finish-work
Skill48 starsChanged yesterdayMITView on GitHub
Close a substantive task with the right checks for its kind of work, an honest outcome summary and a written receipt. Use when a task is about to be called done, at a meaningful checkpoint, when work stops partial or blocked, or when someone says finish this, close this out, wrap this up, or asks whether it is really done.
3 other projects use a version of this.
Adopted by
Most recently taken up by
- serionist/claude-session-restoretheir own version, first seen Nov 8, 2026
- futex-ai/skillstheir own version, first seen Oct 25, 2026
- lxyer/multi-agent-collaboration-systemtheir own version, first seen Apr 16, 2026
Other versions3 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 3
- futex-ai/skills★ 1same name
- lxyer/multi-agent-collaboration-system★ 1same name
- serionist/claude-session-restore★ 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 repository13 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- llms.txtllms.txt
- adversarial-reviewSkill
- ambitiousSkill
- claude-conductorSkill
- email-triage-proSkill
- goalSkill
- hill-climbSkill
- loopSkill
- save-10-percentSkill
- strategy-promptSkill
- t3-code-conductorSkill
The file
We split this skill into the rules it gives an agent
4 rules. 0 are common, 4 are distinctive, and 0 are very specific to this project.
0
Common
Most of the 3 edited versions we read keep these.
4
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
Read only the relevant closeout guidance:
- Distinctive: line 16
Every substantive closeout or checkpoint: receipt.md.
- Distinctive: line 23
Keep proof states apart. Local, visible, staged, scheduled, published and delivered are different claims.
- Distinctive: line 25
After two failed attempts at the same fix, stop patching.
“Common” is measured: kept by at least 6 in 10 of the 3 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
Finish Work
Finish the requested outcome, verify its proof, and record the actual result. Follow the project's own verification rules and required checks first. Reuse checks already completed in this task.
Read only the relevant closeout guidance:
- Research, advice or document-only work: research.md.
- Code or executable configuration changes: code.md.
- User-visible interface changes: also ui.md.
- An authorized release or external change: also release.md.
- Every substantive closeout or checkpoint: receipt.md.
For overlapping work, use the relevant combination. Do not load every reference by default. A research task does not require code tests; a local code task does not imply deployment.
Proof
- Match the evidence to the change: inspect the text for a copy-only edit; exercise the real workflow for a behavior change.
- Keep proof states apart. Local, visible, staged, scheduled, published and delivered are different claims. For anything outside this machine, use the provider's or runtime's own readback.
- Review the final diff or artifact for regressions and unmet requirements. Investigate relevant warnings and silent failures.
- After two failed attempts at the same fix, stop patching. Write down what you learned, then retry from a clean brief or change approach from fresh evidence.
Continue through routine authorized fixes until the agreed outcome and relevant checks are complete. If essential access, intent or authority is missing, prepare everything independent of it and name the remaining step. Preserve source truth. A passing check, saved configuration or draft is evidence only for that step.
Close
Close with the outcome, material verification limits, remaining work if any, and the receipt's real state and path. Keep routine tool details in the receipt. Report complete only when the owned scope and its evidence are finished; name remaining work and unavailable verification honestly.
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.