---

name: check user_invocable: true description: | Code quality checker skill. This skill should be used when the user wants to: (1) Run static analysis tools (mypy, pyright, ruff) on the codebase - triggered by "/check", "/check q", or "/check quality" (2) Find dead/unused code (functions, imports, variables, unreachable code) - triggered by "/check d" or "/check dead"

Code Quality Checker

This skill provides two main functions for maintaining code quality in this project.

Commands

Quality Check (/check, /check q, /check quality)

Run all three static analysis tools on src/cube/ in parallel:

Copy# Run these 3 commands in parallel using the Bash tool
.venv/Scripts/python.exe -m ruff check src/cube
.venv/Scripts/python.exe -m mypy -p cube
.venv/Scripts/python.exe -m pyright src/cube

Output Format

After running all checkers, provide a summary:

1. Summary table showing pass/fail status for each tool
2. Grouped issues by severity (errors first, then warnings)
3. Actionable suggestions for fixing the issues

Offering to Fix

At the end, if there are fixable issues:

1. For ruff issues: Offer to run .venv/Scripts/python.exe -m ruff check --fix src/cube to auto-fix
2. For mypy/pyright issues: Group by category and offer to fix them one by one or in batches

Example conclusion:

Copy## Summary
| Tool    | Status | Issues |
|---------|--------|--------|
| ruff    | FAIL   | 5      |
| mypy    | PASS   | 0      |
| pyright | FAIL   | 3      |

Would you like me to:
1. Auto-fix the 5 ruff issues with `ruff check --fix`?
2. Address the 3 pyright type errors?

Dead Code Detection (/check d, /check dead)

Find unused code across the entire codebase using comprehensive analysis.

Analysis Strategy

Use the Task tool with subagent_type=Explore to perform thorough dead code analysis:

1. Unused imports - Imports that are never used in the file
2. Unused functions/methods - Functions defined but never called
3. Unused variables - Variables assigned but never read
4. Unused classes - Classes defined but never instantiated or subclassed
5. Unreachable code - Code after return/raise/break/continue statements

Execution Steps

1. Launch an Explore agent with thoroughness="very thorough" to analyze:

   • Search for all function/method definitions
   • Cross-reference with all call sites
   • Check __all__ exports in __init__.py files
   • Consider dynamic usage patterns (e.g., getattr, reflection)

2. For each category, report:

   • File path and line number
   • The unused element name
   • Confidence level (high/medium/low)
   • Reason why it appears unused

Output Format

Copy## Dead Code Analysis Results

### Unused Functions (High Confidence)
- `src/cube/module.py:42` - `_old_helper()` - No callers found
- `src/cube/other.py:15` - `deprecated_func()` - No callers found

### Unused Imports (High Confidence)
- `src/cube/file.py:3` - `from typing import Optional` - Never used

### Potentially Unused (Medium Confidence)
- `src/cube/utils.py:88` - `helper_func()` - Only called via getattr

Would you like me to remove any of these?

Caveats to Consider

• Functions referenced in __all__ are public API (not dead)
• Methods may be called via inheritance or protocols
• Some code may be used dynamically (plugin systems, etc.)
• Test files may reference code that appears "unused" in src/