kit-strip-diff
Skill6 starsChanged 27 days agoMITView on GitHub
Remove extras from a finished diff. Use when the outcome already works and the diff still has unused additions, extra instruction lines, format-only hunks, or rare-case handling the outcome does not need.
In this repository20 other files it gives its agents.
- AGENTS.mdAGENTS.md
- rules/AGENTS.mdAGENTS.md
- kit-audit-metricsSkill
- kit-audit-skillsSkill
- kit-capture-lessonSkill
- kit-challenge-designSkill
- kit-check-blastSkill
- kit-create-prSkill
- kit-debate-decision-treeSkill
- kit-debug-prodSkill
- kit-decisionsSkill
- kit-evolveSkill
- kit-explainSkill
- kit-learn-repoSkill
- kit-minimizeSkill
- kit-optimize-contextSkill
- kit-planSkill
- kit-proveSkill
- kit-reviewSkill
- kit-triage-commentsSkill
The file
We split this skill into the rules it gives an agent
12 rules. About 0 are common, 6 are distinctive, and 6 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈6
Distinctive · estimate
This author's own choices. Worth a look.
6
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 13
Not choosing a smaller path before coding (kit-minimize)
- Very specific: line 14
Not proving the change works (kit-prove)
- Very specific: line 15
Not a bug or design review (kit-review)
- Very specific: line 16
Not checking what else the change could break (kit-check-blast)
- Very specific: line 17
Not a token audit of what loads before the user types (kit-optimize-context)
- Distinctive: line 18
Not moving working logic from a shared file to the one caller
- Very specific: line 19
Not the always-on habit while typing (kit-keep-diff)
- Distinctive: line 23
Default: uncommitted changes, staged and unstaged.
- Distinctive: line 27
Read the outcome and the surrounding code.
- Distinctive: line 34
Leave a change the outcome needs, even when it looks large.
- Distinctive: line 40
Stop. Wait for all, a subset, or no.
- Distinctive: line 48
Check result, or that you could not run one
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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
Strip diff
The outcome already works. Cut what this diff does not need. Ask before you delete or restore anything.
What this is NOT
- Not choosing a smaller path before coding (
kit-minimize) - Not proving the change works (
kit-prove) - Not a bug or design review (
kit-review) - Not checking what else the change could break (
kit-check-blast) - Not a token audit of what loads before the user types (
kit-optimize-context) - Not moving working logic from a shared file to the one caller
- Not the always-on habit while typing (
kit-keep-diff)
Scope
Default: uncommitted changes, staged and unstaged. If they name a branch, a commit range, or a path, use that. If the boundary is unclear, ask once.
What to cut
Read the outcome and the surrounding code. Flag a hunk only when you can show it is extra.
- Unused addition. A symbol, type, or file this diff added that nothing calls. Deleting it leaves behavior the same.
- Extra instruction. A new or edited line in a skill, rule, or agent doc that repeats a line already there, or that a competent model would do without being told. Cut that line only.
- Format-only hunk. Order, wrap, or whitespace changed, and the meaning matches the base version. Restore the base text. If you cannot show the match, leave the hunk.
- Rare case. A branch or helper this diff added for a case the agreed outcome does not need. Deleting it leaves that outcome the same.
Leave a change the outcome needs, even when it looks large. A clean hunk gets no finding.
Flow
- State the scope in one line.
- List findings: file, what is extra, which of the four kinds, and the cut.
- Stop. Wait for all, a subset, or no.
- Apply only the approved cuts. Do not rewrite working behavior.
- Re-run the check the repo already uses for the files you touched. If it fails, undo that cut.
Hand back
- What you removed, by kind
- Lines removed and added
- Check result, or that you could not run one
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.

