Concat: an open-source, local-first CapCut replacement that Claude Code can drive through a built-in MCP server
A Rust video editor with local Whisper captions, text-to-speech and voice cloning, background removal, keyframes and 4K export, with no account, watermark or upload. The part we care about: `concat --mcp` lets an agent cut a project with the same commands the window uses.
Concat is an AGPL-3.0 editor for macOS, Windows, Linux and Android, at about 10.8k GitHub stars on Oct 22. It has been climbing since it hit trending lists in mid-October (one curator logged about 1,200 stars in a day on Oct 16), and v0.2.8 shipped on Oct 20. It covers the usual CapCut jobs: auto-captions with local Whisper, text-to-speech and voice cloning, background removal, keyframes, effects and titles, multi-track cutting and export up to 4K60. The models download once from Settings and then run offline, so footage never leaves your disk.
The agent angle is what makes it different from other open editors. The app ships an MCP server: add it to Claude Code with `claude mcp add concat -- <path to concat> --mcp`. The agent can create projects, import media, cut, add titles and captions, keyframe, apply effects, look at frames and export. There is also a JSON-RPC API for scripts. In 'window' mode (turn on Settings > Automation > Scripts and apps) the agent goes through the open app and can't save over the project on your screen. 'Headless' mode runs the engine with no window, and project folders are locked so the two don't clash.
Why it matters: most agent editing today means Remotion or HyperFrames code, or raw FFmpeg. Concat gives an agent a real timeline that a human can then open and tweak by hand, which is closer to how editors actually hand off work.
Caveats: it's beta. Kunal's thread mentions a report of auto-captions landing about 7 seconds early on 0.2.7 for Windows, so check caption timing before export. The 0.2.7 notes also admit a blur or glow export stall on Windows is still open. Mac builds are unsigned. AGPL matters if you plan to build a hosted service on top of it. We haven't run the MCP flow ourselves yet.
Related tools on ReelStack
Skills & repos in our directory
- Concat · MCP, Code-to-video engine
- EffectCraft · Repo, Motion graphics
- VoxCPM2 · Repo, Audio & music
- Kapwing MCP · MCP, Audio & music
- gemini-api CLI (Google) · Repo, Generation
- Eyeline (YouTube AI Video Editing System) · Repo, Code-to-video engine
More from the Lab
- Kapwing's MCP lets Claude or Codex caption, translate and voice-clone-dub a video, and the team published three slash-command skillsBased on the creator’s video · not yet tested by us
- Sirio open-sourced his YouTube editing system: a teleprompter that logs your takes, an AI cut you review, and 267 animated boardsSpotted, not yet tested
- Voxtral 3 for video people: cheaper captions that know your product names, and narration with emotion tagsSpotted, not yet tested
- Tandem: an MIT video editor with a CLI and MCP, where agent edits stay purple until you've watched themBased on the creator’s video · not yet tested by us
Topics: Audio, voice & music · Video MCP servers