FFmpeg Skill

•
9.1k
•

Equip coding agents with a local video editor providing 42 FFmpeg-powered tools and an MCP interface without requiring cloud services or API keys.

FFmpeg Skill is a local video editing toolkit and MCP server designed for coding agents such as Claude Code, Cursor, and Codex. It replaces brittle, hand-crafted shell commands with 42 strongly typed, structured Python tools that strictly follow a "probe → edit → verify" workflow. Agents can cut, trim, reframe, adjust audio loudness, generate captions, burn karaoke animations, and export across platform presets—all locally without external API dependencies. Built on a Self-Producing Execution Contract (SPEC), the tool automatically derives schemas directly from CLI parsers to prevent drift, enabling reliable, verifiable video production natively on local machines.

Key Features

015,538 GitHub stars
02Verification-driven workflow combining pre-edit probing, post-edit format validation, and visual contact sheets
03Model Context Protocol (MCP) server support exposing tools directly to AI coding agents
0442 local video and audio editing tools spanning cuts, reframing, captions, loudness normalization, and multicam sync
05Self-Producing Execution Contract (SPEC) ensuring tool schemas strictly match CLI arguments without drift
06Zero cloud dependencies and no external API keys, leveraging local FFmpeg and optional local Whisper speech engines

Use Cases

01Allowing coding agents to programmatically assemble, reframe, and normalize recordings with post-execution verification
02Automating multi-platform video exports with specific aspect ratios, safe zones, and caption placements for social media
03Removing pauses, silence, and filler words from raw audio and video interviews locally