refactor-clean

/refactor-clean - Dead Code Removal

Detect and safely remove dead code, unused dependencies, and duplicates.

Instructions

You are a codebase cleaner. Remove dead weight methodically — one item at a time, tests after each removal.

Step 1: Detect Dead Code

Run language-appropriate detection tools:

JavaScript/TypeScript: npx knip, npx depcheck, npx ts-prune
Python:               vulture, pip-extra-reqs, autoflake --check
Go:                   staticcheck, deadcode
Ruby:                 debride, unused
General:              grep for unused imports, unreferenced files

Collect all findings before removing anything.

Step 2: Categorize by Safety Tier

Sort every finding into one of three tiers:

SAFE     — Unused private functions, unreferenced local variables,
           unused test helpers, dead branches behind constant flags

CAUTION  — Unused exports (may have dynamic importers),
           unused dependencies (may be peer/optional),
           unused CSS classes (may be referenced in JS)

DANGER   — Config files, entry points, public API exports,
           anything referenced in build scripts or CI

Step 3: Remove SAFE Items

For each SAFE item:

1. Remove the dead code
2. Run {{test_command}}
3. If tests fail, revert immediately and recategorize as CAUTION
4. Record what was removed and lines saved

Step 4: Review CAUTION Items

For each CAUTION item:

1. Search the entire codebase for dynamic references (string literals, reflection, config)
2. Check build scripts, CI configs, and documentation
3. If confirmed dead, remove and run {{test_command}}
4. If uncertain, skip and note for human review

Step 5: Skip DANGER Items

List all DANGER items in the report for human decision. Do NOT remove them.

Step 6: Consolidate Duplicates

After dead code is cleaned:

1. Search for near-duplicate functions (same logic, different names)
2. Propose consolidation — do NOT auto-merge without confirmation
3. If approved, consolidate one at a time, testing after each