estonia-ai-kit

Estonia AI Kit provides MCP servers that connect AI models to Estonian digital services and open data via the MCP (Model Context Protocol). The package suite includes multiple server implementations (e.g., for the Business Register, Tax & Customs, and Open Data portals) that expose government data and APIs in a standardized MCP interface so AI assistants like Claude, GPT, and other models can query structured data and perform actions through natural language prompts. You can start with the RIK (Business Register) MCP server to experiment with company and entity data, then explore EMTA (Tax & Customs) and Open Data MCP endpoints as your use case requires. The servers are designed to be typed, well-documented, and production-ready within the MCP framework, with attention to data privacy (e.g., anonymization where applicable) and modular integration.

To use the tools, install the desired MCP server packages from the Estonia AI Kit, then run them as MCP services in your environment. Once running, your AI model can request data via the MCP API, resolve external data sources (like the RIK dataset or Open Data portals), and incorporate results into its reasoning or responses. The toolkit emphasizes modern TypeScript implementation, strict type safety, and a monorepo workflow for sharing utilities and types across servers. Depending on the server, you may access real data feeds, mock data for demos, or curated datasets suitable for experimentation.

Prerequisites:

• Node.js (recommended v18+)
• Package manager (bun, npm, or pnpm)
• Git

Installation steps:

1. Clone the repository or install the MCP server package individually:

   • git clone https://github.com/your-org/estonia-ai-kit.git
   • cd estonia-ai-kit

2. Install the desired MCP server packages. For example, using Bun (recommended in this project):

   bun add @estonia-ai-kit/rik-mcp-server
   bun add @estonia-ai-kit/emta-mcp-server
   bun add @estonia-ai-kit/open-data-mcp-server
   

   If you prefer npm:

   npm install @estonia-ai-kit/rik-mcp-server @estonia-ai-kit/emta-mcp-server @estonia-ai-kit/open-data-mcp-server
   

3. Start the MCP server(s). The exact command may vary by package, but a common approach is to run using npx for quick startup as demonstrated:

   npx -y @estonia-ai-kit/rik-mcp-server
   

   You can replace the package with the other MCP servers as needed, or run multiple servers in parallel.

4. Verify the MCP endpoints are responding (e.g., via curl or an HTTP client against the provided MCP interface).

Notes:

• This project is modular; you can start with one server (e.g., RIK) and add EMTA or Open Data servers later.
• Some packages may be in WIP or POC status; check the package README for data sources and readiness.

Tips and caveats:

• Environment variables: certain data sources or backends may require API keys, service URLs, or credentials. Check each server package for env vars like API_BASE_URL, API_KEY, or XROAD_CONFIG and provide them in your environment or .env files as appropriate.
• Data readiness: RIK uses real daily snapshots and may require data download steps; EMTA is a POC and currently may use mock data and require X-Road authentication for live data.
• Data privacy: personal identifiers are typically handled with privacy measures; ensure your deployment complies with local regulations and the project’s data handling guidelines.
• Performance: when querying large datasets, consider enabling caching and streaming where supported by the MCP SDK.
• Expansion: the Estonia AI Kit supports adding more MCP servers over time; monitor the repository for updates to data sources, endpoints, and MCP protocol support.