Run the whole AI clipping pipeline — find viral moments, caption, reframe to 9:16 — from your terminal, over the Katto API. The same agent lives in the app, an MCP server, the REST API, and this CLI.
Status: v0.1.0 scaffold — core commands work against the live API.
katto uploadand richer progress land next.
npm i -g katto-cli # installs the `katto` command
# or one-off: npx katto-cli <command>Get a key at katto.tech/dashboard/api-keys (sk_live_…), then:
katto login # stores it in ~/.katto/config.json (chmod 600)
# or, non-interactive:
export KATTO_API_KEY=sk_live_xxxkatto clip https://youtube.com/watch?v=… --lang en --clips 5 --wait
katto jobs
katto job <id>
katto clips <id>
katto transcript <id>
katto usage
katto cancel <id>
katto mcp # prints MCP client config for Claude/Cursor/…Add --json to any command for machine-readable output. Every clip sends an idempotency key automatically.
Prefer a chat? The same account works over MCP — point Claude/Cursor/ChatGPT at https://mcp.katto.tech (katto mcp prints the snippet), or the REST API.
MIT · katto.tech