Expert agent framework for the clewcrew portfolio.
clewcrew-agents provides AI-powered expert agents for different domains including security, code quality, testing, build, architecture, and model validation. Each agent specializes in detecting specific types of hallucinations in code.
# Install with pip
pip install clewcrew-agents
# Install with UV
uv add clewcrew-agentsfrom clewcrew_agents import SecurityExpert, BaseExpert
# Initialize a security expert
security_expert = SecurityExpert()
# Detect security hallucinations
result = await security_expert.detect_hallucinations("/path/to/project")
# Check results
print(f"Found {len(result.hallucinations)} security issues")
print(f"Confidence: {result.confidence}")
print(f"Recommendations: {result.recommendations}")
# Get fix suggestions
fixes = await security_expert.suggest_fixes(result.hallucinations)
for fix in fixes:
print(f"Fix: {fix['fix']}")
print(f"Priority: {fix['priority']}")- BaseExpert: Abstract base class for all expert agents
- HallucinationResult: Standardized result format for all agents
- SecurityExpert: Specialized agent for security vulnerability detection
Each expert agent provides:
- Hallucination Detection: Domain-specific issue identification
- Confidence Scoring: Reliability assessment of findings
- Recommendations: Actionable advice for issue resolution
- Fix Suggestions: Specific solutions for detected problems
- Risk Assessment: Priority and severity evaluation
- clewcrew-common: Shared utilities and patterns
- clewcrew-framework: Base classes and abstractions
- pydantic: Data validation and settings management
# Run tests
pytest tests/
# Run with coverage
pytest tests/ --cov=src/clewcrew_agents --cov-report=htmlWe welcome contributions! Please see our Contributing Guide for details.
This project is licensed under the MIT License - see the LICENSE file for details.
Ready to deploy expert agents for the clewcrew revolution! 🤖✨