Setup & Installation

What It Does

At its core, the MCP Server acts as a bridge between Lara’s Models and the MCP ecosystem. It allows you to offload translation tasks from general-purpose language models to a specialized engine built specifically for translation, ensuring higher quality—especially for non-English content—and more consistent domain handling.

The server supports automatic source language detection, context injection, custom instructions, translation memories, and glossaries. If you plan to use instructions, please carefully read Adapt to Instructions.

Setup

For the hosted Lara MCP endpoint, OAuth 2.0 is the default authentication method: add the server URL and log in through your browser. Lara also supports Access Key authentication for clients or deployments that need it.

Hosted endpoint:

https://mcp-v2.laratranslate.com/v1

Claude

Go to Settings > Connectors.
Click Add Custom Connector.
Enter the name: Lara.
Enter the URL: https://mcp-v2.laratranslate.com/v1.
Click Add, then Connect.
Log in with your Lara Translate credentials in the browser.

ChatGPT

To integrate Lara Translate into ChatGPT:

Open ChatGPT on the web.
Go to Settings > Apps > Advanced settings and enable Developer Mode.
Go to Settings > Apps and click Create. In supported workspaces, admins/owners may instead use Workspace settings > Apps > Create.
Enter a name such as Lara and set the MCP server URL to https://mcp-v2.laratranslate.com/v1.
Complete the authentication configuration and log in with your Lara Translate credentials when the browser OAuth flow opens.
Click Scan Tools so ChatGPT can discover the tools exposed by Lara.
Click Create to save the app.
Start a new chat, enable or select Lara from the available apps/tools, and ask ChatGPT to translate something using Lara.

Note: Custom MCP apps use ChatGPT Developer Mode. Availability and allowed actions depend on your ChatGPT plan and workspace policy.

OpenAI Codex

Add Lara as a remote MCP server:

codex mcp add lara --url https://mcp-v2.laratranslate.com/v1

Authenticate with Lara:

codex mcp login lara

Complete the OAuth flow in your browser, then verify the configuration:

codex mcp list

Codex stores MCP configuration in ~/.codex/config.toml. The equivalent manual configuration is:

[mcp_servers.lara]
url = "https://mcp-v2.laratranslate.com/v1"

Cursor

For the current Cursor setup, see the Client Setup Guide.

If Lara Translate is available in your Cursor marketplace/plugin directory, you can install it there. Otherwise, use the manual MCP configuration described in the guide.

Claude Code

For the current Claude Code setup, see the Client Setup Guide.

If Lara Translate is available in the Claude Code plugin marketplace, you can install it there. Otherwise, use the manual MCP configuration described in the guide.

📘
Authentication tip
If the first OAuth attempt fails, log out of the Lara website and try connecting again.
If you see this error, it is the same known authentication issue:
{"error":"invalid_request","error_description":"refreshToken is required"}

Other Clients

For step-by-step OAuth setup on VS Code (GitHub Copilot), Windsurf, Cline, Continue, and other clients, see the Client Setup Guide.

If your client supports remote MCP servers and MCP OAuth, add the hosted Lara endpoint and follow that client’s authentication flow:

https://mcp-v2.laratranslate.com/v1

If your client requires Access Key credentials, STDIO, a local process, or another transport, use the configuration documented in the Client Setup Guide or the Lara MCP repository.

Verify It Works

After setup, test with a simple prompt:

Translate with Lara "Hello world" to Spanish

Your client should invoke Lara Translate and return the translation.

Available Tools

The hosted Lara MCP server exposes translation, language-detection, translation-memory, and glossary tools.

Translation

Tool	Description
`translate`	Translate text between languages with support for context, instructions, translation memories, glossaries, and translation styles.

Language Detection

Tool	Description
`detect_language`	Detect the language of a text or array of texts.
`list_languages`	List supported language codes.

Translation Memories

Tool	Description
`list_memories`	List translation memories in your account.
`create_memory`	Create a translation memory.
`update_memory`	Update a translation memory name.
`delete_memory`	Delete a translation memory.
`add_translation`	Add a translation unit to a memory.
`delete_translation`	Delete a translation unit from a memory.
`import_tmx`	Import a TMX file into a memory.
`check_import_status`	Check the status of a TMX import job.

Glossaries

Tool	Description
`list_glossaries`	List glossaries in your account.
`get_glossary`	Get details of a glossary.
`create_glossary`	Create a glossary.
`update_glossary`	Update a glossary name.
`delete_glossary`	Delete a glossary.
`add_glossary_entry`	Add or replace a glossary entry.
`delete_glossary_entry`	Delete a glossary entry.
`import_glossary_csv`	Import glossary entries from CSV.
`check_glossary_import_status`	Check the status of a glossary CSV import.
`export_glossary`	Export a glossary as CSV.
`get_glossary_counts`	Get the number of entries in a glossary.

Authentication

OAuth 2.0 (default)

OAuth is the default for the hosted endpoint. Add the server URL to a compatible client, complete the browser login with your Lara Translate account, and the client can connect without placing an API key directly in the MCP configuration.

For client-specific OAuth instructions, see the Client Setup Guide.

Access Key (alternative)

If you prefer to authenticate with Lara Access Key credentials instead of browser OAuth, configure the Access Key ID and Secret as described in the Client Setup Guide.

Translation Workflow

The main translation tool is translate. It accepts structured input, can process multiple text blocks, and can use source-language detection, context, custom instructions, translation memories, and glossaries.

For example, if you're translating the Italian phrase “la terra è rossa” while chatting with a tennis player, you can pass that context into the request. Instead of interpreting “terra” generically as ground or earth, Lara can understand that it refers to clay and return “The clay is red.”

Example

Prompt:
Translate with Lara: "la terra è rossa", I'm talking with a tennis player.

Request:

{
  "text": [
    { "text": "la terra è rossa", "translatable": true }
  ],
  "target": "en-US",
  "context": "Conversation with a tennis player"
}

Response:

[
  {
    "text": "The clay is red.",
    "translatable": true
  }
]

For advanced usage, local development, Access Key configuration, self-hosting, and the latest tool list, visit the lara-mcp GitHub repository.

What It Does

Setup

Claude

ChatGPT

OpenAI Codex

Cursor

Claude Code

Authentication tip

Other Clients

Verify It Works

Available Tools

Translation

Language Detection

Translation Memories

Glossaries

Authentication

OAuth 2.0 (default)

Access Key (alternative)

Translation Workflow

Example

More