AusEcon MCP for ABS | RBA | APRA data

Documentation  ·  Getting started  ·  Tool reference  ·  Changelog

Why this exists

Australian economic data is authoritative but awkward to reach — scattered across portals, formats, and identifiers you have to memorise. ausecon is the open, free, no-API-key way to put it in front of any AI assistant: every series comes back fresh and fully source-traceable, stamped with its provenance, in one consistent shape (metadata · series · observations) across all three regulators. Ask for “the cash rate” or “quarterly real GDP growth” and get research-grade data back without leaving the conversation.

Related MCP server: BCB MCP

What you get

Data sources

Try it instantly (no install)

Prefer not to install anything? A hosted, read-only, no-API-key instance speaks MCP over Streamable HTTP at:

https://mcp.auseconmcp.com/mcp

Point any MCP client that supports remote (Streamable HTTP) servers at that URL — for example, in Claude Code:

claude mcp add --transport http ausecon https://mcp.auseconmcp.com/mcp

The hosted instance may take a few seconds to wake on the first request. The previous https://ausecon-mcp-server.onrender.com/mcp URL continues to work and points at the same instance.

Install

The package lives on PyPI and is designed to be launched on demand by your MCP client via uvx:

uvx ausecon-mcp-server

The server speaks MCP over standard input/output. Launched on its own, it simply waits for a client to connect.

Connect your client

claude mcp add --transport stdio ausecon -- uvx ausecon-mcp-server

codex mcp add ausecon -- uvx ausecon-mcp-server

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "ausecon": {
      "command": "uvx",
      "args": ["ausecon-mcp-server"]
    }
  }
}

Add to ~/.cursor/mcp.json (global) or .cursor/mcp.json (project):

{
  "mcpServers": {
    "ausecon": {
      "command": "uvx",
      "args": ["ausecon-mcp-server"]
    }
  }
}

Or paste this one-click link into your browser:

cursor://anysphere.cursor-deeplink/mcp/install?name=ausecon&config=eyJjb21tYW5kIjoidXZ4IiwiYXJncyI6WyJhdXNlY29uLW1jcC1zZXJ2ZXIiXX0=

Add to ~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "ausecon": {
      "command": "uvx",
      "args": ["ausecon-mcp-server"],
      "env": {}
    }
  }
}

Or add to .vscode/mcp.json (workspace) or your user mcp.json:

{
  "servers": {
    "ausecon": {
      "type": "stdio",
      "command": "uvx",
      "args": ["ausecon-mcp-server"]
    }
  }
}

Hosting it instead? smithery.yaml and Dockerfile.smithery ship a Streamable HTTP deployment at /mcp. See the Smithery guide.

A quick taste

Find the concept you want, then ask for the series:

list_economic_concepts(query="cash rate")

get_economic_series(
    concept="cash_rate_target",
    start="2020-01-01",
)

Need a transparent, formula-based indicator? Call the derived surface directly:

get_derived_series(concept="real_cash_rate", last_n=12)

Connected to an AI agent, you can skip the syntax entirely — ask for “quarterly real GDP growth” and it maps your request to the right tool calls for you.

Develop locally

Python 3.12 is recommended; the CI matrix supports 3.10+.

uv sync --python 3.12
uv run pytest
uv run ruff check src tests scripts

The repo also ships a manual benchmark (evals/) that measures the server's impact on model answers across 52 Australian-economics questions, comparing a bare model, web search, and the ausecon tools. Ground-truth resolution is free to check:

uv run --group evals python -m evals.run_eval --dry-run

A full run makes paid API calls — see the evaluation harness guide before running one.

auseconmcp.com  ·  Issues  ·  MIT Licence  ·  Made for the Australian data community