- Add explicit type annotations for variables in responsibility modules
- Fix Pydantic model constructor calls to include optional fields with defaults
- Fix B904 exception handling by adding 'from e' clauses in agent.py
- Fix RET504 unnecessary assignments before return statements
- Fix threading.Lock type annotation issue in rpm_controller.py
- Update pyproject.toml to ignore S101 assert statements in test files
- Add set_responsibility_system method to BaseAgent class
All responsibility tracking tests pass (58/58) and type-checker shows no issues.
Co-Authored-By: João <joao@crewai.com>
- Add capability-based agent hierarchy with mathematical scoring
- Implement responsibility assignment algorithms (greedy, balanced, optimal)
- Add comprehensive accountability logging and tracking
- Implement performance-based capability adjustment with learning rates
- Integrate with existing Agent and Crew classes seamlessly
- Add 58 comprehensive tests covering all functionality
- Include example usage demonstrating all features
Addresses issue #3491 with four key features:
1. Capability-Based Agent Hierarchy
2. Mathematical Responsibility Assignment
3. Accountability Logging
4. Performance-Based Capability Adjustment
The system is fully backward compatible and optional - existing crews
continue to work without modification.
Co-Authored-By: João <joao@crewai.com>
