LangGraph, CrewAI)

Security

Endpoint	Method	Description
/skills/scan	POST	Scan a skill for security vulnerabilities
/skills/scan/all	GET	Scan all skills for security vulnerabilities
/security/summary	GET	Get security monitoring summary

🛠️ Development

Creating New Skills

# From CLI
python scripts/scaffold_skill.py my_skill DOMAIN --description "Description"

# From SKILL.md spec
python scripts/scaffold_skill.py --from-spec domains/ML_AI/SKILL.name/SKILL.md

# Validate
python scripts/validate_skill.py src/flywheel/skills --recursive

# Test
python -m pytest tests/ -v

📜 License

Licensed under the MIT License.