diffusionstudio /
editor
A video editor where every edit is code, so coding agents can cut and render footage
- 3.4k stars
- 188 stars since 3 Oct
- TypeScript
- Updated 4 days ago
- Some experience
- Image, video & audio
Editing video with an agent is hard because the agent cannot see or hear the footage, and most editors are not built to be driven by code. Diffusion Studio stores each project as a folder of JSX files. The agent edits the code, and the editor canvas redraws.
It works in both directions. If you drag a shape on the canvas or trim a clip on the timeline, the change is written back to the code on the element that owns it, so the agent always sees the current version. The README likens it to an IDE that renders a video canvas instead of text.
The desktop app ships a diffusion command-line tool, which is also exposed as MCP tools. They let an agent probe media, grab frames, view waveforms, transcribe speech, ask a multimodal model about a clip, and check a composition for mistakes like black-frame gaps. It works with Claude Code, Codex, Cursor, Copilot, Gemini CLI, OpenCode or Pi.
What it does
Compositions as JSX code
A project is a folder of SolidJS-style JSX with a stage, scenes, video, text and more. Saving the file recompiles it and mounts it into the editor.
Two-way editing
Changes made on the canvas or timeline are written back to the element's code by id. A person can pick up exactly where the agent left off.
Media inspection tools
Commands like probe, grab, filmstrip, waveform, transcribe and listen let an agent examine footage it cannot watch. Each returns one JSON object.
Generative assets in code
generate.imageandgenerate.videodeclare AI-made images and video in the composition and place them in the timeline. Voiceover is also listed.Check and capture
diffusion capturerenders frames of a scene as an export would.diffusion checkreports structural problems such as gaps or failed sources.
What you could build
- Drop raw footage in a folder and ask an agent to cut it, remove filler words, add subtitles and render.
- Build motion graphics such as product promos, explainers and title sequences from a prompt.
- Clip a highlight from a long video and reformat it vertically for social.
- Ask for summaries, scene search or quotes with timestamps from a video.
- Script a video composition by hand in JSX and fine-tune it on the visual timeline.
Get started
Download the desktop app for macOS (Apple Silicon) or Windows (x64) and follow its setup, which registers its MCP server with your agent. Then ask your agent in plain language to edit a folder of footage. The README also lists a skills install command.
npx skills add diffusionstudio/skillsCommands are copied from the project's README. Read it on GitHub for the full setup.
More in Image, video & audio
VoiceStudio
Desktop app for cloning voices, dubbing video and dictating, run on your own machine
voicebox
Local voice studio for cloning voices, generating speech and dictating into any app
upscayl
Desktop app that enlarges low-resolution images using AI, on Linux, macOS and Windows