sarvam

Sarvam MCP Server provides language-aware text processing powered by the Sarvam API. It exposes tools for language identification, deep text analytics, translation with control over speaker attributes and scripts, and transliteration between scripts. You can send structured requests to the server's tools to identify the language and script of input text, analyze content and answer questions about it, translate text with options like source/target language codes, speech gender, mode and script, and transliterate text between scripts with optional numerals handling and spoken form considerations. This makes it suitable for building multilingual Indian language applications that need robust NLP capabilities without managing multiple services separately.

To use the server, run it locally (via npx) or via Docker where configured. The server expects specific input fields per tool (e.g., input, text, language codes, questions) and returns results accordingly. For example, use language-identification with input text to detect language and script; use text-analytics with text and questions to extract insights; use translate-text with input and source/target language codes to translate content; and use transliterate-text to convert text between scripts while optionally handling numerals and spoken forms. Ensure you provide your Sarvam API key via the SARVAM_API_KEY environment variable when running the server so requests to the Sarvam API are authenticated.

Prerequisites:\n- Node.js (LTS) and npm installed on your machine.\n- Optional: Docker if you prefer running the Docker image.\n\nInstall from npm and run locally:\n1) Ensure Node.js and npm are installed. Verify with:\n node -v\n npm -v\n\n2) Install the Sarvam MCP Server package:\n npm install -g sarvam-mcp-server OR npm install sarvam-mcp-server (local project install)\n\n3) Start the server (this will use npx by default as configured):\n sarvam-mcp-server OR npm run start (depending on setup)\n\n4) Provide your Sarvam API key via environment variable when starting the server, e.g.:\n SARVAM_API_KEY=YOUR_API_KEY_HERE npm start\n\nRunning with Docker:\n1) Pull the image:\n docker pull shbhtngpl/sarvam-mcp-server\n\n2) Run the container with your API key:\n docker run -i --rm -e SARVAM_API_KEY=YOUR_API_KEY_HERE shbhtngpl/sarvam-mcp-server\n\nNote: If you use Claude Desktop integration, you can reference the same environment variable (SARVAM_API_KEY) in your Claude config as shown in the README.

Environment variables and configuration tips:\n- SARVAM_API_KEY is required to authenticate with the Sarvam API. Do not commit your key to public repos.\n- When using Docker, pass the API key with -e SARVAM_API_KEY or via the environment in your orchestration tool.\n- If you need to run multiple instances, ensure unique port mappings or use a reverse proxy to avoid conflicts.\n- The available tools (language-identification, text-analytics, translate-text, transliterate-text) expect specific input fields; consult the README for the exact parameter names.\n- For translation and transliteration, you can control scripts, numerals formats, and additional processing options to tailor outputs for your target audience.\n- If you encounter API rate limits, consider increasing timeouts or caching frequent queries on the client side.