For developers

Connect your AI assistant

Connect our MCP server and your assistant can search and browse every listing in natural language — “find me an agent that qualifies inbound leads” — and read the details, right in your chat. Read-only, no account, no key.

What your assistant can do once connected

Search listings in plain language (search_agents).
Open a listing’s full public details (get_agent).
Browse by category (list_agents, list_categories).

It cannot buy, subscribe, publish, or access anything private — it’s a read-only window on the public catalogue. Buying still happens here on the website.

Add it to your client

Any client that supports remote (HTTP) MCP servers — e.g. Cursor, and Claude Desktop via Connectors — add:

{
  "mcpServers": {
    "buy-my-agent": {
      "type": "http",
      "url": "https://www.buy-my-agent.com/api/mcp?ref=connect"
    }
  }
}

Stdio-only clients (older Claude Desktop and similar) — bridge to the remote server with mcp-remote:

{
  "mcpServers": {
    "buy-my-agent": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://www.buy-my-agent.com/api/mcp?ref=connect"]
    }
  }
}

Restart your client, then ask: “Search Buy My Agent for an agent that follows up with cold leads.” Try the built-in prompts too — find-agent, compare-agents, and browse-category.

Also listed on

The official MCP Registry (io.github.BUYMYAGENT-TECHNOLOGIES-INC/marketplace), Smithery, and other MCP directories — so many clients can add it in one click from their own directory.

Building an agent yourself?

List it on Buy My Agent — free to list, you keep the customer relationship, and we take 10% only on a sale. Prefer the machine surface? See /llms.txt and the capability manifest.