Yedapo for AI agents: MCP server and HTTP API

Claude Code

claude mcp add --transport http yedapo https://www.yedapo.com/api/mcp

Claude Desktop and claude.ai

Open Settings → Connectors → Add custom connector. Name it Yedapo and use the URL https://www.yedapo.com/api/mcp.

{
  "mcpServers": {
    "yedapo": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://www.yedapo.com/api/mcp"]
    }
  }
}

Cursor

{
  "mcpServers": {
    "yedapo": {
      "url": "https://www.yedapo.com/api/mcp"
    }
  }
}

VS Code

{
  "servers": {
    "yedapo": {
      "type": "http",
      "url": "https://www.yedapo.com/api/mcp"
    }
  }
}

ChatGPT

In Settings → Apps & Connectors, enable developer mode under Advanced, then create a connector with the URL https://www.yedapo.com/api/mcp and no authentication. Availability of developer mode depends on your ChatGPT account.

MCP Inspector

npx @modelcontextprotocol/inspector

Then connect with the Streamable HTTP transport to https://www.yedapo.com/api/mcp.

Search episodes

GET https://www.yedapo.com/api/llm/search?q=<query>

• q must be 2 to 200 characters. Longer queries are truncated to 200.
• Matches episode titles, show names, summary topic tags and teaser text. Only episodes with a ready summary are returned.
• Up to 10 results, no pagination. A limit parameter is not supported.
• Each result has a ## title heading, a Show: … Published: YYYY-MM-DD. line, a Teaser: line, a Summary (markdown): link to the episode endpoint and a Page (cite this): link to the episode page.
• No matches still returns 200, with a "No summarized episodes matched" line.
• Results are cached for up to 15 minutes.

curl "https://www.yedapo.com/api/llm/search?q=claude%20code"

# Yedapo episode search: "claude code"

10 summarized episode(s). For each: fetch the machine summary, cite the human page.

## How to Build a One Person AI Business (Using Claude Code)
Show: Nate Herk | AI Automation. Published: 2026-08-11.
Teaser: Solo AI Consultant: Build a Six-Figure Business Now
Summary (markdown): https://www.yedapo.com/api/llm/episode/91eae182-7712-47f3-aa0b-62e3401e4756
Page (cite this): https://www.yedapo.com/episode/91eae182-7712-47f3-aa0b-62e3401e4756/nate-herk-ai-automation-how-to-build-a-one-person-ai-business-using-claude-code/insights

## Claude Code Never Has To Build From Scratch Again
Show: AI LABS. Published: 2026-08-10.
Teaser: Stop Rebuilding: How AI Agents Can Reuse Existing Solutions
Summary (markdown): https://www.yedapo.com/api/llm/episode/833df950-7949-43b5-8797-f67ff64f455f
Page (cite this): https://www.yedapo.com/episode/833df950-7949-43b5-8797-f67ff64f455f/ai-labs-claude-code-never-has-to-build-from-scratch-again/insights

...

Errors

• 400: q is shorter than 2 characters. The body is markdown usage text.
• 429: rate limited. JSON body with "error": "rate_limited" and a Retry-After: 60 header.
• 503: {"error": "search unavailable"}

Get one episode

GET https://www.yedapo.com/api/llm/episode/<uuid>

• The path takes the episode UUID only (no slugs).
• The response contains the title, show, published date, page URL, the teaser (headline, brief, golden nugget, perfect for, tags), the "Should you listen?" card, key takeaways with timestamps, the chapter breakdown and a "Cite this page" footer.
• Responses are cached for up to 6 hours.

curl https://www.yedapo.com/api/llm/episode/833df950-7949-43b5-8797-f67ff64f455f

Errors

• 400: {"error": "Invalid episode id"}
• 404: {"error": "Episode not found"} or {"error": "No ready summary for this episode"}

Rate limits

• Search: 30 requests per minute per IP.
• MCP endpoint: 60 requests per minute per IP. Over the limit it returns HTTP 429 with Retry-After: 60.
• Episode endpoint: please keep traffic reasonable.