🤖 Bring Your Own AI

Create a custom GPT that can search Yard transcripts for you.

How to set it up:

1. Go to ChatGPT GPT Editor
2. Click "Create new GPT"
3. Go to the "Configure" tab
4. Scroll down to "Actions" and click "Create new action"
5. Click "Import from URL" and enter:

https://yard-search.com/openapi.json

ChatGPT will import our API spec and you can then ask it questions like "Search for episodes about a topic" or "Find memorable stories from the show."

Our search API is open and free to use. No API key required.

Search Transcripts

GET https://yard-search.com/api/search?q=your+search+query

List Episodes

GET https://yard-search.com/api/episodes

Get Episode Transcript

GET https://yard-search.com/api/episodes/{videoId}

We now expose a read-only MCP endpoint for transcript search, episode listing, and transcript retrieval.

Endpoint

POST https://yard-search.com/mcp

Discovery Metadata

GET https://yard-search.com/.well-known/mcp/server-card.json

The MCP server is stateless and read-only. It currently exposes three tools: transcript search, episode listing, and episode detail retrieval. The server card is published at the standard well-known path used by Cloudflare Agent Readiness.

Public pages feature-detect document.modelContext.registerTool() and register the same search and episode tools on page load so WebMCP-enabled browser agents can discover them without scraping the DOM.

This is an experimental browser-side surface. Each tool is registered independently with a page-owned abort signal, and pages without the WebMCP API continue to work normally.

We support the emerging llms.txt standard—a simple text file that helps AI models understand what our site does and how to use it.

Think of it like robots.txt, but for AI assistants.