Calibre MCP connects your AI assistant to your RUM, CrUX and Synthetic data. Ask about any page in plain language. Your assistant finds the slow element, script or URL, and a coding agent can then fix it in your codebase.
It works with ChatGPT, Claude, Claude Code, Codex, Cursor, GitHub Copilot and any other client that supports remote Model Context Protocol servers. It’s read-only, needs nothing installed, and is included on every plan.
Your assistant reads only what your Calibre account can see: the Sites in your Teams, in the organisation you choose when you sign in. See what it can access.
Set up#
Every client connects to the same address:
https://calibreapp.com/mcpThe first time your assistant uses Calibre, a browser window opens. Sign in, then choose the organisation to connect.
ChatGPT#
Add a custom connector in ChatGPT with the address above. Custom connectors depend on your ChatGPT plan.
Claude#
Add a custom connector in Claude with the address above. Or, in Calibre, go to Profile → Connected apps and choose Connect to Claude to open Claude with the connector filled in.
Claude Code#
claude mcp add --transport http calibre https://calibreapp.com/mcpThen run /mcp in Claude Code and choose calibre to sign in.
Codex#
codex mcp add calibre --url https://calibreapp.com/mcpThis opens a browser window to sign in. If you close it, run codex mcp login calibre to sign in later.
Cursor#
Save this as .cursor/mcp.json in your project, or as ~/.cursor/mcp.json to use it in every project.
{
"mcpServers": {
"calibre": {
"url": "https://calibreapp.com/mcp"
}
}
}GitHub Copilot#
In VS Code, save this as .vscode/mcp.json, or run MCP: Add Server from the Command Palette and paste the address.
{
"servers": {
"calibre": {
"type": "http",
"url": "https://calibreapp.com/mcp"
}
}
}Other clients#
Add https://calibreapp.com/mcp as a remote MCP server. Your client needs to support signing in with OAuth.
Share it with your team#
Commit the config file to your repository: .cursor/mcp.json for Cursor, .vscode/mcp.json for Copilot, or .mcp.json in the project root for Claude Code. Everyone signs in with their own Calibre account.
{
"mcpServers": {
"calibre": {
"type": "http",
"url": "https://calibreapp.com/mcp"
}
}
}What you can ask#
- “Which pages are failing Core Web Vitals?”
- “Why is LCP slow on
/checkout?” - “Fix the slowest interaction on
/search.” - “What shifts the layout on
/products?” - “Which script blocks the Add to cart button?”
- “How does mobile LCP compare to desktop?”
- “Which countries have the slowest LCP?”
- “Which Budgets are failing right now?”
- “Did yesterday’s Deploy make anything slower?”
- “Which metrics got worse after the last Deploy?”
Answers come from your own data and link to the matching report in Calibre.
Get better answers#
- Install RUM: answers about real visitors, slow interactions and countries need the RUM snippet. Without it, your assistant uses CrUX and Synthetic tests.
- Name your key elements: add a
data-nameattribute, such asdata-name="add-to-cart". RUM uses it to name the element, so your assistant can point straight to it. See Element attribution. - Record your Deploys: your assistant can then tell you whether each Deploy made your Site faster or slower, and which metrics regressed. RUM Deploys and Synthetic Deploys show the same comparison.
Next, see what your assistant can access.