McptubeSAFE
Transform YouTube videos into a compounding knowledge base with transcripts, vision analysis, and agentic search. Works as an MCP server for Claude, Copilot & more.
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
YouTube video knowledge engine — transcripts, vision, and persistent wiki.
[](https://pypi.org/project/mcptube/) [](https://pypi.org/project/mcptube/) [](https://opensource.org/licenses/MIT)
mcptube-vision transforms YouTube videos into a persistent, structured knowledge base using both transcripts and visual frame analysis. Built on the Karpathy LLM Wiki pattern: knowledge compounds with every video you add.
Evolved from [mcptube](https://pypi.org/project/mcptube/) v0.1 — mcptube-vision replaces semantic chunk search with a persistent wiki that gets smarter with every video ingested.
🧠 How It Works
Traditional video tools re-discover knowledge from scratch on every query. mcptube-vision is different:
mcptube v0.1 mcptube-vision ┌─────────────────────┐ ┌─────────────────────────┐ │ Query → vector search│ │ Video ingested → LLM │ │ → raw chunks → LLM │ │ extracts knowledge → │ │ → answer (from scratch│ │ wiki pages created → │ │ every time) │ │ cross-references built │ └─────────────────────┘ │ │ │ Query → FTS5 + agent │ │ → reasons over compiled │ │ knowledge → answer │ └─────────────────────────┘
31b0639c852bOBSERVED · 2026-10-07Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control.
claude mcp add mcptube -- uvx mcptube
{
"mcpServers": {
"mcptube": {
"command": "uvx",
"args": [
"mcptube"
]
}
}
}Exposed tools (21)
18 read · 2 write · 1 destructive. Blast radius: 1 tool can delete or overwrite — an agent that can be talked into calling a tool can be talked into calling this one.
| Tool | Risk | Description |
|---|---|---|
add_video | write | Ingest a YouTube video and build wiki knowledge pages. |
ask_video | read | Ask a question about a single video |
ask_videos | read | Ask a question across multiple videos. |
classify_video | read | Get metadata for classification — YOU classify it. |
discover_videos | read | Search YouTube for videos on a topic. |
generate_report | read | Get data to generate an illustrated report for a video. |
generate_report_from_query | read | Search the library and return data for a cross-video report. |
get_frame | read | Extract a frame from a video at a specific timestamp. |
get_frame_by_query | read | Search a video |
get_frame_data | read | Extract a frame and return as base64 for embedding in reports. |
get_info | read | Get full details for a video including transcript and chapters. |
list_videos | read | List all videos in the library. |
remove_video | destructive | Remove a video from the library and clean wiki references. |
save_tags | write | Save classification tags for a video. |
synthesize | read | Get data for cross-video synthesis on a topic. |
wiki_ask | read | Ask a question — answered via agentic wiki retrieval. |
wiki_history | read | View version history for a wiki page. |
wiki_list | read | Browse wiki pages with optional filtering. |
wiki_search | read | Full-text search across all wiki pages. |
wiki_show | read | Read a specific wiki page in full. |
wiki_toc | read | Get the wiki table of contents. |
Trust audit
SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | PASS |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | WARN |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (2 observation(s))
- Network
- declared (2 observation(s))
- Shell
- none-observed
- Dependencies
- pinned
- Secrets in source
- none-found
Findings (7)
remove_video
"url": "http://127.0.0.1:9093/mcp"
claude mcp add mcptube --transport http http://127.0.0.1:9093/mcp
"url": "http://127.0.0.1:9093/mcp"
"serverUrl": "http://127.0.0.1:9093/mcp"
url = "http://127.0.0.1:9093/mcp"
Gates applied: no_behavioural_pass, no_license.
31b0639c852bfull audit observations/trust-audit/mcp-server/0xchamin__mcptube.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-07 | 31b0639c852b | SAFE | B | 89 | first audit |
Questions
What is the Mcptube MCP server?
Transform YouTube videos into a compounding knowledge base with transcripts, vision analysis, and agentic search. Works as an MCP server for Claude, Copilot & more.
What tools does Mcptube expose?
21 in total: 18 read-only, 2 that write, and 1 that can delete or overwrite (remove_video). Every one is listed on this page with its risk.
Is Mcptube safe to connect to an agent?
The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean server reads B. Separately from the audit: 1 of its tools can destroy data, so scope the token you give it to what you actually need.
What credentials does Mcptube need?
No credential environment variables were found in its source, so it appears to need none.
How does Mcptube run?
It speaks stdio and streamable-http, so it runs as a local process your client starts. It is published on PyPI as mcptube.
How current is this page?
The grade is for one exact copy of the source (31b0639c852b), read on 2026-10-07. The repository is watched and re-audited when it changes.