MCP Connector

The MCP Connector lets you connect an AI assistant (Claude, ChatGPT, Cursor, Perplexity, and other MCP clients) directly to your Jellypod account. Once connected, the assistant works inside your account: create podcasts and hosts, add sources, generate and edit episodes, and publish them, all from a chat.

The MCP Connector is available to all users on every plan.

Server details

Add Jellypod as a remote MCP server in your client with these settings:

Server URL:  https://mcp.jellypod.com/mcp
Transport:   HTTP (Streamable)
Auth:        OAuth (sign in on first connect)

Most clients connect over OAuth: the first time your assistant connects, you sign in through Jellypod and authorize access. The assistant then acts on your account using your own permissions, so it only ever sees the podcasts, episodes, and hosts you already have access to.

Connect your client

Claude (Web and Desktop)

1. In Claude, go to Settings, click Connectors, then click Add custom connector.
2. Set Name to Jellypod and Remote MCP server URL to https://mcp.jellypod.com/mcp, then click Add.
3. Click Connect, sign in to Jellypod if prompted, and approve access to finish.

Claude Code

1. Install Claude Code, then add the Jellypod MCP server:

   claude mcp add --transport http jellypod https://mcp.jellypod.com/mcp

2. Start a Claude Code session, run /mcp, select the Jellypod server, then sign in to Jellypod and approve access.

ChatGPT (Web)

1. In ChatGPT, open Settings, find the Apps and Connectors section, then toggle on Developer mode under Advanced settings.
2. Add a new app with Name Jellypod, MCP Server URL https://mcp.jellypod.com/mcp, and Authentication set to OAuth.
3. Click Create. If a Jellypod sign-in window opens, approve access to complete the connection.

Codex CLI

1. Install the Codex CLI, then add the Jellypod MCP server:

   codex mcp add jellypod --url https://mcp.jellypod.com/mcp

2. A browser window opens. Sign in to Jellypod if prompted and approve access.

Cursor

1. Open Cursor Settings and click the Tools and MCP tab.

2. Click New MCP Server and add this configuration:

   {
     "mcpServers": {
       "jellypod": {
         "url": "https://mcp.jellypod.com/mcp"
       }
     }
   }

3. Save the configuration. On first use, sign in to Jellypod and approve access.

Perplexity (Web and Desktop)

1. In Perplexity, go to All settings, then Connectors, and choose Custom connector.
2. Set Name to Jellypod and MCP Server URL to https://mcp.jellypod.com/mcp, check the custom-connector consent box, then click Add.
3. Find Jellypod under your custom connectors, click Add connector, then sign in to Jellypod and approve access.

Raycast

Raycast uses an API key. Create one in the studio under Settings, then API Keys first (Creator plan or higher).

1. In Raycast, search for Install Server and press Enter.
2. Set Name to Jellypod, Transport to HTTP, and URL to https://mcp.jellypod.com/mcp.
3. Under HTTP Headers, add an item with Key Authorization and Value Bearer YOUR_API_KEY, replacing YOUR_API_KEY with your key.
4. Click Install (or press Cmd+Enter) to finish.

Zapier

Zapier uses an API key. Create one in the studio under Settings, then API Keys first (Creator plan or higher).

1. In Zapier, add an MCP Client by Zapier connection.
2. Set Server URL to https://mcp.jellypod.com/mcp, Transport to Streamable HTTP, OAuth to No, and Bearer Token to your API key.
3. Finish connecting the MCP Client by Zapier.

n8n

n8n uses an API key. Create one in the studio under Settings, then API Keys first (Creator plan or higher).

1. Inside a workflow, add an MCP Client node.
2. Set Server Transport to HTTP Streamable, MCP Endpoint URL to https://mcp.jellypod.com/mcp, and Authentication to Bearer Auth.
3. Create a new Bearer Auth credential, paste your API key, then save.
4. Choose the Jellypod tools you want the workflow to use and configure them as needed.

Any other MCP client

If your tool supports remote MCP servers but does not have a dedicated guide above, add an HTTP (Streamable) MCP server pointing at https://mcp.jellypod.com/mcp. Authenticate with OAuth (sign in and approve on first connect) if your client supports it, or send an API key as an Authorization: Bearer header if it does not.

What your assistant can do

The connector exposes your account's full content lifecycle:

• Podcasts: list, create, update, and delete podcast series
• Episodes: generate from sources or an imported script, list, update, and delete
• Hosts: browse the voice library, create AI hosts, update, and delete them
• Sources: add research from a URL, YouTube video, or pasted text, and list or remove them
• Cover art: generate episode and podcast cover art from a prompt
• Analytics and transcripts: read podcast and episode analytics, and export transcripts as JSON, SRT, or VTT

Example prompts

• List all my podcasts and tell me which episodes are not published yet
• Create a podcast about the latest AI research and generate the first episode
• Show me download analytics for my most recent episode
• Generate cover art for one of my podcasts

Guardrails

A few deliberate guardrails keep the assistant from making changes you did not intend:

• Nothing publishes automatically. Episodes are generated as drafts. Publishing only happens when you explicitly ask, and the assistant confirms first.
• Deletes are protected. Deleting a podcast or episode that is published or scheduled requires an explicit confirmation, so live content cannot be removed by accident.
• No billing, team, or account changes. The connector covers content only. Subscription, team, and account settings stay behind the studio UI.

File uploads and voice cloning are not yet available over the connector. Use the studio for those.