Tubask
Youtube, readable in AI chats.
README
<p align="center"> <img src="assets/logo.svg" alt="Tubask" width="64" height="64" /> </p>
<h1 align="center">Tubask</h1>
<p align="center"> <strong>YouTube, readable in chat.</strong><br /> Hosted MCP server for Claude, Cursor, ChatGPT, and any Streamable HTTP client. </p>
<p align="center"> <a href="https://tubask.app"><img src="https://img.shields.io/badge/website-tubask.app-111110?style=flat-square" alt="Website" /></a> <a href="https://tubask.app/mcp"><img src="https://img.shields.io/badge/MCP-Streamable%20HTTP-111110?style=flat-square" alt="MCP transport" /></a> <a href="https://tubask.app/docs"><img src="https://img.shields.io/badge/docs-quick%20start-111110?style=flat-square" alt="Docs" /></a> <img src="https://img.shields.io/badge/tools-3-111110?style=flat-square" alt="3 tools" /> <img src="https://img.shields.io/badge/auth-OAuth%202.0-111110?style=flat-square" alt="OAuth 2.0" /> </p>
<p align="center"> <a href="https://tubask.app/signup"><strong>Try free →</strong></a> · <a href="#quick-start">Install</a> · <a href="#tools">Tools</a> · <a href="https://tubask.app/docs/examples">Example prompts</a> </p>
Tubask is a hosted remote MCP server that lets your AI assistant search YouTube, summarize videos, and pull timestamped quotes — from inside Claude or Cursor, in one message. No local Python. No 15-tool chains. Three tools, smart routing, OAuth, and a hosted trial (25 searches + 3 summaries) — then upgrade to Pro (no Google key) or stay on Free with your own key.
| Without Tubask | With Tubask |
|---|---|
| Paste a link → “I can't access YouTube” | Paste a link → summary, chapters, quotes |
| Open 5 tabs to research a topic | Ranked results in chat |
| Chain search → transcript → summarize | One summarize_video or goal=advice call |
MCP endpoint
https://tubask.app/mcp
Transport: Streamable HTTP · Auth: OAuth 2.0 (browser signup on first tool call) · Read-only tools
Quick start
npx @tubask/mcp
Interactive setup: pick your client, write config, optionally install the Cursor skill.
No args needed.
npx @tubask/mcp init # non-interactive config write
npx @tubask/mcp skill install # Cursor routing skill
npx @tubask/mcp status # check existing setup
Then Settings → Tools & MCP → Connect in Cursor and complete OAuth on first use.
- Create a free account at tubask.app/signup (no credit card).
- Add the MCP server to your client (
npxabove, or snippets below). - Send any YouTube URL in chat — e.g. “Summarize this talk and list the main argument.”
- On first use, complete OAuth in your browser.
- (Optional) Add your free YouTube Data API key in the Tubask dashboard after the hosted trial.
npx CLI
| Command | What it does |
|---|---|
npx @tubask/mcp |
Interactive onboarding (recommended) |
npx @tubask/mcp init |
Write .cursor/mcp.json in the current project |
npx @tubask/mcp init --global |
Write ~/.cursor/mcp.json for all projects |
npx @tubask/mcp init --client claude-code |
Write .mcp.json + print claude mcp add command |
npx @tubask/mcp skill install |
Copy tubask-youtube skill to .cursor/skills/ |
npx @tubask/mcp status |
Check if Tubask is already configured |
npx @tubask/mcp init --dry-run |
Preview config without writing files |
The server runs at https://tubask.app/mcp — the CLI only writes client config.
Cursor
Add to .cursor/mcp.json (project) or ~/.cursor/mcp.json (global):
{
"mcpServers": {
"tubask": {
"url": "https://tubask.app/mcp"
}
}
}
Then Settings → Tools & MCP → Connect and authorize.
Claude Code
claude mcp add --transport http tubask https://tubask.app/mcp
Or add to .mcp.json:
{
"mcpServers": {
"tubask": {
"type": "http",
"url": "https://tubask.app/mcp"
}
}
}
Claude Desktop / Claude.ai
Settings → Connectors → Add custom connector → paste:
https://tubask.app/mcp
ChatGPT
Settings → Apps & Connectors → Create app → paste the same URL.
Smithery
smithery mcp add https://tubask.app/mcp
No extra config — OAuth and API keys are handled in your Tubask account after connect.
Other clients
Any MCP host with remote HTTP support: paste https://tubask.app/mcp and sign in on first tool call.
Client-specific notes: tubask.app/docs/connect
Tools
| Tool | Use when |
|---|---|
youtube_query |
Paste anything — URL, @handle, playlist, search text. Search, latest uploads, comments, creator advice across videos. |
summarize_video |
Understand one video — summary, key_points, chapters, quotes with timestamps. |
get_transcript |
Exact words — keyword search, time slices, pagination. |
Every response includes next_steps[] so the model doesn't guess the next tool.
Example prompts
Summarize this and give me 3 quotable lines with timestamps:
https://youtube.com/watch?v=...
What's @Fireship published about AI agents in the last month? Skip shorts.
Find every time they say "product-market fit" and give timestamps:
https://youtu.be/VIDEO_ID
More: tubask.app/docs/examples
Why three tools?
Most YouTube MCP servers expose 10–20 granular tools. Models chain search → get_video → get_transcript → summarize, burning tokens, latency, and YouTube API quota.
Tubask collapses the surface:
Paste anything → youtube_query (intent=auto)
One video → summarize_video
Exact words → get_transcript
Creator advice across multiple videos → youtube_query(goal=advice) in one call (up to 8 recent long-form uploads).
Pricing
| Plan | Price | Includes |
|---|---|---|
| Trial | Free | 25 searches + 3 summaries · hosted · no API key |
| Free | $0 | 100 credits/mo (~12 summaries or ~100 searches) · your Google API key |
| Pro | $8/mo | 800 credits/mo (~100 summaries) · hosted · 3 MCP clients |
| Plus | $19/mo | 2,500 credits/mo (~312 summaries) · hosted · unlimited clients |
Credits: 1 search = 1 credit · 1 summary/transcript = 8 credits · metadata-only summary = 1 credit · advice = 1 + 8×transcript-covered videos. Hard stop at zero.
Details: tubask.app/docs/quota
Security
- OAuth 2.0 for every MCP connection
- Your YouTube API key encrypted at rest (Fernet)
- Argon2id password hashing
- Read-only tools — no writes to your YouTube account
tubask.app/docs/security · SECURITY.md
Comparison
| Tubask | Local uvx YouTube MCPs |
|
|---|---|---|
| Install | Paste URL | Python/uv, local deps |
| Summaries | Built-in (chapters, quotes) | Often transcript-only |
| Channel / search / advice | youtube_query |
Varies; often many tools |
| Auth | OAuth + dashboard | Usually none |
| API key | Your key in Tubask dashboard | Optional / none |
Discovery metadata
MCP scanners that cannot authenticate may read server-card.json or the live endpoint at tubask.app/.well-known/mcp/server-card.json.
Support
- Troubleshooting
- GitHub Issues — bugs, feature requests, and public discussion
- support@tubask.app
- Privacy · Terms
License
This repository contains documentation and discovery metadata for the Tubask hosted service. The Tubask service software is proprietary. See LICENSE.
<p align="center"> <sub>v0.6.4 · Model Context Protocol · Made for Claude, Cursor, and builders who live in chat.</sub> </p>
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
Neon Database
MCP server for interacting with Neon Management API and databases
E2B
Using MCP to run code via e2b.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.