MCP server
Connect Claude, Cursor, VS Code and other AI assistants to Podzone with your API key.
Podzone's MCP server lets an AI assistant that speaks the Model Context Protocol search podcasts, read transcripts and manage alerts for you. Ask in plain language — "what have podcasts said about Acme this month?" — and the assistant calls Podzone's tools to answer.
The server is at:
https://api.podzone.fm/mcp
It authenticates with a Podzone API key, sent as an Authorization: Bearer pzk_live_... header. Create one on the app's API keys page; Authentication explains the difference between team and account keys.
Claude Code #
Add the server once, then start claude and ask in plain language. /mcp shows whether it is connected.
claude mcp add --transport http podzone https://api.podzone.fm/mcp \
--header "Authorization: Bearer pzk_live_..."
Claude on the web and desktop #
Open Settings > Connectors and choose Add custom connector. Enter https://api.podzone.fm/mcp as the server URL, choose token authentication rather than signing in with OAuth, and add an Authorization header whose value is Bearer pzk_live_... — including Bearer and the space.
The Podzone tools are usually available in a chat straight away. If not, turn the connector on for that chat from its + menu.
Cursor #
Add the server to .cursor/mcp.json in your project, or to ~/.cursor/mcp.json for every project:
{
"mcpServers": {
"podzone": {
"url": "https://api.podzone.fm/mcp",
"headers": { "Authorization": "Bearer pzk_live_..." }
}
}
}
VS Code #
Add the server to .vscode/mcp.json in your workspace:
{
"servers": {
"podzone": {
"type": "http",
"url": "https://api.podzone.fm/mcp",
"headers": { "Authorization": "Bearer pzk_live_..." }
}
}
}
Other clients #
Any MCP client that connects to a remote server over HTTP and lets you set the Authorization header works the same way: Windsurf, the Claude API's MCP connector, and agent frameworks built on the MCP SDKs among them. Clients that only offer OAuth sign-in, such as ChatGPT, cannot connect yet.
Tools #
| Tool | What it does | Effect |
|---|---|---|
search_episodes |
Search episodes | Reads only |
get_episode |
Get an episode | Reads only |
find_in_transcript |
Find words in a transcript | Reads only |
read_transcript |
Read a transcript | Reads only |
search_podcasts |
Search podcasts | Reads only |
get_podcast |
Get a podcast | Reads only |
list_podcast_episodes |
List a podcast's episodes | Reads only |
list_alerts |
List alerts | Reads only |
get_alert |
Get an alert | Reads only |
list_alert_matches |
List an alert's matches | Reads only |
get_alert_match |
Get an alert match | Reads only |
create_alert |
Create an alert | Changes data |
update_alert |
Update an alert | Changes data |
whoami |
Describe this API key | Reads only |
list_teams |
List teams | Reads only |
list_team_members |
List a team's members | Reads only |
No tool deletes anything. Each tool calls one route of the API, so what a tool can see and do is what the key can see and do through the API.
What it costs #
Every tool call is one API request, counted against your account's limits exactly as if you had called the route yourself. Reading a transcript from its first page also counts as one of your plan's transcript downloads; later pages, and finding words in a transcript, do not.
With an account-scoped key, the tools that act for a team take a team_id argument in place of the X-Podzone-Team header. The assistant can find the ids with the list_teams tool.