hygiene
Skill6 starsChanged 2 months agoMITView on GitHub
Restore the environment after agent work: close only tabs you opened, kill only processes you started, delete only scratch you created, revert state you changed. Use when finishing any task or ending a session, after creating temp files, opening browser tabs or workspaces, or starting background processes.
10 other projects use a version of this, 1 of them edited.
Adopted by
Most recently taken up by
- RohiRIK/skillstheir own version, first seen Nov 15, 2026
- NomadicDaddy/aiddtheir own version, first seen Nov 15, 2026
- e-reitbauer/habitstheir own version, first seen Nov 8, 2026
- NeuralInverse/neuralinversetheir own version, first seen Oct 8, 2026
- OpenCortexIDE/cortexidetheir own version, first seen Oct 8, 2026
- gitpod-io/openvscode-servertheir own version, first seen Oct 8, 2026
Used with
Projects with it also use add-policy, auto-perf-optimize 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.
OpenCortexIDESkilladd-policy
Use when adding, modifying, or reviewing VS Code configuration policies. Covers the full policy lifecycle from registration to export to platform-specific artifacts. Run on ANY change that adds a `policy:` field to a configuration property.
Also in 4 projects that use this
MicrosoftSkillauto-perf-optimize
Run agent-driven VS Code performance or memory investigations. Use when asked to launch Code OSS, automate a VS Code scenario, run the Chat memory smoke runner, capture renderer heap snapshots, take workflow screenshots, compare run summaries, or drive a repeatable scenario before heap-snapshot analysis.
Also in 4 projects that use this
MicrosoftSkillchat-customizations-editor
Use when working on the Chat Customizations editor — the management UI for agents, skills, instructions, hooks, prompts, MCP servers, and plugins.
Also in 4 projects that use this
- +9 more from microsoft
Other versions10 projects use a version of this.
Word for word · 0
None found.
Edited · 1
- kartikkabadi/skills★ 9100% same
Wrote their own version · 9
- gitpod-io/openvscode-server★ 6.2ksame name
- NeuralInverse/neuralinverse★ 221same name
- OpenCortexIDE/cortexide★ 175same name
- cpkt9762/opencode-vscode-ide★ 17same name
- xAmirHamza77/OmniCode★ 12same name
- e-reitbauer/habits★ 1same name
- NomadicDaddy/aidd★ 1same name
- RohiRIK/skills★ 1same name
- CommsConnectgmbh/claude-starter-kit★ 0same name
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 repository10 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- commitSkill
- gh-stackSkill
- grade-itSkill
- install-anti-slopSkill
- principle-prove-it-worksSkill
- review-commentSkill
- react-doctorSkill
- verifySkill
The file
We split this skill into the rules it gives an agent
5 rules. 0 are common, 4 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 8 edited versions we read keep these.
4
Distinctive
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 20
Scratch goes in /tmp or the system temp dir: Never the home directory, never the repo root, never ~/Downloads unless the user asked for a …
- Distinctive: line 21
Do not create files the user did not ask for: No summary.md, notes.md, plan.md unless they requested a document.
- Distinctive: line 27
Before you report done, walk the mess:
- Distinctive: line 33
State you changed: If you toggled a setting, checked out a branch, or pointed a config somewhere temporary, put it back.
- Distinctive: line 37
If cleanup deleted or closed things, one line in the final report: "cleaned up: removed /tmp scratch, closed 3 tabs, killed the dev …
“Common” is measured: kept by at least 6 in 10 of the 8 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
Hygiene
The rule: leave the environment as you found it, minus the one artifact the user asked for. The cleanup pass covers only what you created — files, tabs, workspaces, processes, state. Run it when work ends, not someday.
Own it, or leave it
Cleanup only ever touches what you created. The test for anything you find is: did I make this? If not, leave it. If you are unsure, leave it and mention it.
User things are off limits: their files, their tabs and workspaces, their processes, their uncommitted changes. Deleting is for scratch only. The deliverable, verification artifacts the user will want (a diff, a log of a failure you are reporting), and anything referenced in your report all stay.
During the work
Prevention beats cleanup, and the habits that prevent mess are also the record of what you created:
- Scratch goes in /tmp or the system temp dir. Never the home directory, never the repo root, never ~/Downloads unless the user asked for a file there. Name scratch files clearly:
/tmp/hygiene-audio-probe.pybeats~/test2.py. - Do not create files the user did not ask for. No summary.md, notes.md, plan.md unless they requested a document. The chat reply is the summary.
- One browser tab per task; one workspace per session. Whatever browser surface you used — an in-app browser, ego-browser, Playwright, a CLI automation tool — keep its tabs and workspaces tidy by the same rule: tabs you opened close after the check; workspaces, spaces, or contexts you created close with the task. Use the browser tool's own teardown API when it has one (ego-browser's
completeTaskSpace, Playwright's context close); that tool's skill documents its cleanup contract. - Kill what you start. Background servers, watchers, tail -f, dev processes: stop them when done or hand them off explicitly ("left the dev server running, it's on :3000").
At the end
Before you report done, walk the mess:
- Files. List what you created outside the deliverable: scripts, downloaded fixtures, extracted archives, build output, screenshots. Delete the scratch ones. Keep only what the user asked for or what a future session genuinely needs, and put keepers somewhere sensible (the repo, a project folder), not /tmp and not home.
- Tabs, workspaces, windows. Close the tabs you opened and the workspaces, spaces, or contexts you created, using the browser tool's own teardown API where one exists (ego-browser
completeTaskSpace, Playwright context close, in-app browser close). Close preview panes and terminals you spawned for one-off checks. What you did not open stays: the user's tabs, workspaces, windows, and the browser's restore history are theirs. - Processes. Kill anything still running that isn't meant to outlive the session.
- Bloat inside files. The cleanup applies to content too, not just file count. If you wrote 40 lines where 10 work, trim. Delete debug prints, commented-out dead code, and "notes to self" comments before calling it done.
- State you changed. If you toggled a setting, checked out a branch, or pointed a config somewhere temporary, put it back.
The report
If cleanup deleted or closed things, one line in the final report: "cleaned up: removed /tmp scratch, closed 3 tabs, killed the dev server." If nothing needed cleaning, say nothing.
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.

