SkillFoxx

mcp-video-analyzer

MCP server, /video plugin and CLI: transcript, key frames, on-screen text and metadata for a YouTube, Loom or local video

  • 122 ★ on the repository
  • Last commit
  • MIT
MCP server

Extended access

Why this level

  • The server downloads videos from links the agent passes and stores frames in a disk cache
  • The MCP_ALLOW_PRIVATE_URLS variable opens private-network addresses to the agent
All reasons and checks

guimatheus92/mcp-video-analyzerAuthor:guimatheus92

Install

In your terminal, with SkillFoxx CLI

npx skillfoxx add mcp/mcp-video-analyzer

Detects the agents on your machine, shows the access level and pins the version.

Other ways to install

Access and checks

How we review

Extended accessThe tool runs code or makes network calls and only reads external data.

  • The server downloads videos from links the agent passes and stores frames in a disk cache
  • The MCP_ALLOW_PRIVATE_URLS variable opens private-network addresses to the agent

This is third-party code. Review the repository files before installing.

What it does

mcp-video-analyzer lets an agent work through a video from a link or a file. The `analyze_video` tool returns a timestamped transcript, key frames, on-screen text via OCR, an annotated timeline and metadata. yt-dlp reads YouTube, Vimeo, TikTok, Instagram, X, Twitch, Dailymotion and Facebook links, while Loom videos, direct URLs and local files the server handles itself. There are eight tools. `get_transcript` returns text only, `get_metadata` returns metadata with comments and chapters, `get_frame_at` pulls one frame by time and `analyze_moment` digs into a time range. When no native transcript exists, Whisper transcribes the audio. There are three ways in: the `/video` plugin for Claude Code, a plain MCP config for any client, or a skill plus CLI for Codex, Cursor and agents without MCP. The server reaches only public addresses.

Who it is for. Developers and analysts who hand an agent screen recordings, tutorials and video bug reports and want text and frames back.

Good fit when

  • A bug is recorded on Loom and the agent must find the second where the UI breaks
  • You look for the moment in a YouTube review where the author shows a specific setting
  • A call recording needs a timestamped transcript with the text from the slides
  • A batch of clips has to be processed so it can resume after a failure: `analyze_videos`

Not a fit when

  • The video sits on a private network or local server: private addresses are blocked by default
  • You need to archive a whole account of reels and build a knowledge base: the author points to a separate project, social-knowledge-base
  • You cannot install Node.js 22.12 or newer

Example request

Analyze the video https://youtu.be/jNQXAC9IVRw: give the transcript and say what is shown at 0:10.

MCP

Transport
stdio
Authentication
not required

Environment variables

TWELVELABS_API_KEY secret
TwelveLabs key for analysing direct URLs with the Pegasus model.
MCP_ALLOW_PRIVATE_URLS
Set to 1 to allow private and local addresses.

Limitations

Node.js 22.12 or newer is required. yt-dlp is mandatory for YouTube, Instagram, TikTok and similar links; without it the server returns a warning. Whisper transcribes videos with no subtitles, so quality depends on the model, and with the optional TwelveLabs Pegasus the transcript comes from a model with approximate timing and lines may be dropped. The server reaches only public addresses. The MCP_ALLOW_PRIVATE_URLS variable opens private networks, and by setting it you accept that the agent can reach your local network. Frames from the CLI stay in the cache directory until you delete them.

How to disable. Remove the server entry from your MCP settings or the `video` plugin through /plugin. The frame cache lives in the user cache directory under mcp-video-analyzer; clear it by hand.

FAQ

How do I install the /video plugin in Claude Code?

Run `/plugin marketplace add guimatheus92/mcp-video-analyzer`, then `/plugin install video@mcp-video-analyzer`. The `/video` command appears right away and the MCP server registers automatically.

How do I use it in Codex or Cursor without MCP?

Run `npx skills add guimatheus92/mcp-video-analyzer`. The `video` skill installs into every detected agent and calls the bundled CLI: `npx -y mcp-video-analyzer@latest analyze "<url>"`.

What happens without yt-dlp?

Direct URLs and local files still work, with frames from the bundled ffmpeg. For YouTube, TikTok and other platforms the server warns you to install yt-dlp, while transcripts and metadata do not depend on it.

How do I allow local addresses?

Set the environment variable `MCP_ALLOW_PRIVATE_URLS=1`. Cloud metadata endpoints stay blocked even with it.

What detail levels exist?

Three: `brief` returns metadata and transcript with no frames, `standard` is the default and `detailed` collects more frames. Set the frame count with `--max-frames`, from 1 to 60.

Discussion

No questions or reviews yet.

MCP servers

MCP server that gives an agent a YouTube video's transcript from its link, with language choice, timestamps and ads filtered out

MCP serverExtended access603

Claude Video Vision

claude-video-vision

An MCP server and Claude Code plugin that let the agent watch video: frames via ffmpeg and audio transcription via Gemini, local Whisper or OpenAI

MCP serverExtended access1.5K

MCP server for YouTube transcripts: an agent gets a video transcript by URL, with timestamps and metadata

MCP serverExtended access496

watch-cli is a skill and CLI that hands an agent the frames and transcript of any social video so it can read the clip itself.

SkillExtended access260

YouTube Data MCP (kirbah)

YouTube Data MCP Server (kirbah)

YouTube MCP with trimmed replies: transcripts without a key, video and channel statistics, trends, comments, a MongoDB cache and growing-channel search

MCP serverExtended access33
Official

An agent in Cursor or Claude Desktop describes video, splits it into chapters and shots and finds moments by speech or picture

MCP serverExtended access14

README in short

The README starts with what sets the server apart: transcript, frames and metadata in one tool. It then covers installation for Claude Code, VS Code, Cursor, Claude Desktop and the CLI, a table of eight tools with examples, detail levels, a cache that lets bulk jobs resume, and supported sources. A separate section describes the guard against private-network requests and the Whisper fallback chain.

README badge

Report a problem