Add GEO / AI-discoverability skill (geo-opt)

[cortega26]

Feature request

I'd like to propose adding a GEO / AI-discoverability skill to the registry. Opening an issue first as CONTRIBUTING.md suggests, before sending the PR.

What

Add the geo-optimization skill from cortega26/geo-opt (published to npm as geo-opt). It's a local, source-available toolkit built on three pillars:

• GEO (Generative Engine Optimization) — content structure, evidence density, verified quotations, primary-source citations and entity clarity for AI answer engines.
• Structured data — generate/inject/validate Schema.org JSON-LD.
• Technical SEO — audit/generate robots.txt and sitemap.xml, plus llms.txt, hreflang/canonical and AI-crawler policy.

Why it's distinct from the existing SEO bonus

The registry already ships addyosmani/web-quality-skills/seo as a frontend bonus. This skill is narrower and complementary: it targets generative-engine discoverability (how content is parsed and cited by ChatGPT / Perplexity / Gemini), not just classic search ranking, and it acts on the concrete discoverability artifacts (llms.txt, robots.txt, sitemap.xml, JSON-LD) rather than page-level SEO heuristics. It's explicitly not a ranking/citation predictor.

Proposed detection + map entry

Detection triggers on AI-discoverability artifacts present at the project root:

{
  id: "geo",
  name: "AI Discoverability (GEO)",
  detect: {
    configFiles: ["llms.txt", "llms-full.txt", "robots.txt", "sitemap.xml"],
  },
  skills: ["cortega26/geo-opt/geo-optimization"],
},

The skill lives at .agents/skills/geo-optimization/SKILL.md in the source repo, so sync-skills.mjs resolves it correctly (the parent directory name matches the skill name, and .agents isn't in the skip list). It works via both the tarball/tree path and the sparse-clone fallback.

License

The source repo uses a custom source-available license (Tooltician Community License 1.0). As the author, I'm happy to grant permission to include and redistribute the skill's content in the autoskills registry under the project's terms, and to relicense the bundled SKILL.md (e.g. MIT) if that's preferred for the registry.

PR

I already have the PR ready — the skills-map.ts entry plus the materialized bundle under skills-registry/geo-optimization/ (mirroring the git-tracked source; the manifest/hashes would be regenerated by your sync:skills run so the OpenAI review is recorded on your side). Happy to open it whenever you're interested, or adjust the detection scope if you'd prefer something narrower (e.g. llms.txt only).

Thanks for the project!