| name | github-repo-seo |
| description | Improve GitHub and open-source repository SEO/discoverability across GitHub metadata, README/docs, package registries, GitHub Pages or docs sites, Google Search Console, and release verification. Use when the user asks why a repo/package cannot be found on Google, npm, crates.io, PyPI, etc.; asks to do SEO for a GitHub repo; needs README/package metadata optimization; wants Search Console submission; needs branded package scope planning for public repos; or wants evidence-backed repo/package/site discoverability audits. |
GitHub Repo SEO
Use this skill to make a public repository findable as a product, not just correct as code. Treat SEO as five connected surfaces: GitHub repo metadata, README/docs content, package registry metadata, public site crawlability, and external search/indexing.
Keep claims evidence-based. Separate locally healthy, merged on GitHub, published to registries, and indexed by Google; they move on different clocks.
Workflow
- Start from current state. Run
git status --short --branch and fetch/sync before calling anything latest.
- Run the baseline collector:
python3 ../../scripts/repo_seo_baseline.py --root . --json
- If a public site exists, run:
python3 ../../scripts/site_meta_audit.py <homepage> --json
- Search exact brand, repo, package, and docs-site names when search visibility is the question.
- Record repo, docs site, package registry, and release discoverability separately.
- Read
../../references/data-sources.md before recommending paid APIs or MCP providers.
Check Surfaces
- GitHub About description, homepage, topics, social preview, releases, and community profile files.
- README first screen: name, what it does, who it is for, install command, demo/proof, and authoritative links.
- Package metadata: npm, crates.io, PyPI, Homebrew or other actual distribution channels.
- Docs site: title, meta description, canonical, OG/Twitter cards, JSON-LD, robots, sitemap, and crawlability.
- Search Console: verification, sitemap submission, and indexing requests only when credentials/browser access are available.
Rules
- Do not keyword-stuff descriptions, topics, package keywords, or README headings.
- Do not claim Google has indexed a page immediately after sitemap submission.
- Do not treat a successful crawl as proof of ranking.
- Do not force unrelated projects under one package scope.
- Do not expose tokens. Keep npm/GitHub/provider credentials outside reports and commits.
Decision Table
| Symptom | Likely cause | Action |
|---|
| Google cannot find the repo/site | No crawlable site, weak README text, new page, missing sitemap | Improve README/site metadata, verify robots/sitemap, submit Search Console |
| Package is published but not discoverable | Weak name/description/keywords/readme | Improve package metadata and republish a patch release |
| Branded search shows another product | Name collision, weak entity signals, no disambiguation | Add brand qualifiers, sameAs links, and a disambiguation page or issue |
| Site checks pass but ranking is unknown | No SERP/rank data | Use Search Console for owned properties or DataForSEO/SE Ranking/Ahrefs for SERP evidence |
Done Criteria
Finish with changed files or issue links, exact verification commands and outcomes, public URLs, and remaining external/manual blockers.
