Getting Started

ausecon is an open, free, no-API-key MCP server for official Australian economic data — ABS, RBA, and APRA — with source-traceable provenance and one consistent response shape (metadata · series · observations).

This page takes you from zero to a live answer in about two minutes.

Try it instantly (no install)

A hosted, read-only, no-API-key instance is available over Streamable HTTP at https://mcp.auseconmcp.com/mcp. Point any MCP client that supports remote servers at that URL, or 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 locally

Requirements: Python 3.10+ and uv for the uvx launcher.

The package is published to PyPI and is intended to be launched on demand:

uvx ausecon-mcp-server

On first use, uvx downloads the package into an isolated cached environment. The process then waits for an MCP client to connect over standard input/output.

Connect your client

ausecon speaks MCP over stdio, launched on demand with uvx.

Claude Code

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

Codex

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

Claude Desktop — add to claude_desktop_config.json:

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

Cursor — add to ~/.cursor/mcp.json, or use the one-click link Add to Cursor:

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

Windsurf — add to ~/.codeium/windsurf/mcp_config.json:

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

VS Code — add to .vscode/mcp.json, or use the one-click link Install in VS Code:

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

Ask your first question

Ask your agent: “What’s Australia’s real cash rate right now?”

The agent calls one tool:

get_derived_series(concept="real_cash_rate", last_n=3)

and gets back the RBA cash-rate target less year-ended monthly CPI inflation, computed transparently from the two official upstream series (abbreviated):

{
  "metadata": {
    "source": "derived",
    "title": "Real cash rate",
    "retrieved_at": "2026-07-26T13:12:24Z",
    "server_version": "1.14.0",
    "derived": {
      "formula": "cash_rate_target - monthly_inflation",
      "description": "RBA cash-rate target less complete monthly CPI year-ended inflation. This is an ex-post real rate (the nominal rate less realised year-ended CPI inflation), not the ex-ante Fisher rate (less expected inflation).",
      "operands": [
        { "name": "cash_rate", "source": "rba", "dataset_id": "a2", "series_ids": ["ARBAMPCNCRT"] },
        { "name": "inflation", "source": "abs", "dataset_id": "CPI", "abs_key": "3.10001.10.50.M" }
      ],
      "units": "percentage points"
    }
  },
  "series": [
    { "series_id": "real_cash_rate", "label": "Real cash rate", "unit": "percentage points" }
  ],
  "observations": [
    { "date": "2026-03", "series_id": "real_cash_rate", "value": -0.5 },
    { "date": "2026-04", "series_id": "real_cash_rate", "value": -0.1 },
    { "date": "2026-05", "series_id": "real_cash_rate", "value": 0.35 }
  ]
}

This one response shows the whole design: a semantic concept resolved to official sources, a transparent formula with both operands named down to the exact RBA/ABS series identifiers, provenance stamps (retrieved_at, server_version), and an explicit caveat that this is an ex-post real rate. Every retrieval tool returns this same metadata · series · observations shape.

Where to next

Use list_economic_concepts for ordinary economic requests such as GDP, CPI, unemployment, wages, cash rate, credit, exchange rates, or yields, then get_economic_series with the selected concept.
Use get_derived_series for transparent formula-based indicators such as real cash rate, yield-curve slope, real wage growth, mortgage-rate spreads, or credit-to-GDP.
Use search_datasets, list_catalogue, get_abs_dataset_structure, get_abs_data, get_rba_table, and get_apra_data when you need source-native ABS/RBA/APRA control.

Retrieval responses include metadata, series, and observations. Semantic retrievals also include metadata.semantic, recording the resolved source target and normalised date bounds; derived retrievals include metadata.derived, recording the formula, operands, units, and alignment method.

For more worked calls see Examples, and if you are using the server through an AI agent, see Prompting AI Agents for natural-language requests and the MCP tool calls they usually trigger. For the trust story — freshness stamps, caching, and staleness flags — see Data Freshness & Provenance.