Endpoint
GET /v1/youtube/transcript on https://api.scority.ai.
Open →Use Scority transcript data in Claude-style workflows through server-side API calls or a local MCP-style wrapper around the same endpoint.
The stable integration path is a server-side call to https://api.scority.ai/v1/youtube/transcript. A local MCP-style wrapper can use the same contract when you want Claude Desktop to call a transcript tool.
For most teams, the cleanest Claude workflow is an internal tool or server route that calls Scority and returns only the transcript text and segments needed by the assistant.
A local wrapper should expose one real tool, get_youtube_transcript, backed by the Scority Transcript API. Keep it narrow so Claude receives transcript data without extra product surfaces.
Replace the path with the built local tool location on your machine and provide the API key through the environment.
{
"mcpServers": {
"scority": {
"command": "node",
"args": [
"/absolute/path/to/scority/packages/scority-mcp/dist/index.js"
],
"env": {
"SCORITY_API_KEY": "YOUR_API_KEY"
}
}
}
}This integration should point users to the real transcript API contract. Do not add public package installs, speech-to-text guarantees, search tools, channel tools, billing tools, or dashboard tools to the Claude setup.
Read the broader MCP workflow notes and local tool shape.
Open →Handle unauthorized, rate_limited, quota_exceeded, transcript_not_available, and upstream_transcript_failed.
Open →Use language or lang for language selection, but do not send both.
Open →