Code Cleanup

Overview

Performs post-review cosmetic cleanup to make code production-ready. This workflow is now integrated as Phase T-7 of /skill:specs-kit-task-implementation. It can also be invoked manually using --action=cleanup.

Input: docs/specs/[id]/tasks/TASK-XXX.md (reviewed status) Output: Cleaned code, task marked completed

When to Use

• Use when asked to clean up code, polish, finalize, tidy up, or remove technical debt after review approval.
• Use to prepare code for completion: remove debug logs, dead code, optimize imports, and improve readability.
• Use as the final quality gate in the specification-driven development workflow.
• Not for refactoring logic or fixing bugs — focused solely on cosmetic and hygiene cleanup.

Arguments

Best Practices

• Clean, not change: Only remove or reorganize — never change functionality
• Preserve behavior: Code must work exactly the same after cleanup
• Use project tools: Discover and run the formatter/linter the project already configures (package.json scripts, Maven/Gradle plugins, composer scripts, pyproject.toml tooling, .pre-commit-config.yaml)
• Use a checklist: Track progress through all 8 phases
• Stop on failure: If tests fail, stop and report — do not proceed

Instructions

Phase 1: Task Verification

1. Parse the skill arguments for parameters:

- --task (required): Task ID or file path - --spec (optional): Spec folder path (used with task ID)

Support two formats: - Format 1 (direct path): --task=docs/specs/001-feature/tasks/TASK-001.md - Format 2 (spec+task): --spec=docs/specs/001-feature --task=TASK-001

If Format 2 is used, construct the task file path as: {spec}/tasks/{task}.md

1. Read the task file. Verify:

- Status is reviewed, implemented, or completed (an earlier cleanup may have already stamped completed, which the loop treats as terminal) - Review report TASK-XXX--review.md exists and is approved

1. If not reviewed → stop and tell user to run /skill:specs-kit-task-review first
2. Extract task ID, title, and provides files

Phase 2: Identify Files to Clean

1. Read TASK-XXX--review.md for files created/modified
2. Read task provides field for file paths
3. Verify files exist; build cleanup list
4. Categorize: source files, test files, config files

Phase 3: Technical Debt Removal

Search files for temporary/debug artifacts with Grep:

• console.log, System.out.println, print(, // DEBUG:, // temp, // hack
• Resolved TODO/FIXME comments (keep unresolved ones)

Review context for each finding. Remove confirmed debt and document what was removed.

Phase 4: Import Optimization

1. Discover and run the project's own import tooling from its configuration (package.json scripts, Maven/Gradle plugins, composer scripts, pyproject.toml, .pre-commit-config.yaml)
2. Manually remove unused imports if no tool is configured
3. Document files changed

Phase 5: Code Readability Improvements

1. Discover and run the project's own formatter/linter from its configuration (package.json scripts, Maven/Gradle plugins, composer scripts, pyproject.toml, .pre-commit-config.yaml)
2. If no formatter is configured: fix indentation, break long lines (>120), fix spacing
3. Remove dead code only if obviously safe
4. Document changes

Phase 6: Documentation Verification

1. Verify class/file headers and public API docs
2. Check remaining TODOs are still valid and have context
3. Remove or update outdated comments
4. Document documentation changes

Phase 7: Final Verification

1. Run linters if available
2. Run tests if available
3. Verify no logic or signature changes were introduced
4. If tests fail → stop and report failures

Phase 8: Task Completion

1. Update the task status:

- Add a ## Cleanup Summary section to the task file - Check any remaining boxes in the DoD section - Inside the loop, hooks update status to reviewed (the canonical terminal the loop expects) and stamp revieweddate + cleanupdate. - Standalone run (outside the loop): no hook fires, so edit the frontmatter yourself — set status: reviewed and stamp revieweddate + cleanupdate (YYYY-MM-DD). Only pending, implemented, reviewed, completed are valid values; anything else (e.g. done) makes the task file fail to load and blocks run/refresh for the whole spec.

1. Append ## Cleanup Summary to task file with:

- Files cleaned - Changes made - Verification checklist (linters, tests, no functionality changes)

1. Mark all todos complete

Examples

Spring Boot Cleanup

/skill:specs-kit-task-implementation --task="docs/specs/001-user-auth/tasks/TASK-001.md" --action=cleanup

Actions:

1. Verify TASK-001 status is reviewed
2. Files: UserController.java, UserService.java, UserRepository.java
3. Remove 5 System.out.println and 2 resolved TODOs
4. Run ./mvnw spotless:apply
5. Run ./mvnw test -q
6. Mark task reviewed

TypeScript Cleanup

/skill:specs-kit-task-implementation --task="docs/specs/002-dashboard/tasks/TASK-003.md" --action=cleanup

Actions:

1. Verify TASK-003 status is reviewed
2. Files: Dashboard.tsx, useDashboard.ts, Dashboard.test.tsx
3. Remove 8 console.log statements
4. Run npm run lint:fix and npm run format
5. Run npm test
6. Mark task reviewed

Constraints and Warnings

• Never change logic or signatures during cleanup
• Stop immediately and report if tests fail
• Verify behavior is unchanged before marking complete