Public MCP Server

Let your AI assistant search the show and read transcripts

https://myweirdprompts.com/mcp

Free · read-only · no account or API key · Streamable HTTP

What it is

My Weird Prompts is an AI-generated podcast. Daniel Rosehill records short voice prompts, and two AI hosts, Corn the sloth and Herman Poppleberry the donkey, discuss each one in a full episode.

The Model Context Protocol (MCP) is the standard way AI assistants plug into outside data. Add this server to Claude, ChatGPT, Cursor or any other MCP client, and you can ask it about the show in plain language. It finds the right episodes, reads the transcripts and quotes from them, so you don't have to search the archive yourself.

The server only reads. It can't change anything, and it doesn't see or store anything about you beyond the request itself.

Connect

Claude (web and desktop)

Settings → Connectors → Add custom connector. Name it "My Weird Prompts" and paste the URL above. No authentication.

ChatGPT

Settings → Apps & Connectors → Advanced → enable developer mode, then Create a connector with the URL above and authentication set to none. Availability depends on your plan.

Claude Code

claude mcp add --transport http my-weird-prompts https://myweirdprompts.com/mcp

Cursor, VS Code, Windsurf and other clients

Add this to the client's MCP config file (for Cursor, ~/.cursor/mcp.json). Some clients spell the type "streamable-http".

{
  "mcpServers": {
    "my-weird-prompts": {
      "type": "http",
      "url": "https://myweirdprompts.com/mcp"
    }
  }
}

Tools

The server offers 5 tools. Your assistant chooses which to call, so you don't need to name them. They're listed here so you know what it can do.

`search_episodes` Search episodes

Search the whole archive, including transcripts. mode "fulltext" (default) matches words and phrases, supports "quoted phrases", OR and -negation, and resolves a bare number as an episode number. mode "semantic" matches by meaning — use it for questions or fuzzy topics. Returns slugs, titles and excerpts; follow up with get_episode or get_transcript.

Parameter	Type	Notes
`query`required	string	What to search for.
`mode`	"fulltext" \| "semantic"	Default: "fulltext"
`limit`	integer	Default: 10 Max: 50

`list_episodes` List episodes

List episodes newest first, optionally filtered. With no arguments returns the latest 10. Filters combine: tag (exact tag name), category (category id, e.g. "ai-core"), keyword (substring of title, description, prompt or tags — not transcripts; use search_episodes for those), episode_number. Page with offset.

Parameter	Type	Notes
`limit`	integer	Default: 10 Max: 100
`offset`	integer	Default: 0
`tag`	string
`category`	string
`keyword`	string
`episode_number`	integer

`get_episode` Get episode

One episode's full record: title, description, publish date, tags, duration, audio URL, Daniel's original prompt (promptTranscript) and show notes. Identify it by slug or episode_number. Set include_transcript to also get the full transcript (often 20,000+ characters).

Parameter	Type	Notes
`slug`	string
`episode_number`	integer
`include_transcript`	boolean	Default: false

`get_transcript` Get transcript

The full episode transcript as plain text: Daniel's prompt followed by the hosts' conversation, one speaker turn per paragraph. Identify the episode by slug or episode_number.

Parameter	Type	Notes
`slug`	string
`episode_number`	integer

`about_the_show` About the show

What My Weird Prompts is, who makes it (Daniel Rosehill and the AI hosts), how episodes are produced, licensing, and links for listening and for bulk data.

No parameters.

Things to ask

Once connected, try something like:

What has My Weird Prompts said about sleep? Summarise the three most relevant episodes.
Find the latest episode and give me the key points from its transcript.
Which episodes are tagged "ai-agents"? Pick a recent one and quote what Herman says in it.
What was Daniel's original prompt for episode 500, and did the hosts actually answer it?

Want to try the show format itself? The AI Sandbox has the system prompts and character cards to experiment with, and this server can pull real transcripts to use as examples.

For agents and developers

Endpoint: https://myweirdprompts.com/mcp
Transport: Streamable HTTP, stateless. POST JSON-RPC (one message or a batch) and get application/json back. No session ID, no SSE stream. GET returns 405.
Protocol versions: 2025-06-18, 2025-03-26, 2024-11-05
Auth: None. CORS is open to all origins, so browser-based clients work.
Server name: my-weird-prompts v1.0.0
Discovery: /.well-known/mcp.json, /llms.txt, /for-ai-agents

Raw call with curl

curl -s https://myweirdprompts.com/mcp \
  -H 'Content-Type: application/json' \
  -H 'Accept: application/json, text/event-stream' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"list_episodes","arguments":{"limit":3}}}'

Not using MCP? The same data is available as plain HTTP: /api/episodes.json, /api/episodes/{slug}.json and /transcripts/{slug}.txt. For the whole archive at once, use the Hugging Face datasets instead of calling the API in a loop.

Limits & license

Rate limits: about 60 calls a minute per IP address, and 30 a minute for semantic search. If you go over, the tool returns a "rate limited" message. Wait a minute and try again.

License: episode content is CC BY 4.0. You can reuse it, quote it and build on it. Credit it as "My Weird Prompts by Daniel Rosehill" and say that it is AI-generated.

Who runs this: Daniel Rosehill. Questions or bugs: contact page.