cleanup
Skill1 starsChanged 9 days agoApache-2.0View on GitHub
Sweep a repository for duplication, unnecessary complexity, structural problems, and maintainability improvements. Use only when the user explicitly invokes cleanup or requests a repository cleanup sweep. Never select this skill automatically for ordinary implementation or review.
78 other projects use a version of this.
Adopted by
Most recently taken up by
- travissmithv1/claude-skillstheir own version, first seen Nov 16, 2026
- leejianrong/claude-skillstheir own version, first seen Nov 16, 2026
- dipaksavaliya/mac-dev-cleanuptheir own version, first seen Nov 16, 2026
- gregwebs/skills-sdlctheir own version, first seen Nov 15, 2026
- ekirton/Pouletheir own version, first seen Nov 15, 2026
- willcoliveira/mac-cleanup-agenttheir own version, first seen Nov 15, 2026
Used with
Projects with it also use chips, react-query-best-practices and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
udecodeSkillchips
Run the project's suggested-task ledger, the chips. /chips alone brings this session's open tasks up to date: it applies what the ledger helper says, settles stale and split tasks, cards what can start and reports the board, and /chips all also takes the ship-list tasks and the ones no live session owns, and merges duplicate tasks across the board. Use for /chips, /chips offer, start or drop, \"what chips are open\", \"suggest a task for X\", \"update the chips\", \"clean up the chips\", \"which chips are stale\", \"is this chip still worth doing\", \"merge duplicate chips\", at the last hand-back of a run that offered a task, and whenever pstack would file a tracked issue, a follow-up, a reflect Backlog item or an out-of-band PR.
Also in 3 projects that use this
bibektimilsina00Skillreact-query-best-practices
Audit and fix React Query usage in apps/web — enforce query key factories, signal forwarding, staleTime, and proper mutation patterns. Use when reviewing or fixing server-state code.
Also in 3 projects that use this
bibektimilsina00Skillyou-might-not-need-a-callback
Analyze and fix useCallback anti-patterns in your code
Also in 3 projects that use this
Scylla23Skillredo
Re-run the last routed task one model tier up (haiku → sonnet → opus → fable). User-invoked escalation when the previous result missed. Pass --dry-run to preview without running.
Also in 3 projects that use this
Aetheria-Labs1Skillplan-page
Plan one storefront page from strategy through final copy, section order, proof, offers, template direction, responsive layout, and production asset requirements. Stops at an approved execution-ready specification.
Also in 3 projects that use this
SocialKit-APISkillvideo-transcripts
Extract transcripts from social videos through SocialKit across YouTube, TikTok, Instagram, Facebook, X/Twitter, LinkedIn, and direct video file URLs. Use when the user asks to transcribe, get captions or subtitles, pull the spoken text, or read what a video says, including one video or many at once. Returns full text plus timestamped segments. Not for posting, private accounts, or unrelated website scraping.
Also in 3 projects that use this
- +3 more from bibektimilsina00
Other versions78 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 78
- simstudioai/sim★ 30ksame name
- pipecat-ai/pipecat★ 17ksame name
- udecode/plate★ 17ksame name
- civitai/civitai★ 7.3ksame name
- mampfes/hacs_waste_collection_schedule★ 2.3ksame name
- trycompai/comp★ 2.1ksame name
- udecode/dotai★ 1.2ksame name
- LedgerHQ/ledger-live★ 627same name
- ericboy0224/learn-docker-and-k8s★ 472same name
- udecode/better-convex★ 452same name
68 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository12 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- approval-gateSkill
- fresh-reviewSkill
- git-styleSkill
- interviewSkill
- no-tombstonesSkill
- orchestrateSkill
- repo-styleSkill
- setupSkill
- task-planSkill
- using-anythingSkill
The file
We split this skill into the rules it gives an agent
12 rules. 0 are common, 12 are distinctive, and 0 are very specific to this project.
0
Common
Most of the 66 edited versions we read keep these.
12
Distinctive
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 10
Run only after an explicit cleanup request.
- Distinctive: line 17
Read applicable project instructions and entry points.
- Distinctive: line 20
Use a plan or coverage record at the scale the work needs.
- Distinctive: line 28
Use the applicable lenses in references/sweep-guide.md.
- Distinctive: line 30
Look for repeated logic or configuration, unnecessary complexity, obsolete material, unclear boundaries, misplaced files, brittle tests, …
- Distinctive: line 31
Look for documentation, comments, and tests that merely memorialize removed implementations.
- Distinctive: line 34
Before removing apparently unused material, check relevant entry points, exports, dynamic loading, configuration, scripts, packaging, and …
- Distinctive: line 41
Use subagents when the available capabilities, permissions, and task size support them.
- Distinctive: line 45
Prefer read-only investigation first.
- Distinctive: line 48
Use subprocess workers only when suitable tooling is already available, permitted, and supports collecting results, tracking completion, …
- Distinctive: line 61
Keep diffs focused. Avoid unrelated formatting churn and premature abstractions.
- Distinctive: line 78
When installed and useful, interview, orchestrate, task-plan, repo-style, no-tombstones, fresh-review, approval-gate, and git-style can …
“Common” is measured: kept by at least 6 in 10 of the 66 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Cleanup
Run only after an explicit cleanup request. Discussing, implementing, or reviewing this skill does not invoke it. If loaded without a cleanup request, return to the user's task without starting a sweep.
Own a bounded repository sweep from discovery through verification. By default, fix clear issues that preserve behavior and discuss larger changes. Honor a request for findings only. This skill works on its own and does not require code, Git, subagents, or another installed skill.
Establish the scope
- Identify the target repository or requested areas, the user's goals, and any focus, exclusions, or resource limits. Ask about consequential unknowns; use the existing context for settled choices.
- Read applicable project instructions and entry points. Map the relevant source, tests, documentation, configuration, tooling, and their boundaries. Follow the existing organization unless a change solves a concrete problem.
- Inspect the current work when available. Preserve unrelated edits, untracked files, and user data; their presence does not make them cleanup candidates.
- Identify useful existing checks and record relevant baseline failures when practical. Avoid checks with unapproved effects. If execution or file access is limited, continue with the accessible evidence and state the limits.
- Use a plan or coverage record at the scale the work needs. The conversation can hold it; a broad or interrupted sweep may benefit from an existing project record. Do not require a new report, tracker, branch, or worktree.
Without writable files, return findings and proposed edits as text and state that nothing was saved.
A full sweep covers the relevant authored areas of the repository. Generated output, vendored code, external repositories, and ignored or sensitive material are normally excluded unless the request and project context make them relevant. Record what was inspected, sampled, or skipped, with meaningful limits. A targeted request narrows the sweep to its named areas or concerns.
Inspect for concrete improvements
Use the applicable lenses in references/sweep-guide.md. Start with read-only investigation so proposed changes reflect the repository as a whole. Trace related patterns across areas before deciding what to consolidate.
- Look for repeated logic or configuration, unnecessary complexity, obsolete material, unclear boundaries, misplaced files, brittle tests, stale documentation, and evidenced performance costs.
- Look for documentation, comments, and tests that merely memorialize removed implementations. Retain references and absence checks with a concrete ongoing purpose; preserve meaningful regression coverage and necessary migration guidance.
- Tie each candidate to a practical consequence: inconsistent fixes, avoidable coupling, difficult navigation, fragile changes, wasted resources, or another observable maintenance problem. Separate facts, hypotheses, and preferences.
- Preserve intentional duplication where it supports independent installation, distinct behavior or ownership, fixtures, examples, generated sources, or required license and attribution files. Similar text alone does not justify a shared abstraction.
- Before removing apparently unused material, check relevant entry points, exports, dynamic loading, configuration, scripts, packaging, and compatibility obligations. An empty reference search is evidence to investigate, not proof of disuse.
- Support performance work with a relevant measurement or a concrete, demonstrable cost. Label suspected bottlenecks as hypotheses and avoid speculative optimization.
Useful outcomes include confirming that an area needs no change. Do not invent findings, impose a preferred architecture, or optimize for fewer lines or files.
Delegate when useful and permitted
Use subagents when the available capabilities, permissions, and task size support them. Otherwise perform the same sweep directly, sequencing the relevant areas. Delegation is optional.
- Split by repository area or complementary focus, such as implementation, tests, configuration, and documentation. Keep responsibility for patterns that cross those boundaries with the integration owner.
- Give each worker a bounded brief with accessible inputs, applicable instructions, scope, resource limits, and evidence requirements. The sweep guide includes a self-contained brief.
- Prefer read-only investigation first. Assign implementation after triage, with clear ownership; sequence overlapping edits and checks that share mutable resources.
- Collect actual results and completion status. Verify important claims, combine duplicate findings by root cause, and resolve conflicting recommendations before applying changes. Reviewer agreement is not proof.
Use subprocess workers only when suitable tooling is already available, permitted, and supports collecting results, tracking completion, and stopping the processes you start. Never use a fallback to bypass a delegation restriction or install tooling merely to obtain workers.
Triage and apply changes
For each worthwhile finding, establish its location and evidence, maintenance consequence, proposed correction, likely impact and risk, and a useful verification method. Prioritize concrete benefit relative to disruption and effort. Keep the actionable set focused; report uncertain or low-value ideas accordingly.
- Findings only: Return the prioritized findings and suggested next steps. Keep repository files unchanged, including through formatters or other checks with write effects.
- Clear fixes: Apply authorized changes in coherent batches. Preserve observable behavior, public interfaces, data formats, supported compatibility, and required attribution. Update affected references, tests, and documentation together.
- Larger decisions: Present concrete options and trade-offs for changes to architecture, intended behavior, public contracts, new dependencies, or other consequential boundaries that the user has not already authorized. Continue independent work while the decision is open.
- Defer: Report out-of-scope or unsupported ideas with available evidence and a next step. Ask before adding them to an existing tracker unless already authorized; otherwise use the conversation.
An explicit cleanup request can already authorize routine local fixes throughout its stated scope. Do not ask again for every finding. Broader product work, dependency upgrades, instruction-policy changes, tool installation, and Git or external actions need authorization appropriate to those effects. Ordinary cleanup does not invoke setup.
Keep diffs focused. Avoid unrelated formatting churn and premature abstractions. Treat file moves as changes with consumers: check imports, paths, discovery rules, packaging, and documentation. Do not remove meaningful tests merely to make a check pass.
Verify and finish
Run checks appropriate to the affected artifacts and project rules. Reuse existing coverage; add or update tests when they give meaningful confidence in preserved behavior or a relevant failure. Broaden verification when shared changes or project requirements justify it. Distinguish baseline failures from regressions and explain checks that could not run.
Review the integrated result for unintended behavior changes, broken references, lost coverage, and new coupling. Use a separate review context when useful, available, and permitted; otherwise label the review as self-review. Do not claim independent review or runtime verification that did not occur.
Report:
- Coverage: areas inspected, sampled, or skipped and material limitations.
- Changes: the useful improvements actually made and why.
- Remaining findings: priority, evidence or location, and the decision or next step needed.
- Verification: checks actually performed, their results, review independence, and any remaining uncertainty. Distinguish measured performance changes from expected benefits.
Finish when the bounded sweep and authorized changes are complete and applicable verification is satisfied, or clearly identify outstanding requirements. Do not repeat sweeps solely to find more work or require a repository with zero possible improvements.
When installed and useful, interview, orchestrate, task-plan, repo-style, no-tombstones, fresh-review, approval-gate, and git-style can support the corresponding stages. Apply the guidance above directly when they are unavailable. Setup remains a separate, explicitly requested configuration task.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

