Weblate MCP

What mmntm/weblate-mcp MCP server does

The mmntm/weblate-mcp MCP server exposes Weblate translation management through the Model Context Protocol. It lets an MCP client query projects, components, languages, translation units, statistics, and recent activity without requiring the assistant to navigate Weblate's web interface directly.

Supported workflows include listing projects and their components, finding translations by key or text, reading a translation for a specific key, and writing one or many translation values. Project and component statistics provide completion and string-count information, while language and user statistics support broader progress and contribution analysis. Change-history tools can filter recent activity by project, component, or user.

How it works

The server acts as an MCP-facing layer over Weblate's REST API. An MCP client sends tool calls to the server, and the server uses the configured Weblate instance and API token to perform the corresponding operation. Results are returned to the client for use in an assistant conversation or an automation workflow.

Three transport options are described: STDIO for local MCP client integrations, HTTP/SSE, and Streamable HTTP. The development server uses port 3001 by default and exposes an HTTP MCP endpoint at /mcp when configured for HTTP access.

For translation discovery, searchUnitsWithFilters uses Weblate's native filtering syntax. This supports queries such as finding untranslated strings with state:=0, locating text containing a term with source:"login", or combining a component and state condition. This approach is intended to avoid fetching every key and checking entries individually.

Setup and configuration

The mmntm/weblate-mcp MCP server requires Node.js 18 or newer for the documented development setup, along with access to a Weblate instance and its API. The published package can be launched with npx:

npx -y @mmntm/weblate-mcp

Configure the MCP client with these environment variables:

• WEBLATE_API_URL: the Weblate API base URL.
• WEBLATE_API_TOKEN: the API token used to access Weblate.

The development workflow uses pnpm to install dependencies, build the TypeScript project, and start the server. Optional runtime settings shown in the documentation include PORT, NODE_ENV, and LOG_LEVEL.

Claude Desktop can run the server through the npx command and environment configuration. HTTP clients such as Cursor can connect to a locally running endpoint using http://localhost:3001/mcp.

Tools and capabilities

Available tool groups cover:

• Project listing and project management.
• Component listing and configuration details.
• Translation search, lookup, updates, and bulk writes.
• Language listing for a project.
• Project, component, translation, language, and user statistics.
• Recent changes across projects or filtered by project, component, or user.

The documented tool list includes listProjects, listComponents, searchUnitsWithFilters, searchStringInProject, getTranslationForKey, writeTranslation, bulkWriteTranslations, findTranslationsForKey, and several statistics and history tools. Write operations can update translation values, and the write tool includes approval support.

Limitations and notes

The server requires a reachable Weblate instance and a valid API token; the README does not describe an unauthenticated mode. The provided material documents Weblate API access and the listed MCP tools, but does not establish support for other translation platforms. HTTP client setup assumes that the server is running locally or is otherwise reachable at the configured MCP URL.