Skip to main content
Glama
Sign In
Linux Video Editor MCP

Linux Video Editor MCP

by DeclanJeon
OverviewSchemaSourceRelated ServersDiscussionsAdmin
JavaScript
Local

Linux Video Editor MCP

A local MCP server for basic video editing on Linux, powered by FFmpeg. It provides a small set of editing tools to MCP-compatible agents such as Codex. It is an independent Linux implementation, not Higgsfield's official editor or a reverse-engineered copy of it.

Requirements

  • Linux

  • Node.js 18+

  • FFmpeg with libx264, AAC, MP3 (libmp3lame), and the subtitles filter enabled

  • An MCP client that supports stdio servers

ffprobe is optional. Media inspection falls back to FFmpeg's input report.

Related MCP server: mcp-ffmpeg

Install

git clone https://github.com/DeclanJeon/linux-video-editor-mcp.git
cd linux-video-editor-mcp
npm ci

Register it globally with Codex:

codex mcp add linux-video-editor -- node "$HOME/path/to/linux-video-editor-mcp/server.js"

Replace the example path with the actual clone location. By default the server only reads and writes under ~/Videos. To select another workspace, set VIDEO_MCP_ROOT in the MCP server environment. Restart Codex after registering.

Tools

  • video_probe — report media format, duration, and stream information.

  • trim_video — cut a time range and encode a new MP4 (high, medium, or small quality).

  • extract_audio — extract audio to MP3 (128, 192, or 320 kbps).

  • burn_subtitles — burn an SRT subtitle file into a new MP4.

All paths are relative to the configured workspace. Outputs must use the required file extension, must be in an existing directory, and cannot overwrite existing files. Inputs are left unchanged. Video encoding can take time and CPU, especially for large files.

Development / smoke test

node --check server.js
npm start

The MCP server communicates over stdio; do not print logs to stdout. A client can initialize the server and call tools/list to verify tool discovery.

Limitations

This is a deliberately small local editing toolkit, not a full timeline editor. It does not provide a graphical UI, video generation, cloud uploads, or Higgsfield account integration. Codec/filter availability depends on the local FFmpeg build. The subtitle filter rejects unsupported path characters to avoid FFmpeg filter-string parsing issues.

License

MIT. See LICENSE.

Available Tools

4 tools
burn_subtitlesBurn subtitles into videoA

Render an SRT subtitle file permanently into a video. Both files and output must stay in ~/Videos; saves a new MP4.

Inputs: input (required), output (required), subtitles (required).

extract_audioExtract audioB

Extract audio from a media file in ~/Videos and save it as MP3. Original is unchanged.

Inputs: input (required), output (required), bitrate.

trim_videoTrim videoA

Cut a segment from a video in ~/Videos and save a new MP4. Start and duration are seconds; original is unchanged.

Inputs: input (required), output (required), start_seconds (required), duration_seconds (required), +1 more.

video_probeInspect video or audio fileC

Return media metadata for a file inside the Videos workspace.

Inputs: file (required).

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Deploy Server
    One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

    TDQS

    B3.3/5.0

    Scored across 4 tools

    Disambiguation5/5

    Each tool targets a unique operation on a distinct aspect of media: probing metadata, trimming segments, extracting audio, and burning subtitles. There is no overlap in inputs or side effects, so an agent can pick the right tool without ambiguity.

    Naming Consistency4/5

    Three of four names follow a clean verb_noun pattern (trim_video, extract_audio, burn_subtitles). video_probe inverts this to noun_verb, a minor deviation that is still readable and unambiguous.

    Tool Count4/5

    Four tools is a focused, non-redundant set within the reasonable 3-15 range. However, for a self-described video editor the surface is slightly thin, leaving obvious common operations uncovered.

    Completeness3/5

    Core flows (probe, trim, audio extraction, subtitle burn) are covered, but there is no concat/merge, format conversion, resize/crop, or frame/thumbnail extraction—notable gaps for a video editing domain that agents would frequently hit as dead ends.

    Resources

    Related MCP Connectors

    Related MCP Servers

    • ff-toolkit

      inthepond
      A
      license
      A
      quality
      D
      maintenance
      Enables LLMs to perform FFmpeg operations like clipping, merging, extracting audio, adding subtitles, and transcoding videos via a set of tools exposed as an MCP server.
      5
      8
      MIT

    • mcp-ffmpeg

      kevinten-ai
      A
      license
      A
      quality
      D
      maintenance
      Provides 30+ FFmpeg video and audio editing tools via MCP, enabling AI assistants to perform operations like trimming, transcoding, overlays, and composition directly.
      27
      4
      MIT

    • ffmpeg-skill

      bayshier
      A
      license
      Not graded
      quality
      C
      maintenance
      Gives coding agents a local FFmpeg video-editing engine exposed over MCP, enabling them to probe, cut, join, reframe, caption, normalize audio, and verify the resulting media without cloud services or API keys. Every operation returns structured JSON and self-checks its own output against delivery specs.
      6,944 npm
      MIT