Cardle
Cardle MCP

Connect an AI assistant to your Cardle feed

Cardle has a built-in MCP server that lets Claude, ChatGPT, Gemini, or anything else that speaks MCP read and manage your cards directly. Ask it to add a reminder, check your feed, or tick off a checklist, and it can just do it.

How does connecting Cardle work?

Add Cardle's server address to a compatible assistant, authorize access to your Cardle account, then ask the assistant to save or retrieve a card. Use the same Google account in each assistant to reach the same feed. Connector availability depends on your assistant, plan and workspace settings.

The connection shares your saved Cardle items, not your full chat history. For a complete example, read how to save and revisit your first item.

1

Add Cardle as a connector

Using Claude.ai, Claude Desktop, or ChatGPT? This is all you need — no separate sign-in and no token to copy.

Server address: https://cardle.dev/api/mcp

Claude
ChatGPT
2

Try it

Ask your assistant something like:

It only sees and changes your Cardle data — nobody else's.

Using Claude Code, Gemini, or another assistant instead?

These take a token instead of signing you in directly.

  1. Open your Cardle vault, sign in with Google (top right), and click Connect AI next to your name.
  2. Give the token a name — e.g. “Claude Code” — and click Generate token.
  3. Copy it right away; it's shown only once. Keep it as secret as a password.
Claude Code
claude mcp add --transport http cardle https://cardle.dev/api/mcp \
  --header "Authorization: Bearer YOUR_TOKEN_HERE"
Gemini (CLI or app)

Add an MCP server entry pointing at https://cardle.dev/api/mcp, with an Authorization: Bearer YOUR_TOKEN_HERE header. The exact settings screen varies by version.

Back in the Connect AI panel, you can see every token you've created and revoke any of them any time — whatever was using it stops working right away.

Want to build your own tools on Cardle? Join the builder list and tell us what you'd build.

← Back to Cardle