Mcptube MCP Server
YouTube videos transformed into a persistent, structured knowledge base.
Data last scanned today · Reviewed today
Overview
Mcptube MCP Server acts as a knowledge engine that processes YouTube content by combining transcript analysis with visual frame interpretation. Rather than performing simple semantic search, it utilizes an LLM to extract information from videos and compile them into a persistent, evolving wiki repository. This architectural approach ensures that information compounds over time as more videos are ingested. The system utilizes SQLite FTS5 for efficient indexing and FileWikiRepository for storing structured knowledge in JSON format. By integrating scene-change detection via ffmpeg and LLM-based vision models, the server creates cross-references between videos, allowing your AI assistant to reason over a growing library of compiled information instead of analyzing isolated video chunks on every query.
Our verdict
This server carries a high risk level with a security score of 24/100. The primary security concern is the absence of a read-only mode, meaning every installation grants full write access to the underlying system. Furthermore, the lack of a detected license creates legal ambiguity, and the project is maintained by a single contributor without official vendor oversight, which may affect long-term stability and security auditing. It is recommended for users who understand these permissions risks and are comfortable evaluating third-party Python code.
- Compounding knowledge base architecture avoids redundant re-processing
- Uses FTS5 and LLM reasoning for accurate knowledge retrieval
- Features 0 direct dependencies to reduce supply-chain surface area
- High-risk profile due to lack of read-only mode
- No detected license creates legal usage ambiguity
- Single-contributor project lacks official vendor backing
Tools
| Tool | Description | Risk |
|---|---|---|
| add_video | Add a YouTube video to the wiki, extract knowledge, and update search index. | medium |
| search_wiki | Full-text keyword search across the wiki pages using FTS5. | low |
| get_wiki_page | Retrieve the full content of a specific wiki page by its slug. | low |
| list_wiki_pages | List all pages in the wiki, including entities, topics, and concepts. | low |
| get_wiki_toc | Retrieve the table of contents for the entire wiki. | low |
| remove_video | Remove a video and its contributions from the wiki. | medium |
| list_videos | List library | low |
| wiki_list | Browse wiki pages | low |
| wiki_show | Read a wiki page | low |
| wiki_search | Full-text search | low |
Compatibility
| Client | Local | Docker | Remote | Read-only |
|---|---|---|---|---|
| Claude Desktop | ||||
| Cursor | ||||
| VS Code | ||||
| Windsurf | ||||
| ChatGPT |
Frequently asked questions
›What tools are required for the video ingestion process?
The system utilizes ffmpeg for scene-change frame detection during the ingestion pipeline.
›How does this server handle storage and search?
It stores knowledge as JSON pages on disk in a FileWikiRepository and maintains a SQLite FTS5 index for keyword retrieval.
›Is the information retrieval performed on raw video files?
No, the server relies on a write-once ingestion pipeline that processes transcripts and LLM-generated frame descriptions into a persistent, structured knowledge base.
›How does the system manage knowledge across multiple videos?
The WikiEngine uses merge semantics to append or create new wiki pages, allowing knowledge to compound rather than treating videos as isolated units.
Badge
Maintain this server? Add the live badge to your README.