improvement(cleanup-skill): parallelize analysis, apply fixes sequentially

[TheodoreSpeaks]

Summary

• Rewrote the /cleanup skill to fan out the seven code-quality passes as read-only parallel subagents (fix=false), then converge and apply fixes sequentially in dependency order
• Avoids concurrent-write clobbering on overlapping files while keeping the analysis-phase speedup

Type of Change

• Improvement

Testing

Verified all seven sub-skills support the fix=false propose-only path

Checklist

• Code follows project style guidelines
• Self-reviewed my changes
• Tests added/updated and passing
• No new warnings introduced
• I confirm that I have read and agree to the terms outlined in the Contributor License Agreement (CLA)

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
docs	Ready	Preview, Comment	Jul 12, 2026 2:37am

[cursor]

PR Summary

Low Risk
Changes are limited to agent skills and command markdown; no app, API, or migration runtime behavior is modified.

Overview
/cleanup is reworked from a single sequential chain into four steps: parse scope/fix, run eight quality passes in parallel with fix=false, converge overlapping proposals (tagged by pass), then apply fixes pass-by-pass in dependency order (effects → … → comments) using content-anchored edits so parallel snapshots stay valid. The pipeline now includes you-might-not-need-url-state and you-might-not-need-a-comment.

New .agents/skills for comment hygiene and nuqs URL-state anti-patterns, plus matching .claude/commands and .cursor/commands copies (including full babysit PR-review loop docs). babysit post-push verify clarifies that git log --reverse is required so commit order matches the PR list.

The PR also adds several large agent-only command guides (e.g. db-migrate, memory-load-check, frontend polish / anti-slop design skills) with no application runtime changes.

^{Reviewed by Cursor Bugbot for commit 3a32a53. Bugbot is set up for automated code reviews on this repo. Configure here.}

[greptile-apps]

Greptile Summary

This PR changes the cleanup workflow to keep parallel analysis while applying edits in order. The main changes are:

• Cleanup now parses scope and fix arguments before running passes.
• The analysis passes run concurrently with fixes disabled.
• Proposed edits are grouped by pass and applied sequentially.
• URL-state and comment cleanup passes were added to the cleanup set.
• Several command and skill docs were added or mirrored for Claude and Cursor.

Confidence Score: 5/5

This looks safe to merge.

• No blocking issues found in the changed code.

Important Files Changed

Filename	Overview
.agents/skills/cleanup/SKILL.md	Updates the cleanup skill to parse arguments, run read-only passes in parallel, reconcile proposals, and apply surviving edits in dependency order.
.claude/commands/cleanup.md	Mirrors the cleanup command flow for Claude with explicit scope substitution and sequential apply steps.
.cursor/commands/cleanup.md	Mirrors the cleanup command flow for Cursor with explicit scope substitution and sequential apply steps.
.agents/skills/you-might-not-need-url-state/SKILL.md	Adds a URL-state cleanup skill covering nuqs usage, manual query handling, and shareable view-state patterns.
.agents/skills/you-might-not-need-a-comment/SKILL.md	Adds a comment cleanup skill for removing redundant inline comments and promoting real documentation to TSDoc.

_{Reviews (7): Last reviewed commit: "fix(skills): port url-state + comment pa..." | Re-trigger Greptile}

[TheodoreSpeaks]

@greptile

[TheodoreSpeaks]

@cursor review

[TheodoreSpeaks]

@greptile

[TheodoreSpeaks]

@cursor review

[TheodoreSpeaks]

@greptile

[TheodoreSpeaks]

@cursor review

[TheodoreSpeaks]

@greptile

[TheodoreSpeaks]

@cursor review

[TheodoreSpeaks]

@greptile

[TheodoreSpeaks]

@cursor review

[TheodoreSpeaks]

@greptile

[TheodoreSpeaks]

@cursor review

[cursor]

Cursor Bugbot has reviewed your changes using default effort and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 3a32a53. Configure here.}

[cursor]

Lint check cannot target files

Medium Severity

Step 3 tells the orchestrator to run bun run lint:check on the touched files, but the root script is turbo run lint:check with no per-path API. Following the wording literally may pass stray arguments to Turbo or skip lint while believing only edited paths were checked.

Additional Locations (2)

• .claude/commands/cleanup.md#L52-L53
• .cursor/commands/cleanup.md#L47-L48

 

^{Reviewed by Cursor Bugbot for commit 3a32a53. Configure here.}