BudgetKey

You can query and analyze Israeli budget data through a dedicated MCP server that exposes BudgetKey data via an MCP interface. This server lets AI assistants fetch budget items, contracts, entities, revenues, and more, enabling rich budget analysis and reporting.

How to use

You connect to the BudgetKey MCP server using an MCP client or integration that understands the MCP protocol. Start by discovering dataset structures, then search for relevant records, and finally query for precise results with inline filters. You can navigate datasets such as budget items, contracts, entities, and revenues, and include item_url links in your results for direct data access.

Practical usage pattern: first inspect a dataset’s schema, then perform a text search to locate identifiers, and finally run an SQL-like query to retrieve the exact fields you need. Always include item_url in your results to provide direct data access.

How to install

Prerequisites: you need Python 3.x and pip installed on your machine.

Option A — Run the MCP server locally (Python)

# Start the MCP server locally
python server.py

Option B — Use the remote HTTP MCP endpoint

Configure your MCP client to connect to the remote endpoint at the following URL.

{
  "mcpServers": {
    "budgetkey": {
      "type": "http",
      "url": "https://next.obudget.org/mcp",
      "args": []
    }
  }
}

Additional notes

Project includes a Docker image for quick local deployment. Run the container to expose the MCP endpoint on port 8000.

Example: start the container and access the MCP endpoint at http://localhost:8000/mcp.

Tools and data access patterns

The server provides three primary tools to interact with datasets: DatasetInfo to learn schema and columns, DatasetFullTextSearch to locate textual identifiers, and DatasetDBQuery to execute precise SQL-like queries. Always use DatasetInfo before querying a dataset.

Available tools

DatasetInfo

Get comprehensive information about a dataset, including its columns and database schema. Use this before querying a dataset.

DatasetFullTextSearch

Perform free-text search on a dataset to locate relevant items before querying.

DatasetDBQuery

Execute SQL queries on a dataset's database to retrieve precise information; always include item_url in results.