minimal-diff
Skill30 starsChanged 4 months agoMITView on GitHub
Scope discipline for code changes. Use whenever writing or editing code: implementing a feature, fixing a bug, or modifying existing files. Trigger especially when tempted to refactor, clean up, add error handling, add configuration options, or restructure code beyond what the task asked for.
3 other projects use a version of this.
Adopted by
Most recently taken up by
- aprameyak/coding-skillstheir own version, first seen Oct 2, 2026
- poorvith-mp/skills-developertheir own version, first seen Aug 19, 2026
- vichudo/skillstheir own version, first seen Jul 11, 2026
Other versions3 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 3
- vichudo/skills★ 2same name
- aprameyak/coding-skills★ 1same name
- poorvith-mp/skills-developer★ 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 repository8 other files it gives its agents.
- delegate-and-verifySkill
- evidence-audited-analysisSkill
- finish-the-turnSkill
- lessons-ledgerSkill
- outcome-first-writingSkill
- plain-handoffSkill
- root-cause-firstSkill
- verified-doneSkill
The file
We split this skill into the rules it gives an agent
9 rules. About 0 are common, 9 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈9
Distinctive · estimate
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 8
Do the simplest thing that works well, and only the thing that was asked.
- Distinctive: line 13
No hypothetical futures: Do not design for requirements that do not exist yet.
- Distinctive: line 14
No impossible-scenario handling: Do not add error handling, fallbacks, or validation for situations that cannot happen.
- Distinctive: line 15
Change the code, not around it: Do not add feature flags, compatibility shims, wrapper layers, or "v2" copies when you can simply change …
- Distinctive: line 16
Match the surroundings: Write code that reads like the file it lives in: same naming, same idiom, same comment density.
- Distinctive: line 21
Before finishing, read your own diff hunk by hunk and ask of each one: *does the task fail without this?*
- Distinctive: line 23
If yes, keep it.
- Distinctive: line 24
If no, but it fixes something real and separate: revert it, and note the finding in your summary as a suggested follow-up.
- Distinctive: line 25
If no, and it is stylistic: revert it without comment.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Minimal Diff
Do the simplest thing that works well, and only the thing that was asked. The size of your diff should be proportional to the size of the request.
The rules
- No unrequested work. Do not add features, refactor, or introduce abstractions beyond what the task requires. A bug fix does not need surrounding cleanup. A one-shot operation does not need a helper function. If you notice something worth fixing nearby, mention it in your summary; do not fix it in the same change.
- No hypothetical futures. Do not design for requirements that do not exist yet. Avoid premature abstraction and half-finished generalization. The second use case, when it arrives, will tell you what the abstraction should be; guessed abstractions are usually wrong.
- No impossible-scenario handling. Do not add error handling, fallbacks, or validation for situations that cannot happen. Trust internal code and framework guarantees. Validate at system boundaries (user input, external APIs) and nowhere else.
- Change the code, not around it. Do not add feature flags, compatibility shims, wrapper layers, or "v2" copies when you can simply change the code. Dead paths and parallel versions are debt from the moment they land.
- Match the surroundings. Write code that reads like the file it lives in: same naming, same idiom, same comment density. A correct change in a foreign style is still a costly change.
- Comments state constraints, not narration. Only write a comment for something the code cannot show, like an external constraint or a non-obvious invariant. Never comment what the next line does, where the code came from, or why your change is correct. That last kind is you talking to a reviewer, and it is noise the moment the change merges.
The pre-submit check
Before finishing, read your own diff hunk by hunk and ask of each one: does the task fail without this?
- If yes, keep it.
- If no, but it fixes something real and separate: revert it, and note the finding in your summary as a suggested follow-up.
- If no, and it is stylistic: revert it without comment.
Why this matters
Every line you touch is a line someone must review, a line that can conflict, and a line that can break. Reviewers trust agents whose diffs contain exactly what was asked. The fastest way to lose that trust is a ten-line fix arriving inside a two-hundred-line cleanup.
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.

