dx-standards-discover
Skill8 starsChanged 4 months agoMITView on GitHub
Mine the project's real conventions from tooling, code, and docs and write them into the standards layer.
In this repository21 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- dx-archiveSkill
- dx-brainstormSkill
- dx-diagnoseSkill
- dx-distillSkill
- dx-domain-discoverSkill
- dx-domainSkill
- dx-frameSkill
- dx-implementSkill
- dx-initSkill
- dx-lessonSkill
- dx-newSkill
- dx-planSkill
- dx-refactor-discoverSkill
- dx-referencesSkill
- dx-researchSkill
- dx-reviewSkill
- dx-review-triageSkill
- dx-roadmapSkill
- dx-standards-updateSkill
- dx-tddSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
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 31
Stop. Do not chain into another skill.
≈ “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
dx-standards-discover
Mine this project's actual conventions — from tooling, recurring code, and existing docs — and write them as prescriptive standard files under context/standards/. Record only what the codebase genuinely evidences: no inherited opinions, no aspirational rules. This fills the frontend/, backend/, testing/ layers dx-init left empty.
Guard. context/standards/{frontend,backend,testing}/ must exist. If it is missing, tell the user to run /dx-init first. Never overwrite global/ — dx-init seeded it; only append a global file if a genuinely new project-wide rule surfaces that none of the seeds cover. With --from=PATH, mine that external codebase instead of the working tree.
1 — Read the fixed sources
Read the config that already encodes rules: linter + formatter configs, tsconfig/compiler settings, CI workflows, pre-commit hooks, package.json/build scripts, .editorconfig. Then read READMEs and any docs/ that state conventions. These are the strongest evidence — a rule in a linter config is enforced, not wished-for. Read foundation/glossary.md for naming (a one-line habit — no section).
2 — Mine code patterns by layer
Fan out one built-in Explore subagent per layer present (frontend, backend, testing). Each reports the recurring patterns actually in the code — component/module structure, API shape and error handling, state and data access, test structure and naming, file layout. A convention counts only when it recurs across several files; one example is not a standard. Skip any layer the project doesn't have.
3 — Write the standard files
Group findings by layer + topic into concise, prescriptive files (~20–30 lines each) — e.g. frontend/components.md, backend/api.md, testing/test-writing.md. Match the seeded style: a ## per rule, "do this" phrasing, a short example only when it clarifies. Every rule must trace to something you saw — name the config or pattern in a few words, and drop anything you can't back. Where config and code disagree, follow the enforced config and flag the drift as a lesson candidate — don't average the two.
Done when
The evidenced standards exist under context/standards/, and global/ is untouched (or minimally appended). Print a one-line-per-file summary of what was written (created/updated · layer/topic · what it captures), then:
Standards discovered: <n> files under context/standards/
Next: /dx-plan <id> — planning now matches these standards into the plan.Stop. Do not chain into another skill.
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.

