project-discovery
Skill2 starsChanged 56 days agoMITView on GitHub
Use before editing anything in a repo or asking how a project wants contributions. Finds upstream's own rules.
6 other projects use a version of this.
Adopted by
Most recently taken up by
- upex-galaxy/agentic-qa-boilerplatetheir own version, first seen Nov 7, 2026
- NAVANEETHVVINOD/agent-orchestratortheir own version, first seen Oct 3, 2026
- bhagyamudgal/skillstheir own version, first seen Sep 9, 2026
- ntaffzii/Skill-Agentstheir own version, first seen Sep 2, 2026
- taipm/setup-c2their own version, first seen Apr 12, 2026
- kognar-ai/claude-skillstheir own version, first seen Jan 20, 2026
Other versions6 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 6
- upex-galaxy/agentic-qa-boilerplate★ 26same name
- ntaffzii/Skill-Agents★ 4same name
- bhagyamudgal/skills★ 2same name
- taipm/setup-c2★ 1same name
- kognar-ai/claude-skills★ 0same name
- NAVANEETHVVINOD/agent-orchestrator★ 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 repository20 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- artifact-handoffSkill
- catalogSkill
- changelog-qualitySkill
- change-splittingSkill
- code-structureSkill
- comment-qualitySkill
- evidence-disciplineSkill
- factual-integritySkill
- language-style-samplingSkill
- llm-tellsSkill
- plan-iteration-gateSkill
- pre-commit-checkSkill
- repo-hygieneSkill
- self-review-gateSkill
- skill-authoringSkill
- socratic-specSkill
- systematic-debuggingSkill
- upstream-hygieneSkill
The file
We split this skill into the rules it gives an agent
3 rules. 0 are common, 1 is distinctive, and 2 are very specific to this project.
0
Common
Most of the 5 edited versions we read keep these.
1
Distinctive
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 16
Run before any code edit, changelog draft, or PR.
- Very specific: line 75
Check: git log -p -3 -- <path> beside your own diff — do the comment shape and message shape look like they came from the same project?
- Very specific: line 87
Create .hermes/plans/<date>_<slug>/discovery.md or docs/discovery/<project>.md with:
“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
Project Discovery — Step 0
Run before any code edit, changelog draft, or PR.
Search in order (first hit wins — keep first that exists)
CONTRIBUTING.md/CONTRIBUTING.rst/CONTRIBUTINGat repo root.github/CONTRIBUTING.mddocs/CONTRIBUTING.md/doc/contributing.*HACKING/HACKING.md/ps/HACKINGCLAUDE.md/AGENTS.md/.cursor/rules//.github/copilot-instructions.md(agent contributor notes if repo ships them)README.md/README.rstsection "Contributing" / "Pull Requests" / "How to Contribute"src/CONTRIBUTING(e.g. Exim)- Fallback, and always worth the minute: sample the project's own changesets.
git log --oneline -20 -- <paths you will modify>, thengit showa handful of them, ideally by different authors. Written guides rarely cover comment and message idiom, so read it off the history and mimic what upstream does.
Command
ls -1 CONTRIBUTING* 2>/dev/null; ls -1 .github/CONTRIBUTING* 2>/dev/null; \
ls -1 doc/contributing* docs/CONTRIBUTING* 2>/dev/null; \
ls -1 src/CONTRIBUTING 2>/dev/null; \
ls -1 HACKING* ps/HACKING 2>/dev/null; \
ls -1 CLAUDE.md AGENTS.md 2>/dev/null; \
grep -i -n "contribut\|pull.request\|coding.style\|DCO\|sign.*off" README.md README.rst 2>/dev/null | head -n 30What to extract into spec/plan
- DCO / Signed-off-by required?
- Subject convention (imperative ≤50? conventional commits? scope prefix?)
- Trailer style (
Fixes:,Closes:,Co-authored-by:public name no email vs with?) - Style gates run in CI (formatter, linter,
checkpatch-like) - PR template / issue template in
.github/ - External doc pointer (e.g. devguide.python.org) — follow external doc.
- Agent-specific rules (
CLAUDE.md) that override generic style at file level?
Match the project's idiom — comments and changelog
Fit the change to the codebase you are touching: style, terminology, and length of both comments and the commit message.
- Sample the changesets of the files you will modify, not the repo's whole history — idiom differs by subsystem (driver vs core, tests vs docs, hand-written vs generated).
- Match three things: style (WHY block vs no comment, prose vs bullets, tense and voice), terminology (use the project's own names for its concepts; a synonym reads as a second concept), and length (the paragraph and subject caps this project actually uses — they refine the generic defaults here, they do not replace them).
- Style and terminology come off 2-3 sampled changesets — they are categorical,
so a few samples settle them. Length does not: it is continuous and follows
the size and kind of the change, so read the file's actual envelope
(
git log --format=%B -20 -- <path>) and include one sample of a change like yours. Samples that disagree with each other mark a file mid-transition — sample up. - Written rules win where they exist and are current; when practice diverges from a written rule, follow the written rule and say so in the message.
- Check:
git log -p -3 -- <path>beside your own diff — do the comment shape and message shape look like they came from the same project? - Rules move: never work from a remembered or cached copy of this project's conventions — re-read the guide and re-sample the file history at the start of each task. A convention recalled from an earlier session, or copied out of a summary, may have been superseded since; the current source is the only authority, and stale rules are followed confidently.
Worked example: docs/references/changeset-idiom-example.md.
Write down
Create .hermes/plans/<date>_<slug>/discovery.md or docs/discovery/<project>.md with:
- Which file found (path)
- 5-10 bullet summary of must-obey rules
- Link to source file line range
- Anything ambiguous → open Q for human
When to skip
Trivial one-line fix where recent git log --oneline -10 subject pattern is obvious and repo has no CONTRIBUTING — still note that you looked and found none.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
Reviews can't be read right now.
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.

