Loading...

Let AI answer from your pages

The same list is available to AI agents over MCP, with nothing extra to build.

What MCP gives you

MCP (Model Context Protocol) is the standard way an AI agent plugs into an outside tool. Search Stack runs an MCP server, so any MCP-capable agent — Claude, or your own — can use your reference list as a retrieval tool: it searches your pages and answers from what it finds rather than guessing. It's the same list you just built — nothing re-imported, no second copy of your data.

Connect an agent

Point an MCP client at the Search Stack MCP server. In an agent's MCP configuration that's one server entry:

{
  "mcpServers": {
    "searchstack": {
      "url": "https://mcp.searchstack.dev/",
      "headers": { "X-API-Key": "YOUR_READ_ONLY_KEY" }
    }
  }
}

Use a read-only key (search-result:read) scoped to the reference list, so the agent can search these pages and nothing else. It then gains Search Stack's search tools:

You: How do I change a field without breaking apps that already use my search?

Agent: Renaming or removing a field creates a new version, while the old version keeps working, so existing apps aren't affected… — drawn from your Versioning page, which the agent found by searching your reference list.

The url on each result is the citation, so an answer traces back to a real section of a real page. The console Playground is an agent reading your account's data over this same MCP server.

Go deeper: The Playground — an agent reading your data over MCP in the reference.

Top