The interface is localized. Skill descriptions and documents are machine-translated by AI into your language; code, commands, paths and technical terms stay in the original, and the author's document governs.
← AI on an actual editing timeline
Open source · Local-first · Agent / MCP

OpenChatCut · Conversational AI video editing.

Turn natural language into edits on a real timeline. Video tracks, captions, transitions and audio remain editable, with previews and undo under your control.

Opens localhost:5199 on your own computer. Install and start the open-source edition first. This site does not connect automatically or host a shared online editor.

AI on an actual editing timeline

01Real multitrack edits, captions, transitions and undoable changes.

02Transcript-based cuts, pause cleanup, caption and audio previews.

03The built-in agent and Codex / Claude Code share the same project through MCP.

04Export video, audio, SRT, FCPXML or editable projects as supported by the release.

Start with a small project

  1. Choose the open-source build for your OS from the author's GitHub Releases and read the release notes.
  2. Launch locally, create a project and import a short video you have permission to use.
  3. Describe the edit, review the proposal and preview, then approve and export. Keep media and project backups.
Make this interview a 30-second cut. Remove repetition and obvious pauses, add captions and fades. Submit a draft for preview; do not apply or export it until I approve.

Developers: run from source

The open-source build requires Node.js 24.x and npm. This is the official POSIX shell example; on Windows, copy the file instead of using cp. Inspect source and dependencies before running. Configure only the services you need; keys stay on your local server. Installation and editing were not tested here.

git clone https://github.com/0xsline/OpenChatCut.git
cd OpenChatCut
npm install
cp .env.example .env.local
npm run dev

Connect Codex / Claude Code

Start the local open-source editor and open your project before adding its Streamable HTTP MCP. Prefer manual edit sessions: begin_edit_session → review_edit_session, then preview and approve in the editor. Examples contain no keys. Use the editor's local connection settings if authentication is required; do not submit tokens to this site.

Agent Skill

npx skills add 0xsline/OpenChatCut

Codex · config.toml

[mcp_servers.openchatcut]
url = "http://localhost:5199/api/external-mcp/mcp"

Claude Code · CLI

claude mcp add --transport http openchatcut \
  http://localhost:5199/api/external-mcp/mcp

Local-first is not always offline

Projects and media are local by default. Cloud models, transcription, generation, media search and remote storage send relevant inputs to the services you select. Model or API usage may cost extra. Use authorized media and keep backups.

The open-source edition uses AGPL-3.0-or-later, not MIT. Redistribution, modifications and deployment must respect its license and corresponding-source obligations; dependencies have their own terms. OpenChatCut Pro is separately licensed, so its trial and pricing do not define the open-source edition.

This site offers navigation and setup guidance, not installers, automated edits or media/key collection. The tool is not counted as a new Skill ZIP. Public hosting needs separate user isolation, authentication, storage and license compliance; do not expose a single-user MCP as a shared service.

Sources checked · 2026-10-14

Saved skills