ChatCut Desktop brings AI video editing to Codex and Claude Code

Editing video with AI usually means wrestling with web timelines, dragging clips into complex browsers, and manually clicking export. ChatCut is taking a very different route by handing the timeline straight to your coding agents.

The team just rolled out ChatCut Desktop, introducing local Model Context Protocol (MCP) support that plugs directly into agents like OpenAI's Codex and Anthropic's Claude Code. Instead of treating video generation as an isolated task, your agent can now treat video editing as just another tool call.

Giving coding agents a video timeline

The standout feature here is how deep the agent integration actually goes.

Once installed, ChatCut Desktop runs a local STDIO MCP server named chatcut_desktop on your machine. The app automatically registers its core tools and bundled skills with supported local agents, bypassing the need for manual config tweaking or UI automation hacks.

From there, agents can query active projects, inspect billing credits, and execute full editing workflows. The headline capability is reference-video recreation, also known as remaking or cloning footage.

Through the bundled video-gen skill and the submit_video_remake tool, an agent can take an existing reference clip and preserve its motion paths, cuts, and timing while swapping out actors, products, clothing, or entire environments. If you already have raw footage, the agent can also load editing skills to match the pacing, captions, and transitions of a target video automatically.

While the local skill handles workflow orchestration and tool definitions, the heavy lifting of video processing and generation runs through cloud MCP infrastructure.

Setup and platform limits

ChatCut designed the installation to be agent-friendly from the start, using native command-line scripts.

Mac users on macOS 13 or newer can install it with a simple curl script in Terminal, covering both Apple Silicon and Intel hardware. Windows users on 64-bit Windows 10 or newer get a single PowerShell command.

Linux is officially unsupported for now. Once installed, the desktop client handles Authenticode and Apple Gatekeeper signatures properly before guiding the user through browser-based sign-in for Google or email accounts.

Why it matters

We think turning a desktop video editor into an MCP server is a remarkably sharp move.

Until now, autonomous agents have been great at writing scripts, churning out code, and pinging text-to-video APIs. But stitching everything into a cohesive edit with precise timing still required human intervention inside traditional editing suites. By exposing a local MCP interface, ChatCut effectively gives agents hands inside the video editing timeline.

For solo creators, growth marketers, and e-commerce teams, this radically lowers the cost of turning one successful ad format into dozens of localized variations. You can ask an agent to duplicate a viral layout, swap in your product, and match the cut points without touching an editing tool.

That said, the reliance on cloud credits means heavy iteration could get expensive fast. And with Linux still off the table, developers hoping to run fully automated headless video pipelines on cheap cloud servers will have to wait.