LILT MCP Server

Enterprise AI translation with optional human verification, exposed as a remote MCP server so AI assistants can request brand aligned, multilingual translations.

Collaboration by LILT OAuth2 active

Connect (Remote Server) Docs

Overview

LILT is an enterprise translation and localization platform that combines adaptive AI translation models with expert human linguists. In March 2026 LILT launched a remote MCP server that lets AI assistants like Claude send text and files to LILT for translation, optionally route mission critical content (legal copy, technical manuals, marketing assets) to LILT's verified linguist network, and monitor job progress directly from a chat interface.

The server is hosted by LILT at https://mcp.lilt.com/mcp and exposes a focused tool set for translating text, uploading and translating files with human verification, checking credit balances, polling job status, and downloading completed jobs. Translations use customer specific adaptive AI models trained on company style guides and terminology, so output remains brand aligned across languages.

LILT MCP is provider hosted (no GitHub repo to self host) and is designed to work with any MCP compatible client. Authentication is handled through a sign in at mcp.lilt.com using SSO or email and password, with a separate self serve "LILT MCP App" account available for pay as you go usage alongside the existing LILT Enterprise offering.

Tools

Tool	Description
`translate_text`	Performs instant machine translation of a text string between supported language pairs.
`translate_files_with_verification`	Creates a formal translation job for uploaded files, optionally routed through LILT's human verification (expert linguist review).
`upload_files`	Uploads one or more files for use in a verified translation job. Returns success and error lists.
`check_job_status`	Returns the current status of a verified translation job (queued, in progress, completed, etc.).
`download_job`	Downloads the completed translated files for a verified translation job.
`get_credit_balance_information`	Returns the current LILT credit balance for the authenticated account.

Setup Guide

LILT runs the MCP server, so there is no package to install. You connect your MCP client to the remote URL.

Prerequisites:

An MCP capable client (e.g. Claude Desktop with Pro, Team, or Enterprise plan)
A LILT MCP App account at https://mcp.lilt.com (self serve, separate from lilt.com accounts) or a LILT Enterprise account

Claude Desktop (UI):

Open Settings, then Connectors
Click Add custom connector
Name: LILT
Remote MCP Server URL: https://mcp.lilt.com/mcp
Save, refresh, then click Connect and complete the SSO or email sign in at the LILT login page
Grant access when prompted

Generic MCP client config (using mcp-remote proxy):

{
  "mcpServers": {
    "lilt": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://mcp.lilt.com/mcp"
      ]
    }
  }
}

After installation, restart your client. The first call will trigger the OAuth style sign in flow at mcp.lilt.com. Future sessions reuse the authenticated session.

Use Cases

Translate AI drafted blog posts, social copy, or release notes into multiple languages from inside Claude without leaving the conversation
Route legal policies, technical manuals, or contracts to LILT's verified linguist network for human review while letting the assistant track the job
Quote and budget upcoming localization work by checking credit balance and job status before launching campaigns
Localize batches of documents (e.g. help center articles) by uploading files and downloading verified translations once jobs complete
Build agent workflows that combine content generation with brand aligned translation using a company's custom LILT adaptive models

Example Prompts

"Translate this product launch announcement into French, German, and Japanese using LILT."
"Upload these three PDFs to LILT and start a verified translation job into Spanish (LATAM)."
"Check the status of my LILT job 12345 and download the files when it's done."
"How many LILT translation credits do I have left this month?"
"Translate the attached privacy policy with human verification and notify me when reviewers finish."

Pros

Official, vendor hosted MCP server with no self hosting or local install required
Combines instant AI translation with optional human linguist verification, addressing the quality gap of pure machine translation
Uses customer specific adaptive models so translations follow corporate terminology and style
Job lifecycle tools (upload, status, download) let assistants manage end to end localization workflows

Limitations

Requires a paid LILT account (self serve pay as you go, or LILT Enterprise) plus a paid Claude plan for the official setup path
Supported language pairs, file formats, and rate limits are not enumerated in public documentation
Closed source and provider hosted, so there is no way to self host or audit the server implementation

Alternatives

DeepL API (community MCP wrappers exist) for fast machine translation without human review
Google Cloud Translation via community MCP servers for broad language coverage at lower cost
Smartling or Phrase, which offer similar enterprise localization platforms (no official MCP server at time of writing)