Skip to main content
Lexful can be exposed to AI agents and tools via the Model Context Protocol (MCP). This lets you use Lexful capabilities from compatible agent runtimes without hand-writing all of the plumbing. This page gives a high-level overview of how Lexful works with MCP and the endpoint you will use to connect.

MCP endpoint

Your MCP endpoint is:
Replace {POD} with the pod your account is hosted in. You can tell from the address you use to sign in to Lexful: Your account ID is the same value used for Lexful API access (for example, in the X-Account-ID header). You can find it in your account settings. Authentication to this endpoint is handled via OAuth. On first connection from a compatible MCP client, you will be prompted to authorize access.
If your account has IP restrictions enabled, they also apply to the MCP endpoint. Requests from an IP address outside your allowlist are rejected with a 403 error. Hosted clients such as claude.ai, ChatGPT, and Copilot Studio connect from their provider’s servers, not your network, so their IP addresses must be allowed for these clients to work.
If you are using a hosted demo environment, your team may provide a preconfigured demo account ID and corresponding MCP endpoint.

What MCP enables

At a high level, MCP allows you to:
  • Describe Lexful operations as tools that an AI agent can call.
  • Configure authentication and base URLs once, then reuse them across tools.
  • Standardize how agents access Lexful alongside other systems.
You can connect Lexful to:
  • Chat interfaces that expose MCP via Connectors or Apps UIs (for example, claude.ai or ChatGPT).
  • Developer tools that support raw MCP JSON config (for example, Cursor, Claude Code, or IDE extensions).
In chat UIs, you typically paste the MCP URL into a connector form and complete OAuth in the browser. In developer tools, you usually add a small JSON block to a config file or settings panel.

Where to go next

  • See Available tools for the full list of tools exposed by the MCP server.
  • See Workflows for everyday tasks you can ask an assistant to do.
  • See MCP setup for configuration details.
  • See MCP examples for client-specific setup instructions.