Video Research MCP
@Galbaz1
Give Claude Code 51 research, analysis & media production tools with one command. Video analysis, deep research, academic papers, TTS voice-over, AI video generation, and Weaviate knowledge store — powered by Gemini 3.5 Flash.
Overview
What is Video Research MCP?
Video Research MCP is a Claude Code plugin that bridges Claude’s text-only capabilities with Google Gemini for video understanding, deep research, web search, and media production. It provides 51 tools, 17 slash commands, 12 skills, and 7 sub‑agents across three MCP servers.
How to use Video Research MCP?
Install with a single npx command (npx video-research-mcp@latest) and export a GEMINI_API_KEY. The installer copies commands, skills, and agents to ~/.claude/ and configures the MCP servers to run via uvx from PyPI. It also works as a standalone MCP server with any MCP client by configuring the mcpServers section.
Key features of Video Research MCP
- Analyze local video files and YouTube videos with timestamps and concept maps
- Deep research with evidence‑tier labeling (Confirmed, Strong Indicator, Inference, Speculation)
- Web search via Gemini grounding with source citations
- Analyze PDFs, URLs, raw text, and entire directories of documents
- Persistent semantic search via optional Weaviate knowledge store
- Explainer video synthesis from research content
- Parallel scene generation and interactive visualizations
Use cases of Video Research MCP
- Create meeting minutes and action items from a recorded video call
- Analyze a YouTube tutorial for timestamps, concept map, and comment sentiment
- Research a technical topic with graded evidence and an interactive evidence network
- Compare and synthesize findings across multiple academic papers or reports
FAQ from Video Research MCP
What API keys are required?
A Google AI API key (GEMINI_API_KEY) is required. Download it from Google AI Studio.
Does Video Research MCP require Weaviate?
No. Weaviate is optional. Without it the plugin works the same, but persistent semantic search is not available.
Can Video Research MCP be used without Claude Code?
Yes. The tools are standard MCP and can be called by any MCP client using the uvx command configuration provided in the README.
How are large video files handled?
Files over 20 MB are uploaded to Gemini’s File API and context‑cached; follow‑up questions reuse the cache instead of re‑uploading.
What are the system requirements?
Python 3.11+, uv, Node.js 16+, and a Google AI API key.
