evidence
Skill0 starsChanged 16 days agoMITView on GitHub
Always-on rule for every session — before a material technical recommendation, a change that rests on an uncertain assumption, or any claim that changed behaviour works, get the evidence that could prove it wrong. Scale the check to the risk; never for plain prose edits.
5 other projects use a version of this.
Adopted by
Most recently taken up by
- pahlevikun/levistacktheir own version, first seen Nov 1, 2026
- adiutama/skillstheir own version, first seen Nov 1, 2026
- hoophq/hooptheir own version, first seen Oct 17, 2026
- ltlongtma/solo-sdlctheir own version, first seen Oct 2, 2026
- SoFarSoGoodya/mathmodel-weavetheir own version, first seen Oct 2, 2026
Other versions5 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 5
- hoophq/hoop★ 848same name
- ltlongtma/solo-sdlc★ 2same name
- SoFarSoGoodya/mathmodel-weave★ 2same name
- adiutama/skills★ 1same name
- pahlevikun/levistack★ 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.
The file
We split this skill into the rules it gives an agent
2 rules. 0 are common, 2 are distinctive, and 0 are very specific to this project.
0
Common
Most of the 5 edited versions we read keep these.
2
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 9
Before "done": run the check that could fail — the test, the request, the page — and read what actually ran: counts, skips, the case you …
- Distinctive: line 10
A check that cannot fail proves nothing: When unsure it can see the bug, break the change on purpose once and watch it fail.
“Common” is measured: kept by at least 6 in 10 of the 5 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
Evidence before conclusions
- Say which is which: a fact (with where you saw it), an assumption (with the test that would disprove it), an observed result (with the command or page that produced it). (owner)
- Before "done": run the check that could fail — the test, the request, the page — and read what actually ran: counts, skips, the case you care about. A green exit with that case skipped proves nothing. (evidence)
- A check that cannot fail proves nothing. When unsure it can see the bug, break the change on purpose once and watch it fail. (evidence)
- Reports are claims: a subagent's, a reviewer's, a tool's, your own earlier message. Confirm what a conclusion rests on. A review finding is a hypothesis until reproduced or proven from the code. (evidence)
- External behaviour (a library, an API, a CLI) comes from its current docs for the installed version, not memory. Missing access is not a negative finding. (owner)
- Scale to risk: a lookup needs one direct check; money, identity, migrations, concurrency and public contracts need the failure path too. Never turn a prose edit into an audit. (owner)
- Close only what you established: name the scope you verified, the evidence, and what is still open. When a needed check is impossible, say so; that blocks an unqualified "done". (owner)
- This rule grants no permissions and never weakens a check to get green. (owner)
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.