hf.audit-code

Code Quality Audit

Run a comprehensive code quality audit across the entire repo. Dynamically analyzes source code for dead code, complexity, duplication, error handling gaps, type safety issues, class cohesion, and method size. Creates GitHub issues for findings so HydraFlow can process them.

Instructions

1. Resolve configuration before doing anything else:

   • Run echo "$HYDRAFLOW_GITHUB_REPO" — if set, use it as the target repo (e.g., owner/repo). If empty, run git remote get-url origin and extract the owner/repo slug (strip https://github.com/ prefix and .git suffix).
   • Run echo "$HYDRAFLOW_GITHUB_ASSIGNEE" — if set, use it as the issue assignee. If empty, extract the owner from the repo slug (the part before /).
   • Use hydraflow-find as the label for created issues.
   • Store resolved values as $REPO, $ASSIGNEE, $LABEL.

2. Discover project structure:

   • Use Glob to find all *.py source files, excluding .venv/, venv/, __pycache__/, node_modules/, dist/, build/.
   • Separate into SOURCE files (production code) and TEST files (anything under tests/ or matching test_*.py).
   • Count total source files, total lines, and identify the main modules.

3. Launch agents in parallel using Task with run_in_background: true and subagent_type: "general-purpose":

   • Agent 1: Dead code & unused exports — Finds unreachable code, unused functions, unused imports, and stale modules.
   • Agent 2: Method size, class cohesion & duplication — Enforces small methods, single-concept classes, and DRY.
   • Agent 3: Error handling & robustness — Finds bare excepts, swallowed errors, missing error paths, and fragile patterns.
   • Agent 4: Type safety & API consistency — Finds missing type annotations, Any overuse, inconsistent return types, and public API gaps.
   • Agent 5: Convention drift & over-engineering — Reads docs/wiki/gotchas.md (or CLAUDE.md fallback) at runtime and sweeps the codebase for documented avoided patterns plus over-engineering accumulation.