TubeText for Muse

Bring YouTube transcripts, AI summaries, and translations into Meta Muse. TubeText runs a public, read-only MCP server — point Muse at it once and ask about any video from any chat.

Set it up (takes a minute)

1. Open Muse (iOS, Android, or web) and start a new chat.
2. Paste the setup prompt below and send it.
3. Muse connects, lists the tools, and saves TubeText as a skill.

Set up TubeText as a custom connector.

- Endpoint: https://tubetext-mcp-production.up.railway.app/mcp (hosted MCP server over streamable HTTP, JSON-RPC 2.0).
- Auth: none needed. Read tools are public; AI tools run on a shared daily quota.
- After connecting, list every tool it gives you and show me the results.
- Then save it as a reusable skill called "TubeText" so I can ask about YouTube videos from any conversation.

Confirm success by listing the available TubeText tools.

What Muse can do with it

• get_transcript

  Fetch a YouTube video's transcript with timestamps.

• list_languages

  List caption languages available for a video.

• search_videos

  Search YouTube by topic or list a channel's latest videos.

• resolve_channel

  Find the right channel when given just a creator name.

• summarize

  Summarize a transcript into key takeaways.

• translate

  Translate a transcript into another language.

Example prompts

• “What are the most relevant videos about Meta Muse right now?”
• “Tell me about Fireship's last 5 videos.”
• “Summarize the latest videos on AI coding agents.”

Access & limits

• No account needed. Transcript and search tools are unlimited.
• AI tools (summarize, translate) run on a shared quota of 20 calls per day per IP.
• Read-only: TubeText never posts, uploads, or modifies anything.

Technical details

Source code and integration docs: github.com/anomalyco/muse_connectors

How TubeText handles your data: Privacy Policy · Terms of Service