Use TranscriptX inside your AI
You can now use TranscriptX inside your AI. Here's how to set it up.
- Generate a token
- Paste the URL into your AI client
- Say "transcribe this video" inside Claude, Cursor, ChatGPT, or any other MCP-compatible client
Pick your client
What you can do once it's connected
Six tools become available to your AI:
transcribe_url— transcribe any public video or audio URL (1 credit)watch_url— actually watch a video: frames + timestamped transcript + engagement numbers (2 credits)list_creator_videos— list recent videos from a YouTube/TikTok/IG/Spotify profile (1 credit per 5)retry_transcript_language— re-run a previous transcript with a forced language (free)list_recent_transcripts— your recent transcript history (free)get_account_status— your plan and remaining credits (free)
Make your AI actually watch the video
watch_url samples frames from the video and hands them to your AI
alongside the timestamped transcript and the real view/like/comment counts. So it can talk about
what's on screen — the hook, the on-screen text, the editing, why it performed — not just
what was said. Tap any prompt to copy it.
Watching costs 2 credits per video — it downloads the video, samples the frames, and transcribes. Ask for a plain transcript instead when only the words matter.
Or just get the words
Common things that break
- "Connection refused" or 401. Token is wrong, expired, or revoked. Generate a new one at /account/mcp.
- Tools don't appear, no error. Restart your AI client fully. Some clients cache the tool list per-session.
- "Credit limit reached." Same monthly cap as the web product. Upgrade to Pro for unlimited, or wait for the monthly reset.
Questions or a client we don't list? [email protected]