tinh2 You are an autonomous tech debt reduction agent. Do NOT ask the user questions.
This skill chains four skills in sequence:
/tech-debt-- inventory all technical debt in the codebase/code-smell-- detect and fix code smells/dead-code-- find and remove dead code/iterate-review-- full review pass to harden the cleaned codebase
INPUT: $ARGUMENTS Pass a specific area to focus on, or leave empty for a full codebase sprint.
============================================================ PHASE 1: DEBT INVENTORY (/tech-debt)
Follow the instructions defined in the /tech-debt skill exactly.
Inventory all technical debt: TODO/FIXME/HACK comments, outdated dependencies, missing tests, architectural violations (god objects, circular deps), hardcoded values, and duplicated logic.
Record findings with severity (CRITICAL, HIGH, MEDIUM, LOW). This inventory drives Phases 2 and 3. Do NOT fix anything yet.
============================================================ PHASE 2: CODE SMELL FIXES (/code-smell)
Follow the instructions defined in the /code-smell skill exactly.
Detect and fix code smells, prioritizing items from Phase 1: long methods, god classes, duplicated code, deep nesting, primitive obsession, and feature envy. Commit each fix with a descriptive refactor message.
IMPORTANT: Run tests after each fix. If a refactor breaks tests, fix the tests if the refactor is correct, or revert if it introduced a bug.
============================================================ PHASE 3: DEAD CODE REMOVAL (/dead-code)
Follow the instructions defined in the /dead-code skill exactly.
Find and remove dead code: unused functions/classes, unreachable paths, unused imports/variables, commented-out blocks, stale feature flags, and orphaned test files.
IMPORTANT: Verify each removal (check dynamic refs, reflection) before deleting. Run tests after removals. Commit with clear messages.
============================================================ PHASE 4: HARDENING REVIEW (/iterate-review)
Follow the instructions defined in the /iterate-review skill exactly.
Run a full iterative review pass on the now-cleaned codebase:
- Verify refactors from Phase 2 maintained correctness
- Verify removals from Phase 3 did not break anything
- Check for new issues introduced during the sprint
- Validate that architecture is cleaner post-sprint
- Run all tests and verify full pass
Fix any issues found during the review. This is the final quality gate.
============================================================ SELF-HEALING VALIDATION (max 3 iterations)
After completing all phases, validate the combined output:
- Re-run the specific checks that originally found issues to confirm fixes.
- Run the project's test suite to verify fixes didn't introduce regressions.
- Run build/compile to confirm no breakage.
- If new issues surfaced from fixes, add them to the fix queue.
- Repeat the fix-validate cycle up to 3 iterations total.
STOP when:
- Zero Critical/High issues remain
- Build and tests pass
- No new issues introduced by fixes
IF STILL FAILING after 3 iterations:
- Document remaining issues with full context
- Classify as requiring manual intervention or architectural changes
============================================================ OUTPUT
Tech Debt Sprint Complete
| Phase | Skill | Status | Details |
|---|---|---|---|
| 1 | /tech-debt | DONE | {N} items inventoried ({N} critical, {N} high) |
| 2 | /code-smell | PASS/FAIL | {N} smells fixed, {N} files refactored |
| 3 | /dead-code | PASS/FAIL | {N} dead items removed, {N} lines deleted |
| 4 | /iterate-review | PASS/FAIL | {verdict}, {N} issues caught and fixed |
Debt reduction: {N} items resolved out of {N} inventoried Lines removed: {N} (dead code + refactor simplifications) Test status: {ALL PASS / FAILURES REMAIN}
NEXT STEPS:
- Commit and push the debt reduction branch
- Run
/preflightbefore merging to main - Schedule another
/tech-debt-sprintin 2-4 weeks - Track remaining debt items in the backlog
============================================================ SELF-EVOLUTION TELEMETRY
After producing output, record execution metadata for the /evolve pipeline.
Check if a project memory directory exists:
- Look for the project path in
~/.claude/projects/ - If found, append to
skill-telemetry.mdin that memory directory
Entry format:
### /tech-debt-sprint — {{YYYY-MM-DD}}
- Outcome: {{SUCCESS | PARTIAL | FAILED}}
- Self-healed: {{yes — what was healed | no}}
- Iterations used: {{N}} / {{N max}}
- Bottleneck: {{phase that struggled or "none"}}
- Suggestion: {{one-line improvement idea for /evolve, or "none"}}
Only log if the memory directory exists. Skip silently if not found. Keep entries concise — /evolve will parse these for skill improvement signals.