Cleanup

A skill for identifying and removing unused code, dead files, and unnecessary dependencies. This helps reduce technical debt, improve build times, and make the codebase easier to navigate.

Workflow

Phase 1: Investigation

Identify potential candidates for removal using automated tools or manual inspection.

Unused Dependencies

Check for packages listed in project metadata (e.g., package.json, requirements.txt, go.mod) that are not actually imported or used in the source code.

• Node.js: Use depcheck or knip.
• Python: Use pip-check-reqs or deptry.
• Go: Use go mod tidy.

Unused Code & Files

Check for functions, classes, variables, or entire files that are defined but never referenced.

• Node.js/TypeScript: Use knip or ts-prune.
• Python: Use vulture.
• Go: Use unused (from staticcheck).
• General: Use grep or ripgrep to search for references across the project.

Phase 2: Verification

Before deleting anything, perform a thorough check to avoid breaking the application.

1. Check Dynamic Usage: Look for strings used in reflection, dynamic imports, or runtime lookups that might not be caught by static analysis tools.
2. Check Configuration: Ensure the item is not used by build tools, CI/CD pipelines, or configuration files (e.g., ESLint plugins, Tailwind CSS content paths).
3. Public API Check: If the repository is a library, verify if the "unused" code is part of the public API intended for external users.

Phase 3: Removal & Testing

1. Minimal Removal: Remove one component at a time if possible.
2. Run Tests: Execute the test suite to ensure no functional regressions.
3. Run Build: Perform a full build to verify that dependencies are still intact and no compilation errors were introduced.

Usage Examples

Finding unused Node.js dependencies

npx depcheck

Finding dead Python code

pip install vulture
vulture .

Cleaning up Go modules

go mod tidy

Searching for references manually

grep -r "my_function_name" .

Presentation to User

When presenting findings to the user, use the following format:

Candidates for Removal

• Type: [Dependency | Code | File]
• Item: [Name or Path]
• Evidence: [Tool output or reason why it's believed to be unused]
• Proposed Action: [Remove | Investigate further]

Troubleshooting

• False Positives: Automated tools often miss dynamic references. Always verify with grep before removal.
• Hidden Dependencies: Some dependencies are used implicitly by frameworks or tools. Check configuration files if a tool breaks after removing a "unused" dependency.
• Peer Dependencies: Be careful when removing packages that might be required by other packages but not directly imported.