← LeaderboardThe Skill Leaderboard

techdebt

Find and fix technical debt including duplicated code, dead code, outdated patterns, and code smells. Run at the end of sessions to clean up.

Published by brycewang-stanford ↗brycewang-stanford/Auto-Empirical-Research-Skills

Try your first task

techdebt

Start a fresh conversation. Replace the highlighted placeholders.

Use techdebt to scan [target directory or file] for technical debt, prioritize and fix high-severity easy issues with atomic commits, then run the documented lint and test gates; leave any unresolved findings summarized for a future session.

Required inputs

  • [target directory or file]

User-approved starter prompt based on a pinned source version; check current skill requirements before running. Task evidence ↗

SKILL.md

name: techdebtdescription: Find and fix technical debt including duplicated code, dead code, outdated patterns, and code smells. Run at the end of sessions to clean up.workflow_stage: engineeringcompatibility:   - claude-code  - cursor  - codex  - gemini-cliauthor: Awesome Econ AI Communityversion: 1.0.0disable-model-invocation: truetags:   - refactoring  - code-quality  - maintenance

Technical Debt Finder

Identify and fix technical debt in the codebase.

What to Look For

Code Duplication

  • Functions with similar logic that could be consolidated
  • Copy-pasted code blocks
  • Repeated patterns that should be abstracted

Dead Code

  • Unused imports
  • Unused functions or classes
  • Commented-out code blocks
  • Unreachable code paths

Outdated Patterns

  • Deprecated API usage
  • Old-style string formatting (% or .format) vs f-strings
  • Type hints using typing.List instead of list
  • Missing type hints on public functions

Code Smells

  • Functions longer than 50 lines
  • Too many parameters (more than 5)
  • Deep nesting (more than 3 levels)
  • Magic numbers without constants
  • Overly complex conditionals

Missing Best Practices

  • Missing docstrings on public functions
  • Missing error handling
  • Hardcoded values that should be config
  • Missing tests for critical paths

Workflow

  1. Scan the Codebase

    • Look for patterns matching the issues above
    • Prioritize by impact and ease of fix

  2. Report Findings

    • List issues by category
    • Include file paths and line numbers
    • Estimate severity (high/medium/low)

  3. Fix Issues

    • Start with high-severity, easy fixes
    • Create atomic commits for each fix
    • Run tests after each change

  4. Verify

    • Run linter: ruff check .
    • Run tests: pytest
    • Ensure no new issues introduced

Arguments

Optionally specify a directory or file to focus on.

Usage:

  • /techdebt - Scan entire project
  • /techdebt src/ - Scan specific directory
  • /techdebt src/utils.py - Scan specific file

Output

Provide a summary of:

  • Issues found (by category)
  • Issues fixed
  • Remaining items for future sessions