context-pack
Skill2 starsChanged 7 days agoMITView on GitHub
Writes one small digest of a change — the diff, the files it touches, the API contracts and the test entry points — so review agents read that instead of re-exploring the repository. Use before a review, or when several agents would otherwise re-read the same files.
4 other projects use a version of this.
Adopted by
Most recently taken up by
- oocx/tfplan2mdtheir own version, first seen Oct 8, 2026
- TIKAZI/TIKAZ-Codex-Context-Economytheir own version, first seen Aug 31, 2026
- Mesteriis/Engineering-Bible-AItheir own version, first seen Jun 28, 2026
- kambleakash0/agent-skillstheir own version, first seen Apr 6, 2026
Other versions4 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 4
- oocx/tfplan2md★ 174same name
- kambleakash0/agent-skills★ 10same name
- TIKAZI/TIKAZ-Codex-Context-Economy★ 3same name
- Mesteriis/Engineering-Bible-AI★ 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 repository18 other files it gives its agents.
- approve-planSkill
- check-impactSkill
- design-forgeSkill
- deslopSkill
- diagnose-bugSkill
- falsifySkill
- forgeSkill
- gap-analysisSkill
- interrogateSkill
- make-fixtureSkill
- micro-loopSkill
- phased-loopSkill
- pick-tierSkill
- ship-planSkill
- simple-loopSkill
- taste-testSkill
- unslopSkill
- wait-whatSkill
The file
We split this skill into the rules it gives an agent
3 rules. 0 are common, 2 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 3 edited versions we read keep these.
2
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 77
Every claim carries a file:line anchor, so a reviewer can open exactly one place.
- Distinctive: line 78
No file bodies. If a reviewer needs more, they open that file — the pack's job is to say where, not to replace the file.
- Distinctive: line 80
Write it once per change. If the branch moves, rewrite it rather than patching it.
“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
Context pack
Eight cold-start agents re-reading the same seven files cost 39.5M cache-read tokens in a traced run — 94% of that run's price. The pack replaces the re-reading with one shared page.
Input
- The repo and the base ref (the same base the review uses).
- The plan folder, if there is one:
.plans/<task>/plan.md.
Build it
The diff, complete for the changed lines:
git diff "$BASE"...HEADDrop lockfiles, generated output and vendored directories. If the diff alone is over the budget, split it by file group and say which group is where.
- The files it touches, one line each: path, what the file is, and its role in the change. No file bodies.
- The API contracts the change crosses: endpoint, method, request shape, response shape, and where the shape is declared (the type or interface file, with a line number).
- The test entry points: the test files that cover the changed source files, and the command that runs them.
- The plan's acceptance criteria, verbatim, one per line, with any criterion marked
manual:flagged as outside automated verification. - What not to re-derive: the facts a reviewer would otherwise spend turns rediscovering — which field decides the behaviour, which component owns the call, what the previous behaviour was, and the known-broken things that are not part of this change.
Write it
Write .plans/<task>/context-pack.md:
# Context pack
BASE: origin/master
CHANGED: 7 files, +283/-32
## What this change does
<two or three lines>
## Files
- `components/button/StartClassButton.vue` — start button; emits `started` on a successful start
- ...
## API contracts
- `POST /api/program-batch/start/{batchCode}` → 204; success is what emits `started`
declared in `types/program-batch-student.d.ts:41`
## Test entry points
- `components/button/StartClassButton.test.ts` — `npx vitest run components/button/StartClassButton.test.ts`
- `pages/classes/_components/ClassesCard.test.ts` — `npx vitest run pages/classes/_components/ClassesCard.test.ts`
## Acceptance criteria
1. ...
2. manual: the list stops reporting NOT ATTEMPTED after a start (needs a real backend)
## Do not re-derive
- `batchStatus === "NOT ATTEMPTED"` is the signal; `courseStatus` is not.
- The modal's query key is built once at setup, so it does not follow a changed prop.
## Diff
```diff
<the diff>
## Rules
- **Budget: 3,000 tokens** (about 2,200 words). If it does not fit, cut prose before cutting the diff, and cut the diff by file group before dropping anything.
- Every claim carries a `file:line` anchor, so a reviewer can open exactly one place.
- No file bodies. If a reviewer needs more, they open that file — the pack's job is to say where, not to replace the file.
- The pack states the diff, the change's intent and the criteria. It does not argue that the change is correct; that is the reviewer's job.
- Write it once per change. If the branch moves, rewrite it rather than patching it.
Report one line: `Context pack: <path> (<n> words)`.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.

