🔌

Connect your AI tools

You can connect an AI tool — Claude, ChatGPT, Cursor or another — to Frontitude, so it can work with your copy instead of guessing at it. It gets exactly what you can see and edit, and nothing else.

‍

‍

Before you start

You need an AI tool that supports MCP. Claude, ChatGPT, Cursor, Claude Code, VS Code, Codex and Gemini CLI all do.

‍

💡 If your workspace administrator has turned AI tools off, you'll see a message saying so when you try to connect. Ask them to turn it back on in Settings → AI.

‍

‍

Connect

  1. Open Settings → AI in Frontitude.
  2. Choose your AI tool and copy the setup instructions.
  3. Paste them into your AI tool and let it set up the connection.
  4. Sign in to Frontitude when prompted, and choose which workspace the connection covers.

‍

‍

What it can do

Your AI tool can search your copy library, open a component to see its wording, character limit, variables, translation coverage and status, and see everywhere a component is used. If you allow it to, it can also create components, update values, set status, manage tags and variables, and leave comments.

‍

‍

What it can't do

  • It can't see anything you can't see. Same projects, same library, same access level.
  • It can't see your design files. It reads what has been synced into a project.
  • It can't create projects, pages or frames.
  • It can't manage collaborators or billing.
  • It can't email anyone about a comment it leaves, even one that @mentions somebody.

‍

‍

Staying in control

  • ‍Keep your AI tool's confirmation prompts on. You approve each change before it runs.‍
  • Give a connection only the tools it needs. On the Profile tab you can narrow a connection — to read-only, or to a specific set of tools. You can't widen it later beyond what you approved when you connected.‍
  • Disconnect whenever you like, from the same tab. It takes effect on the agent's next call.

‍

💡 One thing to know about how agents work: an agent acts on what it reads. If a piece of copy, a comment or a note contains something written as an instruction, an agent may follow it. This matters most when several products are connected to the same conversation. Connect the tools you'd be happy showing your copy library to, and keep confirmation prompts on.

‍

‍

If something doesn't work

  • "AI tools are turned off for this workspace" — an admin has switched them off in Settings → AI.
  • The agent can't see a project — check you have access to it yourself. The agent has exactly what you have.
  • It stopped working after a permissions change — that's expected. Access is checked on every call, so a change takes effect on the next one.

‍

Ready to try it out? See prompt samples for Frontitude MCP, or read the full documentation.

Copied!
🍪