Search and ask 236,000+ podcast transcripts; answers cite the episode and minute.
Maintain this server? Claim it
Call this server's tools from your browser. Nothing to install.
Bearer pgk_... Podgist API key. Optional: list_services and find_podcasts are free.
Requests go through mcptry because browsers can't call most MCP servers directly. Header values are sent to the server for that request only and are never stored or logged.
Tool list from our last check. Connect live to refresh it.
Starts a cited answer drawn only from what was said on podcasts. Returns a job_id; call get_answer with it until status is complete (usually under a minute). Needs an API key; billed per question (see list_services).
This tool isn't marked read-only, so it may change something on the server or in an account you connected.
Pick your client. Values in capitals, like YOUR_TOKEN, are yours to fill in.
claude mcp add --transport http podgist https://api.podgist.com/mcp
Headers
Authorization (optional, not in the config above): Bearer pgk_... Podgist API key. Optional: list_services and find_podcasts are free.Starts a cited answer drawn only from what was said on podcasts. Returns a job_id; call get_answer with it until status is complete (usually under a minute). Needs an API key; billed per question (see list_services).
| Argument | Type | Description |
|---|---|---|
| question* | string |
{
"type": "object",
"required": [
"question"
],
"properties": {
"question": {
"type": "string",
"maxLength": 500,
"minLength": 3
}
},
"additionalProperties": false
}Public podcasts whose name contains the text, with their slug (for the podcast filter of search_transcripts). Free; no key needed. At most 20 results.
| Argument | Type | Description |
|---|---|---|
| name* | string |
{
"type": "object",
"required": [
"name"
],
"properties": {
"name": {
"type": "string",
"maxLength": 100,
"minLength": 2
}
},
"additionalProperties": false
}The status and, once complete, the answer and its cited passages for a job from ask_question. Needs the API key that asked. Not billed.
| Argument | Type | Description |
|---|---|---|
| job_id* | string |
{
"type": "object",
"required": [
"job_id"
],
"properties": {
"job_id": {
"type": "string",
"maxLength": 64,
"minLength": 8
}
},
"additionalProperties": false
}One episode's transcript, with speaker turns and timestamps, by podcast and episode id (the sid and pid of a search result). Only what the key's account may read. Needs an API key; billed per transcript (see list_services).
| Argument | Type | Description |
|---|---|---|
| episode_id* | string | |
| podcast_id* | string |
{
"type": "object",
"required": [
"podcast_id",
"episode_id"
],
"properties": {
"episode_id": {
"type": "string",
"format": "uuid"
},
"podcast_id": {
"type": "string",
"format": "uuid"
}
},
"additionalProperties": false
}The key's account balance and monthly allowance, and its calls and cost per day over the last 30 days. Needs an API key. Not billed.
No arguments.
{
"type": "object",
"required": [],
"properties": {},
"additionalProperties": false
}Lists the tools, which need an API key, and the current price per call. Free; no key needed.
No arguments.
{
"type": "object",
"required": [],
"properties": {},
"additionalProperties": false
}Full-text search across transcripts: matching episodes with snippets, newest relevance first, 20 a page. Optional podcast slug and date range (YYYY-MM-DD). Needs an API key; billed per search (see list_services).
| Argument | Type | Description |
|---|---|---|
| page | integer | |
| query* | string | |
| date_to | string | |
| podcast | string | A slug from find_podcasts. |
| date_from | string |
{
"type": "object",
"required": [
"query"
],
"properties": {
"page": {
"type": "integer",
"maximum": 50,
"minimum": 1
},
"query": {
"type": "string",
"maxLength": 500,
"minLength": 1
},
"date_to": {
"type": "string",
"format": "date"
},
"podcast": {
"type": "string",
"description": "A slug from find_podcasts."
},
"date_from": {
"type": "string",
"format": "date"
}
},
"additionalProperties": false
}We read every tool's name, description and inputs the way the model will, looking for instructions hidden from you. A clean result means no warning signs, not a guarantee. Scanned Oct 31, 2026.