Lumen — the open-source AI video editor

Edit video with an AI that can see, hear and cut.

Vibecoded with Claude Opus 5.5 by Mas-Inx

Lumen is a desktop video editor where the AI isn't a chatbot bolted onto a timeline. It watches your edit frame by frame, listens to the mix, reads what's said — and makes real, undoable edits with the same tools you use. Bring your own brain: Claude Code, Codex, or a model with your OpenAI, Anthropic, Gemini, OpenRouter or local Ollama / LM Studio key.

 

Why Lumen

• An AI that edits, not just chats. Every edit in Lumen is a typed command, and the AI gets all of them — plus tools to look and listen. About 70 tools in all, each change a normal undo step. See them all →

• It sees what you'll export. When the AI checks a frame, it's rendered exactly as the export will be: every track, title, effect and transition. It can watch the whole edit as a contact sheet, look inside footage, and hear loudness, silences and speech.

• Bring your own brain. Use the Claude Code or Codex you already pay for, an API key from any major provider, or a local model. Lumen is also an MCP server, so any agent can drive it.

• Local-first. Media is edited in place and never uploaded. Speech-to-text runs on your GPU with Whisper. No account, no watermark, no telemetry.

• A real editor underneath. A magnetic timeline, keyframes, 3D titles and transitions, colour looks, frame-exact export with hardware encoding — so you can finish by hand whatever the AI starts.

It's an open-source alternative to editors like CapCut, Filmora and Descript — built so an AI can do anything you can.

Related MCP server: ViralSpin MCP

Contents

• Features — editing · Copilot · smart edits · generation · export · integrations

• Download · Build from source · Documentation · Roadmap · Contributing · Credits

Features

A real editor

• Magnetic main track that keeps your story edge to edge, with free tracks above and below for B-roll, overlays, titles, captions, voice, music and sound effects.

• Split, trim, ripple, slide into gaps, snapping, markers, speed and reverse, and keyframes with easing on every property.

• 2D and 3D transitions (cube, flip, door, page curl, shatter…), effects, colour looks and grading, and extruded 3D titles rendered live.

• Sound that drives the picture: a Web Audio mix with per-clip volume, fades and keyframes, Studio voice clean-up, on-device noise reduction and 15 built-in sound effects.

• Import MP4, MOV, WebM, MKV, MP3, WAV, FLAC, PNG, JPEG, WebP, GIF and more — by dialog, folder or drag and drop. Files stay where they are; missing ones can be relinked.

• .lumen project files that survive moving folders between drives, crash recovery, and undo for everything. Ctrl+K reaches every command.

The Copilot sees, hears and edits

Ask in plain words — or talk to it. The Copilot plans, uses Lumen's tools, and checks its own work:

You see every step, including the frames it looked at, and Undo all reverts a whole answer.

What the AI sees when it watches your edit — one call to get_contact_sheet:

Brains: your own Claude Code or Codex (Lumen runs them headless, with only Lumen's tools), or a model with your key: Anthropic, OpenAI, Google Gemini, OpenRouter, OpenCode Zen, Ollama, LM Studio or any OpenAI-compatible endpoint. Keys are encrypted by Windows (DPAPI) and never leave the app's main process.

Smart edits

One-click suggestions on the timeline, and tools for every AI:

• Transcribe on the device with Whisper (GPU via WebGPU, or CPU), or with OpenAI or ElevenLabs.

• Remove pauses — the whole timeline closes up around each cut, so captions, B-roll and markers stay in sync while music plays on; with several speakers it only cuts where everyone is quiet.

• Captions from the transcript, on their own track.

• Duck music under speech with volume keyframes.

• Reframe between landscape, vertical, square and portrait.

• Hear the mix — loudness in LUFS against the −14 streaming target, peaks, clipping and silences.

Generate 3D, motion graphics, images, video and voice

• Blender — real extruded 3D titles, looping abstract backgrounds and AI-written scenes, rendered in the background and dropped into your media.

• HyperFrames motion graphics — lower thirds, title cards, kinetic type, stat counters or any HTML + GSAP composition — rendered inside Lumen.

• Images and video from your OpenAI key (GPT Image) or Gemini key (Gemini image models, Veo video), with model lists fetched live — new models show up without an update.

• Voiceovers, sound effects and music with ElevenLabs.

• Anything from an MCP server — Higgsfield, Runway, Replicate, fal.ai or your own.

Effects, transitions and looks

2D and 3D transitions, 15 effects (glow, grain, shake, RGB split, 3D tilt, curved screen…), one-click colour looks and full grading — each previewed on your own footage before you apply it.

Export

From 480p to 4K in your project's shape, 24–60 fps, with hardware encoding where your GPU supports it, a live file-size estimate and progress in the taskbar.

Connect anything

• Lumen is an MCP server. Turn it on and any agent — Claude Code, Claude Desktop, your own — gets the same ~70 tools, pictures included:

  claude mcp add --transport http lumen http://127.0.0.1:47910/mcp --header "Authorization: Bearer <token from Lumen>"

• Lumen is an MCP client. Add any server by URL or command; generator tools show up as models in the Generate panel.

Download

Download the latest Lumen for Windows →

Windows 10 or 11, 64-bit. The installer lets you install just for yourself (no admin needed) or for everyone, and associates .lumen project files.

Optional: Blender 4.2+ for 3D renders. Everything else is built in; the Whisper speech model downloads once (about 200 MB) the first time you transcribe on your device.

Build from source

You need Windows 10/11 and Node.js 24.

git clone https://github.com/Mas-inx/lumen-ai-video-editor.git
cd lumen-ai-video-editor
npm install
npm run dev      # the app, with hot reload

Documentation

• AI tools reference — every tool the AI can use, with parameters

• Architecture — how the editor, media engine, AI and integrations fit together

• Contributing · Changelog · Security · Third-party notices

Roadmap

Ideas the project is heading towards — contributions welcome:

• macOS and Linux builds

• Code-signed installer and automatic updates

• Exporting a marked range, and render queues

• Streaming audio for hour-long recordings

• More languages for the interface

Contributing

Bug reports, ideas and pull requests are welcome — start with CONTRIBUTING.md. Every edit in Lumen is a command in src/editor/commands.ts, and every AI tool lives in src/integrations/tools/, so adding a feature for people usually adds it for the AI too.

Credits

Lumen was vibecoded with Claude Opus 5.5 in Claude Code by Mas-Inx.

License

MIT © 2026 Mas-Inx. Lumen builds on excellent open-source work — see THIRD-PARTY-NOTICES.md.