Skip to main content

MCP server – Integrating Lexeri in LLMs´

Stefan Rohde
  • Updated

The Model Context Protocol (MCP) is an open standard which was developed to provide AI assistants with secure and structured access to external data sources and tools. MCP acts as a bridge between Large Language Models (LLMs) and various applications, databases or services.

Lexeri provides an MCP server that LLMs can use to access your termbase.

Functions of the Lexeri MCP server

The Lexeri MCP server provides AI assistants with a range of terminology management functions. External applications can use standardized MCP interfaces to interact directly with your Lexeri termbase.

The following functions are currently available as tools on the MCP server.

Available functions

  • Search and update terms
  • Create or propose terms
  • Create subject areas or entries for a subject area
  • Text term checks
  • Create usage examples for a term

Besides checking generated texts for correct terminology, these functions can also be used, for example, to extract terms with an LLM and suggest term candidates as term requests.

Configuring the Lexeri MCP server

Lexeri provides an SSE endpoint under the URL: https://mcp.lexeri.com/mcp/sse. An API token is needed for authentication with the Lexeri API. This articles explains how to create an API token in Lexeri: API tokens

Configuring the MCP Server in Claude Desktop

Claude Desktop supports the addition of integrations via an MCP server. The tools provided by the MCP server are then displayed in the Claude Desktop integration menu:

Claude Desktop currently does not support remote MCP servers, so an NPM package named "mcp-remote" is used for configuration. It provides the endpoints of a remote MCP server as a local MCP STDIO server for Claude Desktop. To learn more about mcp-remote, visit: https://www.npmjs.com/package/mcp-remote

The Lexeri MCP server must be configured manually by creating a configuration file. For this purpose, a JSON file named claude_desktop_config.json is saved in one of the following directories:

  • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
  • Windows: %APPDATA%\Claude\claude_desktop_config.json

The configuration file may look like this, for example:

{
  "mcpServers": {
    "Lexeri": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://mcp.lexeri.com/mcp/sse",
        "--header",
        "Authorization: Bearer ${AUTH_TOKEN}"
      ],
      "env": {
        "AUTH_TOKEN": "<LEXER_API_TOKEN>"
      }
    }
  }
}

The text “<LEXERI_API_TOKEN>“ must be replaced with the API token created. This configuration example uses npx to execute the NPM package mcp-remote.

Lexeri MCP server in the ChatGPT API

OpenAI offers support for MCP Remote Server in the Responses API. You will find a detailed description of this in the OpenAI documentation: https://platform.openai.com/docs/guides/tools-remote-mcp

A request can be made in Python, for example, as follows:

from openai import OpenAI

client = OpenAI()

resp = client.responses.create(
  model="gpt-4.1",
  input="Check the following text for correct terminology...",
  tools=[
    {
      "type": "mcp",
      "server_label": "lexeri",
      "server_url": "https://mcp.lexeri.com/mcp/sse",
      "headers": {
        "Authorization": "Bearer $LEXERI_API_TOKEN"
      }
    }
  ]
)

print(resp.output_text)

Use of MCP servers may require approval. The workflow for this purpose is described in the OpenAI documentation.

Related articles