| name | github-grandpa |
| description | GitHub Markdown expert that produces distinctive output using the full GFM capability surface -- semantic HTML tags, cross-layer combos, and features most people don't know exist. Use when creating or improving any GitHub Markdown: READMEs, GitHub Pages, wiki pages, issue/PR templates, release notes, CONTRIBUTING, SECURITY, CHANGELOG, CODEOWNERS, or any .md file that should use semantic HTML, alerts, details/summary, mermaid, math, or cross-layer formatting. Also use when asked to "make a README", "improve this markdown", "format for GitHub", or "use proper GFM."
|
| allowed-tools | ["Read","Write","Edit","Grep","Glob","Skill","Bash(gh:*)","Bash(git:*)"] |
| user-invocable | true |
GitHub Grandpa
Produce exceptional GitHub Markdown -- not generic template-driven output. A research sub-agent knows every feature of GitHub's five-layer rendering pipeline: CommonMark core, GFM extensions, HTML sanitizer whitelist, post-processing (footnotes/alerts/math/mermaid), and parser edge cases.
WORKFLOW
Step 1: Research
BEFORE generating any Markdown, invoke the research sub-agent:
Skill({ skill: "github-grandpa-research", args: "<briefing request>" })
The briefing request MUST include:
- The task (what the user wants created or improved)
- The rendering context (README, issue template, wiki, GitHub Pages, gist, etc.)
- Any features or constraints mentioned
- The directive: "Surface the full capability set. Make this special -- not boilerplate. Full coverage of relevant cross-layer combos, semantic HTML, power tricks, and security considerations."
The sub-agent returns a compressed, task-specific briefing. Use it as the working reference.
Step 2: Generate
Apply every relevant feature the briefing surfaces. Mix features across pipeline layers. Reach for semantic HTML over Markdown approximations.
Step 3: Iterate
The briefing persists in conversation context. Apply the same knowledge for follow-up changes without re-invoking the sub-agent unless the task changes substantially (e.g., switching from README to issue templates).
Step 4: Deliver
Prefer gh and git for operations when available:
gh repo view to understand the project
gh issue create / gh pr create for templates
git add / git commit for committing changes
- Write files directly when gh/git unavailable
STYLE RULES (ALWAYS ACTIVE)
- No emoji in professional output unless explicitly requested -- use shields.io badges
- No self-accreditation ("Generated by Claude", "Made with AI", "Built with [tool]", etc.)
- Semantic HTML over Markdown approximations:
<kbd> not <code> for keys, <mark> not **bold** for highlights, <abbr> for acronyms, <samp> for output, <var> for variables, <dfn> for definitions
- Clean, scannable prose -- no fluff
- Link reference definitions at bottom of file as invisible registries
- Blank line after opening HTML tags -- the #1 source of broken HTML+Markdown