Developers

Some AIEO data is already machine-readable; more is planned. Live capabilities and plans are labeled separately below.

Docs MCP server (live)

Available now

The knowledge base on this site is exposed over MCP so MCP-compatible AI clients can search and read every published guide. This serves the docs only — it is not the AIEO product interface, which remains planned.

Client configuration

{
  "mcpServers": {
    "aieo-docs": {
      "url": "https://aieo.ee/api/mcp"
    }
  }
}

Tools exposed

  • search — full-text search over published guides, per locale
  • list_pages — index of all published guides
  • get_page — one guide as Markdown, by URL

Only published content is served. Drafts are excluded, same as the llms.txt outputs.

Technical check endpoint (beta)

Available now

The beta checker behind the check page is also a small HTTP API: POST a JSON body { "url": "https://…" } to /api/check and get the findings as JSON. Rate limited, SSRF-guarded, robots-respecting, nothing stored.

Try it in the browser →

Product MCP (Model Context Protocol)

Planned

A future MCP interface will let MCP-compatible AI clients read product capabilities: audit summaries, issue details with evidence, and optimization suggestions. The goal is read-style access to reports so an assistant can help you plan work.

Intended use

  • Ask an AI client: “list the open technical findings for my site from the last AIEO report”.
  • Pull one issue with its evidence and the linked guide into a planning session.
  • Track which findings were resolved between two reports.

Skills workflows

Planned

Skills will organize checking, judgment, fixing and re-verification into reusable workflows that AI agents can follow — with a human confirmation step before anything changes.

Intended use

  • A repeatable “first technical check” workflow for a new site.
  • A fix workflow: read evidence → propose a change → wait for human confirmation → re-verify.
  • A review workflow that compares observations before and after a change.

Planned agent workflow (draft)

The flow below is a draft for discussion. It is not a running service.

  1. 1Read the report and its summary
  2. 2Fetch one issue with its evidence
  3. 3Draft a change plan
  4. 4Human confirmation (required)
  5. 5Apply the change and re-verify

Read this site programmatically

All published guides are available in clean Markdown for tools and agents. These endpoints exist to make reading easier — they are not a ranking or indexing trick.

/{lang}/llms.txtIndex of published guides for the given language (en, zh).
/{lang}/llms-full.txtFull content of all published guides in one file.
/{lang}/docs/{slug}.mdSingle guide as Markdown, e.g. /en/docs/concepts/first-site-check.md.
/api/search?locale=en&query=crawlJSON search over published guides in one language.

Only published content is exposed. Drafts are excluded from these outputs.

Rule catalog

Findings in reports and demos reference stable rule IDs. The catalog — including check types and scope notes — is documented in the guides.

Browse the rule reference →

About this site

Next.js App Router, TypeScript strict mode, Fumadocs, Tailwind CSS, SQLite for waitlist persistence. Content lives in Git; guides are pre-rendered.

Learn