KyaniteLabs/kinocut

View on GitHub ↗

Guardrailed video editing MCP server for AI agents. FFmpeg, Hyperframes, repurposing tools, Python client, and CLI. Local, fast, free.

118 ★28 forksPythonUpdated 18d ago

What you need to know

A free, local-first, guardrailed video editing MCP server (plus Python client and kino CLI) that lets AI agents trim, caption, repurpose, and quality-gate video with typed FFmpeg tools.

Install

pip install kinocut (requires FFmpeg on PATH); optional extras such as pip install "kinocut[transcribe]", kinocut[image], kinocut[stems]

Usage

  • Add to any MCP client, e.g. claude mcp add kinocut -- uvx --from kinocut kino, then ask your agent to trim, caption, convert to vertical, and quality-check video.

Key features

  • 194 MCP tools and 165 CLI commands
  • Video Receipts with per-step hashes and resume
  • Preflight guardrails and quality/release checkpoints
  • Workflow engine with validate, plan, render, and inspect
  • Still/plate and image editing tools
  • Repurposing into Shorts, Reels, and TikTok packages

Best for

AI agent builders and local media operators doing repeatable, reviewable video editing workflows.

Caveats

  • Requires local FFmpeg; speech, image, stems, upscale, and audio features are optional extras; native MCPB desktop packaging is still staged.
Platforms: macOS · Linux · WindowsClients: Claude Code · Claude Desktop · Cursor · Windsurf · Cline
Documentation ↗

Reviewed 2026-08-11

Topics

agent-toolsai-agentsai-videoclaudeclaude-codeclicursorffmpeghyperframesmcpmcp-servermcp-toolsmedia-automationmodel-context-protocolpythonpython-librarysubtitlesvideovideo-automationvideo-editing
Stars
118★
Forks
28
Language
Python
License
Apache-2.0
Created
2026-03-21
Last push
2026-08-17