The Lexeri MCP server brings your termbase directly into AI assistants such as Claude Desktop, Claude Code, or any other tool that supports the Model Context Protocol (MCP). You can use it to look up terms in the middle of your workflow, check texts against your terminology, propose new terms, or create tasks – all without leaving your assistant.
This guide explains what the server can do, how to set it up, and what to keep in mind. The Model Context Protocol (MCP) is an open standard designed to give AI assistants secure and structured access to external data sources and tools. MCP acts as a bridge between large language models (LLMs) and a wide range of applications, databases, and services.
Setup guides for other clients can be found here:
- Use Lexeris MCP server in Claude Desktop
- Add Lexeris MCP server to agents in Microsoft Copilot Studio
- Connect Lexeris MCP server to GitHub Copilot
- Use Lexeris MCP server in ChatGPT Responses API
What is the Lexeri MCP server?
MCP is an open standard that lets AI assistants access external tools in a controlled way. The Lexeri MCP server is the official bridge between MCP-capable assistants and your Lexeri terminology. Once it is connected, the most important Lexeri features become available directly in the chat.
The server is reachable at:
https://mcp.lexeri.com/mcp
Requirements
You need:
- A Lexeri account with access to the termbase you want to use.
- An MCP-capable client (for example, Claude Desktop or Claude Code).
- An API token, if you do not want to sign in via OAuth. Tokens can be generated in Lexeri under your user account. Make sure to grant the token only the permissions (scopes) you actually need.
Security note: An API token grants the same permissions as your Lexeri login. Treat it like a password and never store it unencrypted in shared documents.
Features of the Lexeri MCP server
The Lexeri MCP server provides AI assistants with a range of terminology management features. Through standardised MCP interfaces, external applications can interact directly with your Lexeri termbase.
The server exposes about 30 functions, grouped into the following areas:
Searching and looking up terms
Search terms by keyword, language, status, or topic, retrieve detailed information about a term, and view usage examples and definitions.
Checking texts
Run a check on any text against your termbase. The assistant identifies the terms used, flags non-recommended or outdated ones, and suggests the preferred alternatives – across languages, if you wish.
Capturing new terms
Create new terms directly, or propose them for review via a term request. Definitions, usage examples, part of speech, grammatical properties, and source references can be supplied along with the term.
Driving workflows
Create and populate tasks to edit terminology without publishing changes immediately. Manage and review term requests.
Reviewing term extractions
Review automatically extracted term suggestions from documents and confirm or reject them individually.
Structuring the termbase
Create topics, assign term entries to topics, group synonyms and translations under a shared term entry, and add custom fields.
Inspecting imports
List existing imports and review their contents.
Which features are actually available to you depends on the permissions of your token or your Lexeri user.
Manage ontologies
List and view ontologies, create relation types, link term entries via typed relations, and retrieve related term entries.
Permissions and security
- Permissions match exactly those of your Lexeri user. No additional privilege elevation takes place.
- All communication is encrypted via HTTPS.
- The server does not store any content or tokens persistently. Every request is forwarded directly to the Lexeri API.
- Recommendation for production use: create a dedicated token per application with minimal scope, rotate it regularly, and provide it through a secrets manager (e.g. Vault, 1Password CLI, GitHub/GitLab Secrets) – do not check it into source control in plain text.
Frequently asked questions
Which termbase is used? The termbase is determined by the token or the OAuth login you use. If you have access to multiple termbases, the one associated with the token applies.
Does the server work with other AI tools? Yes. Any client that supports the Model Context Protocol can connect. Claude Desktop and Claude Code are the most common clients, but other assistants and your own implementations are also possible.
Tools do not appear in the client Please check that the URL is set correctly to https://mcp.lexeri.com/mcp and that the client has been restarted. If you are using an API token, make sure it is valid and has the necessary scopes.
An action is rejected The token in use does not have the required permissions. Generate a new token in Lexeri with the appropriate scopes (for example, edit, edit_terms, or manage_term_requests) and replace the existing one.
An input is not accepted This is usually a validation error – for example, a duplicate term, an unknown locale code, or a missing required field. The server's detailed error message names the affected field.