Skip to main content
Kino runs agents inside the desktop app. The agent is the runtime that searches footage, edits the timeline, and writes motion graphics; Kino is the studio around it, holding the project, the media library, the canvas, and the export pipeline. Kino does not bill or resell model usage for Claude Code and Codex. Sign in once and messages run on your Claude or ChatGPT plan, not Kino credits.
Choosing an agent in Kino: Kino paid via Kino plan, Claude Code on your Claude subscription, or Codex on your ChatGPT subscription

Agent comparison

The agent decides the runtime, setup path, model access, and billing owner.

Working in chat

Whichever agent you run, you talk to it in chat. Chat is a great place to ask for an edit and a terrible place to watch one. Ask here; results land on the timeline.

Ask questions

“What clips have Sarah in them?” or “Show me all the outdoor shots”

Tag people

@mention people in your library to reference them in conversation

Drag in clips

Drop videos directly into chat to ask about specific footage

Build timelines

Ask Kino to add clips to your timeline: “Add this to the rough cut”

Example conversations

Show me all clips where @Brian is speaking about the product launch
Found 4 clips matching your search…
When you @mention someone, Kino searches across footage containing that person, even if they’re not speaking. Drop several videos into chat at once to ask Kino to compare them, find the best take, or combine them into a sequence.

MCP

External agents can also drive the open project from outside the app over Kino’s MCP server, including Codex, Cursor, and Claude Code running in a terminal.