The Model Context Protocol (MCP) server provides a standardized interface that allows any compatible AI model or agent to access your Caro data in a simple and secure way.

The Caro MCP server allows AI agents like Claude and Codex read your Caro data — headcount, people, cost centers, reporting hierarchy, and more — directly during a conversation.

With the Caro MCP, you can:

<aside> ⚠️

Heads up: data sensitivity. Caro contains confidential information (compensation, confidential headcount). Once connected, the AI agent can read what you can read in Caro, and that data flows into the assistant’s context and any third-party model provider behind it. Treat MCP responses with the same care you’d treat exporting data from Caro.

</aside>

Enabling the MCP server

<aside> 💡

API key access. OAuth is the standard way to connect. If you have a specific use case that needs API key authentication (e.g. CI integrations, automation scripts), reach out to us and we’ll help you set that up.

</aside>

Set up the Caro MCP in any client using the following instructions.

General

Claude Code CLI (Terminal)

Claude Desktop Application (Cowork, Chat)

Codex CLI

Cursor or other MCP-compatible IDEs


Verifying the connection

After connecting, ask the assistant: