MCP server

Transcripts for AI agents (MCP) one MCP server for Claude and Cursor

Our MCP server gives Claude Desktop, Claude Code and Cursor tools that fetch transcripts from links.

1.0Capabilities

Core advantages and the limits that come with them

Provide Claude, Cursor, and LLM autonomous agents with video transcripts and AI transcription via Model Context Protocol.

Zero scraper maintenance
Your agent calls the transcribe and get_video_transcript tools directly, with no browser automation or scraper of its own to maintain.
Token efficient exports
Agents can request raw text, segmented chunks, or timed SRTs to fit context windows efficiently.
Automatic job polling
The wait_for_job tool automatically polls the asynchronous transcription pipeline until results are ready.

ChatToolsServers

transcriptdock

Summarize the talk at youtube.com/watch?v=dQw4w9WgXcQ

get_video_transcriptDone

{ "url": "youtube.com/watch?v=dQw4w9WgXcQ", "format": "segments" }

00:00.0Welcome to this presentation on distributed systems.

00:03.5Today we cover storage architectures and partitioning.

The talk opens with an introduction to distributed systems, then moves on to storage architectures and partitioning.

Ask your agent to fetch a transcript...

Transcript ready· via MCP

Operational limits and transparency

Standard rate limits and workspace quotas apply.
Requires an active API key with workspace billing allowance.
Single region infrastructure.

2.0Implementation

Request and response the exact request and what comes back

Client setup

{
  "mcpServers": {
    "transcriptdock": {
      "command": "npx",
      "args": ["-y", "@transcriptdock/mcp"],
      "env": {
        "TRANSCRIPTDOCK_API_KEY": "td_live_..."
      }
    }
  }
}

Job outcome

{
  "id": "2d3e4f5a-6b7c-8d9e-0f1a-2b3c4d5e6f7a",
  "source": {
    "platform": "youtube",
    "media_id": "dQw4w9WgXcQ",
    "canonical_url": "https://www.youtube.com/watch?v=dQw4w9WgXcQ",
    "title": "Sample YouTube Video"
  },
  "source_origin": "creator_captions",
  "language": "en",
  "text": "Welcome to this presentation on distributed systems...",
  "segments": [
    {
      "start": 0.0,
      "end": 3.42,
      "text": "Welcome to this presentation on distributed systems."
    },
    {
      "start": 3.5,
      "end": 7.15,
      "text": "Today we cover storage architectures and partitioning."
    }
  ],
  "words": null,
  "timing_granularity": "segment",
  "duration_seconds": 212.5
}

Explore documentationAPI quickstart MCP server specs

3.0Questions

Frequently asked

What happens with captions, AI transcription and limits for this source.

Which MCP clients are supported?

Any client implementing the Model Context Protocol over standard input and output, including Claude Desktop, Claude Code, and Cursor.

Can an agent transcribe uploaded files?

Not through the MCP server today: its tools take YouTube and TikTok links. Uploads and direct audio links go through the HTTP API, where they run AI transcription on paid plans.

4.0Tools

Subtitle to text converter

Format subtitle files into LLM ready prompts.

Transcript cleaner

Strip filler words to optimize prompt token usage.