Summary
Add an evaluation harness for measuring how Project CodeGuard guidance affects AI-generated code.
The initial implementation can start small and expand over time with additional tasks, scoring methods, delivery modes, and benchmark integrations.
Goals
- Provide a concrete starting point for CodeGuard evaluation work
- Compare AI-generated output with and without CodeGuard guidance
- Document how contributors can run and extend the harness
Related
Summary
Add an evaluation harness for measuring how Project CodeGuard guidance affects AI-generated code.
The initial implementation can start small and expand over time with additional tasks, scoring methods, delivery modes, and benchmark integrations.
Goals
Related