GitHub Presence โ€” Quality framework for GitHub repos

GitHub Presence

Quality framework for GitHub repos. Type-specific checklists, a design playbook, and a review skill for Claude Code.

Phase 1-3 5 Project Types MIT License

Every repo starts from scratch

There's no shared standard for what a CLI tool README needs vs. a learning path vs. a checklist project. Quality depends on the day, not on the type.

Without vs. With

Without

  • Ad-hoc READMEs, whatever feels right
  • Quality is subjective and inconsistent
  • Manual review, easy to miss things
  • "I should probably add a FAQ..."

With GitHub Presence

  • Type-specific checklists (CLI โ‰  course โ‰  checklist)
  • Tiered scoring: Bronze โ†’ Silver โ†’ Gold โ†’ Platinum
  • Automated skill checks every criterion
  • "For Gold you need: FAQ, Contributing, Mobile"

Three components, one system

Everything you need to make your repos consistent and professional.

๐Ÿ”

Review Skill

Automated quality check via Claude Code. Detects project type, checks against tier-specific criteria, reports PASS/FAIL.

๐Ÿ“–

Playbook

24-section design knowledge base. Terminal mockups, dark themes, scroll animations, FAQ patterns โ€” everything for a professional landing page.

๐Ÿ“‹

Templates

5 README templates + landing page HTML template. Copy, fill placeholders, ship. One template per project type.

Five types, five checklists

Each project type has different needs. A CLI tool is not a course is not a checklist. The review skill detects your type automatically.

๐Ÿ”ง

CLI Tool

Build tools, proxies, linters. Terminal demos, install commands.

๐Ÿ“š

Learning Path

Tutorials, curricula. Chapter navigation, progress tracking.

๐Ÿ—

Framework

Templates, scaffolds. Architecture diagrams, starter vs. full.

โœ…

Checklist

Security guides, references. Interactive elements, printable.

โšก

Mini Tool

Scripts, proof of concepts. Minimal โ€” problem + run command.

Bronze to Platinum

Tiers are cumulative โ€” Gold includes everything from Bronze and Silver. A single FAIL means the tier is not reached.

๐Ÿฅ‰

Bronze

Problem + Install + Usage. Good enough for experiments.

๐Ÿฅˆ

Silver

+ Architecture, Config, Tests. For active projects.

๐Ÿฅ‡

Gold

+ FAQ, Contributing, Landing Page. Public release ready.

๐Ÿ’Ž

Platinum

+ Cross-references, API docs. Ecosystem component.

Three steps to better repos

1
Clone the repo
git clone https://github.com/janrummel/github-presence.git
2
Copy the skill to your Claude Code skills directory
cp -r github-presence/skill/ ~/.claude/skills/website-review/
3
Run /website-review in any repo

Navigate to any of your projects in Claude Code and type /website-review. The skill detects your project type, suggests a target tier, and reports PASS/FAIL for every criterion.

Common questions

What is this exactly?

A quality framework that defines what a good GitHub repo looks like for different project types (CLI tools, courses, frameworks, checklists, mini tools) and provides a Claude Code skill to automatically check your repos against those standards.

Do I need Claude Code?

For the automated review skill, yes. The playbook, project types, and checklists are useful on their own as reference material โ€” no tools needed.

Can I add my own project type?

Yes. Create a new checklist file in skill/checklists/ following the same format (criteria grouped by tier, with IDs). The skill will pick it up automatically.

What counts as Gold?

All Bronze + Silver + Gold criteria for your project type must PASS. One FAIL means the tier is not reached. The review skill shows exactly what's missing.

How do I contribute?

New checklists, playbook improvements, and templates are all welcome. Please open an issue before submitting large changes.

The story

Started with 6 repos that all looked different. Different README structures, different quality levels, no shared patterns. Every new project meant reinventing the wheel โ€” what sections to include, how to structure the landing page, what makes a good FAQ.

So I built a system. Defined 5 project types, created tier-based checklists, wrote a 24-section playbook with everything I learned about landing pages. Then automated the review with a Claude Code skill that checks every criterion and tells you exactly what's missing.

Now quality is consistent and measurable. Not perfect โ€” measurable. And that changes how you think about your repos.