Get Flisk

Docs Connect

Connect your AI (MCP)

Connect Claude, Codex, Cursor or any MCP client to Flisk, so your AI can put scenes on your screens and change them, as you.

Flisk speaks the Model Context Protocol, so your own AI (Claude, Codex, Cursor or any agent that speaks MCP) can put things on your screens and change them, as you: make a scene, show it on your TV, look at what’s there and fix it.

You’ll need a Flisk account: make one in the Flisk app on iPhone or Android first.

On your computer: through the command line#

The easiest way. Install Flisk’s command line, sign in, and connect:

curl -fsSL https://flisk.tv/install | sh
flisk login
flisk connect

flisk connect adds Flisk to Claude Code (or flisk connect codex for Codex). For Cursor, Claude Desktop and others, flisk connect show prints the setting to paste in. The agent runs flisk mcp, signed in as the command line is: there’s no token to copy, and flisk logout stops it.

Anywhere else: Flisk’s MCP server#

Address: https://flisk.tv/mcp (Streamable HTTP, JSON-RPC 2.0). Each request carries an access token of yours:

Authorization: Bearer <your token>

Make a token for each agent, named for what will use it, in the Flisk app (Settings, Connect your agent) or with flisk tokens new "Claude, laptop". It’s shown once. For example, in Claude Code:

claude mcp add --transport http flisk https://flisk.tv/mcp \
  --header "Authorization: Bearer <your token>"

Keeping it yours#

  • A token acts as you, on your screens only. It can’t reach anyone else’s, or read your messages to other people.
  • Treat it like a password: don’t put it in a public repository or a shared file.
  • Revoke one you no longer use: flisk tokens lists them, flisk tokens rm ID ends one, and anything using it stops at once.

Once it’s connected#

Ask your agent for something on one of your screens. It can start with Flisk’s explain tool, which says how to drive Flisk well, then Building scenes has the rest.

For agents: https://flisk.tv/llms.txt.

This page as Markdown