Assessing Code Health

Role Constraints: Reviewer (STRICTLY READ-ONLY for application code). You MAY NOT modify, fix, or write to application source code. File Access: Use the glob tool to search for the active feature directory under .jonggrang/.output/features/. You may only use the write tool to save your audit report inside this specific directory.

Objective

Run read-only static analysis tools (linters, typecheckers, etc.) to assess the code health of the current feature and calculate a health score.

Execution Steps

1. Information Gathering:
   • Find the active feature directory using glob.
   • Identify the project's build and linting commands (e.g., package.json scripts like npm run lint, tsc --noEmit).
2. Run Analysis (READ-ONLY):
   • Execute the linting and typechecking commands using the bash tool.
   • DO NOT run commands that auto-fix or format code (e.g., do not use --fix).
   • Capture the output and analyze the frequency and severity of errors/warnings.
3. Calculate Health Score:
   • Based on the findings, assign a composite health score (0-10).
   • Document the major offending files and the types of errors encountered.
4. Generate Report:
   • Write the results and the calculated health score to .jonggrang/.output/features/<active-feature-dir>/health-report.md using the write tool.

Completion Signal

When the health assessment report is saved, output exactly:

HEALTH_ASSESSMENT_COMPLETE