/sc-cleanup

Clean up the codebase at $ARGUMENTS (or current directory if not specified) after an AI coding session.

Purpose

AI coding sessions often leave behind artifacts:

• Debug statements (console.log, print(), debugger, binding.pry)
• Duplicate implementations (AI wrote new code instead of finding existing)
• Inconsistent naming (AI used different conventions than the codebase)
• Orphaned tests or missing test coverage
• TODO/FIXME comments that should be resolved

This command finds these issues so you can clean them up before committing.

Phase 1: Launch Analysis Agents

Launch 4 agents in parallel with run_in_background: true:

Task(subagent_type: "sc-refactor:sc-dead-code-detector", run_in_background: true,
     prompt: "Analyze [target] for dead code and debug artifacts. Focus on:
              - Debug statements (console.log, print, debugger, binding.pry, byebug)
              - Stale TODO/FIXME/HACK/XXX comments
              - Commented-out code blocks
              - Unused imports")

Task(subagent_type: "sc-refactor:sc-duplication-hunter", run_in_background: true,
     prompt: "Analyze [target] for code duplication. Focus on:
              - Duplicate function implementations
              - Copy-pasted code blocks
              - Reimplemented utility functions")

Task(subagent_type: "sc-refactor:sc-naming-auditor", run_in_background: true,
     prompt: "Analyze [target] for naming inconsistencies. Focus on:
              - Convention violations vs established patterns
              - Inconsistent casing or prefixes
              - Names that don't match the domain language")

Task(subagent_type: "sc-refactor:sc-test-organizer", run_in_background: true,
     prompt: "Analyze [target] test organization. Focus on:
              - Missing tests for new code
              - Orphaned test files
              - Test/implementation file mismatches")

Phase 2: Collect and Prioritize

Wait for all agents to complete via task-notification messages. You MUST use the results delivered in task-notifications directly — do NOT call TaskOutput on completed background agents, as the task registry purges completed entries and TaskOutput will fail with "No task found".

Categorize findings:

# Post-Session Cleanup Report

## Immediate Fixes (before committing)
[Debug statements, obvious duplicates, broken tests]

## Should Address
[Naming inconsistencies, missing tests]

## Consider Later
[Minor duplication, stylistic issues]

## Findings Summary
- Dead code/debug: [count] issues
- Duplication: [count] issues
- Naming: [count] issues
- Test organization: [count] issues

Phase 3: User Choice

Use AskUserQuestion:

Option	Label	Description
auto-fix	Auto-fix immediate issues	Remove debug statements, delete dead code (max 15 changes)
fix-plan	Generate fix plan	Step-by-step instructions for each issue
report-only	Just the report	Output report and stop

If "Auto-fix immediate issues":

1. Create TodoWrite checklist from immediate fixes
2. Implement each fix (max 15), marking complete as you go
3. Only fix high-confidence items (debug statements, obvious dead code)
4. Summarize changes made

If "Generate fix plan":

For each issue: File, Problem, Fix steps, Verification.

If "Just the report":

Output the report. Done.

What This Does NOT Check

• Functional correctness (use tests for that)
• Security issues (use sc-review-pr)
• Structural completeness (use sc-structural-check)
• Code style (use your linter)