agentsclimarketplace

Contributor setup

Skill emaarco/hogwarts/plugins/felix-felicis/skills/contributor-setup

A magical place where my skills, rules, and plugins for AI agents are defined, which magically boost productivity 🏰πŸͺ„

Install
npx -y skills add emaarco/hogwarts --skill contributor-setup

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Analyzes a repo's contributor experience and creates or updates what's missing: GitHub issue-form templates (bug/feature/refactor), an open-source target-group-focused README, CONTRIBUTING.md, and the remaining community-health files (PR template, CODE_OF_CONDUCT, SECURITY, LICENSE, CODEOWNERS). Use when asked to make a repo contributor-friendly, open-source ready, or to set up issue templates / README / CONTRIBUTING.

SKILL.md

7.5 KB, as published. Nobody here has run it

Skill: contributor-setup

Audits everything a first-time visitor or contributor touches β€” README, issue templates, CONTRIBUTING.md, and the surrounding community-health files β€” then creates what's missing and upgrades what falls short. Analysis always comes first: never overwrite an existing file without showing the user what's wrong with it.

Run this when asked to "make this repo contributor-friendly", prepare a project for open-sourcing, set up issue templates, or as the community slice of a repo maturity audit.

Phase 1 β€” Analyze

Start from GitHub's own community-profile scoring, then inspect each file locally:

gh api repos/{owner}/{repo}/community/profile --jq '{health_percentage, files}'
ls .github/ISSUE_TEMPLATE/ 2>/dev/null
ls .github/PULL_REQUEST_TEMPLATE.md .github/CODEOWNERS CONTRIBUTING.md CODE_OF_CONDUCT.md SECURITY.md LICENSE README.md 2>/dev/null

Then judge quality, not just existence. Before writing anything, understand the project well enough to describe it accurately: read the manifest (package.json / pom.xml / …), the entry points, and any existing docs β€” the README must reflect what the project actually does, not a generic template.

README β€” evaluate against the target group (the developer who lands on the repo knowing nothing):

  • Answers what is this and why should I care in the first screen (one-paragraph pitch, not an architecture essay)
  • States the added value: which problem it solves, for whom, and why to pick it over the obvious alternatives β€” features alone don't answer this
  • Install / quickstart that works copy-paste
  • A "what you get" section β€” features, screenshot/GIF for anything visual
  • Badges that carry information (CI status, version, license) β€” not badge walls
  • Links to CONTRIBUTING.md and LICENSE near the end
  • Reference shape (from wardley-maps-modeler): pitch β†’ Install β†’ What you get β†’ structure/packages β†’ Contributing β†’ License

Issue templates β€” issue forms (.yml with a body: of typed fields), not legacy .md templates:

  • .github/ISSUE_TEMPLATE/ with at least a bug report and a feature request; a refactor template for code-heavy projects
  • Each form: name, description, title prefix ('[Fix]: ', '[Feat]: '), labels, and required: true on the fields that make an issue actionable (bug: description + steps to reproduce + expected + actual; feature: summary + motivation)
  • config.yml present (decide blank_issues_enabled with the user; add contact_links for discussions/security if applicable)

CONTRIBUTING.md β€” must let a stranger go from clone to merged PR without asking anything:

  • Setup and inner loop as copy-paste commands (install, build, test, lint)
  • What pre-commit hooks/CI actually enforce vs. what the contributor must run themselves
  • Commit convention (e.g. Conventional Commits) with concrete examples
  • PR process: branch naming, review expectations, where to ask questions
  • Project-specific gotchas (build-before-test, version pins that must move together, …)

Everything else that matters:

  • LICENSE β€” without it the project isn't legally usable; ask the user which (MIT default suggestion)
  • .github/PULL_REQUEST_TEMPLATE.md β€” short: summary, linked issue, checklist mirroring CONTRIBUTING requirements
  • CODE_OF_CONDUCT.md β€” Contributor Covenant is the standard; needs a real contact address
  • SECURITY.md β€” how to report vulnerabilities privately (GitHub private vulnerability reporting or an email)
  • .github/CODEOWNERS β€” default * owner so PRs auto-request review (also required by the sibling skill dependabot-setup)
  • Repo metadata via gh repo view β€” description and topics set; issues enabled

Phase 2 β€” Report

Summarize as a table before touching anything:

ItemStatusFinding
READMEβœ… / ⚠️ needs work / ❌ missinge.g. "no quickstart; assumes internal context"
Issue templates…e.g. "legacy .md templates; no required fields"
CONTRIBUTING.md…
LICENSE / CoC / SECURITY / PR template / CODEOWNERS…

Propose a prioritized fix list and confirm scope with the user (AskUserQuestion) β€” especially: license choice, blank issues yes/no, CoC contact address, and who owns CODEOWNERS.

Phase 3 β€” Create or update

Work through the confirmed list. Rules:

  • Update > replace. For existing files, fix the specific gaps from Phase 1; keep the author's voice and any project-specific content.
  • Ground every claim. Commands in README/CONTRIBUTING must be the repo's real scripts (read package.json scripts, Makefile, CI workflow) β€” run the quickstart yourself if possible; never invent commands.
  • Write for the target group. README sells the project to a potential user first, contributor second. CONTRIBUTING is all mechanics. Don't duplicate content between them β€” link.

Issue templates β€” ready-to-use forms ship with this skill in its reference/ folder (next to this SKILL.md):

  • reference/fix.yml β€” bug report (label bug)
  • reference/feat.yml β€” feature request (label enhancement)
  • reference/refactor.yml β€” refactoring proposal (label refactor)
  • reference/config.yml β€” issue chooser config (blank-issues toggle, optional contact links)

Copy them to .github/ISSUE_TEMPLATE/, replace the <project>/<owner> placeholders with the real repo, and adjust fields only where the project genuinely differs. For config.yml, settle blank_issues_enabled and any contact_links with the user per the Phase 1 checklist.

Ensure the labels exist β€” a form that references a missing label opens the issue silently unlabeled. Check and create what's missing:

gh label list --json name --jq '.[].name'
# default repos ship bug + enhancement, but usually not refactor:
gh label create refactor --description "Code refactoring or structural improvement" --color 6f42c1

Phase 4 β€” Verify

  • Issue forms: valid YAML and valid schema β€” GitHub silently falls back to a blank issue on schema errors, so check field types against the syntax reference
  • Every command written into README/CONTRIBUTING was executed or verified against the repo's actual scripts
  • Re-run gh api repos/{owner}/{repo}/community/profile after pushing to confirm the health percentage improved
  • Internal links resolve (CONTRIBUTING ↔ README ↔ LICENSE)

Sources

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.