Query the directory from your agent (MCP + REST)

AI Agent Pages is built to be used by agents, not only browsed by humans. If you run Cursor, Claude Desktop, or your own MCP client, you can search curated listings, read machine specs, and inspect readiness breakdowns without scraping HTML.

Why query instead of crawl?

Our directory is curated: submissions are scanned, then approved by a human. Each public listing exposes a JSON spec and a per-listing llms.txt. That is higher signal than guessing from search snippets—especially when your agent recommends tools to a user.

Connect in three steps

1. Register at /connect and create a delegation token (aap_…) if you need write tools later.
2. Point your MCP client at https://www.aiagentpages.com/mcp (see the server card).
3. Try the walkthrough on Agent Tool Finder.

Public tools (no token)

These work for read-only discovery:

• search_listings — text and category filters, minimum readiness score
• search_commerce_listings — stores; optional UCP-ready filter
• get_listing / get_listing_spec — canonical JSON for a slug
• get_readiness_report — probe-level breakdown (same signals we show on listing pages)

Example prompts:

• “Search AI Agent Pages for SaaS listings with readiness ≥ 80.”
• “Get the listing spec for {slug} and summarize agent endpoints.”
• “Find UCP-ready e-commerce listings in the directory.”

REST mirror

Prefer HTTP? See /developers:

• GET /api/v1/listings
• GET /api/v1/listings/{slug}

Same data model as MCP reads—pick whichever fits your stack.

Human-readable discovery files

We also publish:

• /llms.txt and /AGENTS.md for this site
• /listings/{slug}/llms.txt for each approved listing

List your site

When your origin ships llms.txt, MCP cards, or related discovery specs, submit for review. We score against site type (content, product, agent-native, or store) so marketing sites are not penalized for missing commerce protocols they will never use.

Questions or integration feedback? Reply via contact paths on the site or open a thread where you found us—we are early and actively improving the agent surface.