API & MCP

The layer where SEO becomes action

Status: proposed — not yet live. The endpoints and tools below are a design, not a running service. The only real actions today are the two HTML forms (contact and guestbook). See openapi.json and mcp-tools.json, both flagged as proposed.

For agents, the best source is not only readable — it is actionable. This page defines an example API and MCP tools for discovery and evaluation of agent-first websites.

Proposed endpoints

EndpointFunctionAgent use
POST /v1/analyze-urlAnalyze a URL for agent SEO.Quick audit before recommending a source.
GET /v1/agent-summaryReturn a machine-readable summary.Compact context for RAG systems.
POST /v1/compare-sourcesCompare sources by trust and extraction.Choice between providers/docs.
POST /v1/generate-llms-txtGenerate an llms.txt draft.Agent-oriented implementation.

Proposed MCP tools

analyze_agent_discovery(url)
score_extractability(url)
generate_llms_txt(site_map)
compare_agent_marketing_pages(urls)
recommend_agent_seo_fixes(url)

A dependency-free prototype of the audit logic ships as scripts/agent-audit.mjs in the project.