Skip to content

starlight-intelligence/starlight-search-intelligence

v0.1.0MIT

Evidence-led SEO, AEO, GEO and AI-search visibility workflows with connected-provider discovery, source provenance, offline deterministic evaluation and reviewable implementation plans. Skills only; no hosted monitoring service.

GitHub

Starlight SEO & AI Visibility

SEO · AEO · GEO · AI Search Visibility | Starlight Growth Intelligence, portable skills surface

This private skills-only ChatGPT/Codex plugin provides 13 bounded workflows, including eight Semrush-aligned methods and original AI visibility measurement, source-backed prioritization, production patching and GenCreator handoffs. The plugin has no bundled Semrush API/MCP endpoint and no hosted scheduler. It uses separately connected, authenticated provider tools only when available.

Fast example

Audit the technical SEO and AI search visibility of example.com in the Netherlands. Verify Semrush access and state every data gap. Prioritize three evidence-backed opportunities; propose a reviewable GitHub PR, not a direct deployment.

Capabilities

  • skills/search-setup: onboarding, source and budget assessment
  • skills/seo-*: 8 independent Semrush-aligned search workflows
  • skills/geo-ai-visibility: observed-prompt citation measurement, not hallucinated rankings
  • skills/growth-opportunity-portfolio: deterministic evidence-preserving index
  • skills/seo-production-changes: PR-first website implementation
  • skills/gencreator-growth-handoff: creator/editorial distribution packet

Offline CLI (Python stdlib, no network)

python scripts/growth_core.py score skills/search-setup/references/examples/opportunities.json
python scripts/growth_core.py visibility skills/search-setup/references/examples/visibility.json
python scripts/growth_core.py semrush-rank skills/search-setup/references/examples/semrush_rank.csv --database us --observed-at 2026-10-27T13:00:00+00:00
python -m unittest discover -s scripts -p 'test_*.py' -v

Each output carries source data or a digest, input validation and explicit limitations. Input fixtures use synthetic data, except when your own user-authorized Semrush report is supplied. These scripts read existing local files only.

OSS provenance and licensing

Our skill adaptations and CLI are MIT licensed. Upstream projects are referenced rather than copied. Attribution, exact upstream revisions and an eight-workflow map: skills/search-setup/references/upstream-registry.md. Semrush official source code is Apache-2.0; Corey Haines and MarketingSkills SEO Operator are MIT. No external shell install commands are executed.

Lifecycle status

Initial capabilities: structured skills + deterministic offline analysis. No customer-facing dashboard, private production data adapter, public directory publication, always-on hosted monitoring, cross-platform AI scraping, or automatic publishing. Those require separately tested cloud services and user authorization.

The architecture contract is in skills/search-setup/references/architecture.md.