$ loading_
Let LLMs watch local videos and search previously viewed content.
Copy the install command and let the AI configure it · recommended for beginners
Please install the "io.github.HUANGCHIHHUNGLeo/claude-real-video" MCP server from askskill: Run: claude mcp add 'io-github-huangchihhungleo-claude-real-video' -- npx -y claude-real-video
Researchers can have an LLM watch local interview videos or recordings first, then search for a topic or scene to reduce manual review time.
Product managers can use it to locate where a feature appears in a demo video and summarize the related explanation.
Developers can connect it to an MCP-enabled LLM workflow so the model can search and answer questions based on previously watched local videos.
This is an MCP tool whose core capability is to let an LLM watch local videos and search content from videos it has already watched.
The provided description explicitly says “locally,” so it supports having the model watch local videos. For online video support, see the source repository.
The provided material does not include installation steps, runtime details, or key requirements. Please see the source repository for specifics.
Enable Claude to analyze video frames and audio for richer understanding.
Help AI agents understand local videos and search moments by content.
Transcribe, search, and analyze videos with structured summaries, key points, and verdicts.
Extract timestamped screenshots from local videos for visual walkthrough generation.
Search YouTube videos, transcripts, channels, and trends for research and analysis.
Upload, transcribe, search, summarize, and chat with social video content.