aihw-mcp

mcp-name: io.ausdata/aihw-mcp

MCP server for Australian Institute of Health and Welfare statistics. Plain-English access to long-term mortality (GRIM), regional mortality (MORT), cancer incidence and mortality (ACIM), national health expenditure, youth justice detention, and the public hospitals register — all from a single uvx command.

Hosted access? For cross-source queries, webhooks, an always-on REST API, and a uniform response envelope across all 9 sources, see ausdata.io — free tier available (500 calls/mo, no card).

"How have diabetes deaths changed since 1980?"
"What's the age-standardised mortality rate in the Sydney - Inner West SA3?"
"Show me breast cancer incidence in women aged 50–54 over time."
"How much did Australia spend on public hospitals in 2022-23?"
"How many young people are in detention in NSW vs VIC?"
"List all Principal referral hospitals in Queensland."
"Top 5 causes of death in 2023."

Sister to abs-mcp (Australian Bureau of Statistics), rba-mcp (Reserve Bank of Australia), ato-mcp (Australian Taxation Office), and au-weather-mcp (Australian weather). The five together cover the macro / regulator / tax / health / climate layer of Australian official data.

Install

# Run on demand via uvx (recommended)
uvx --upgrade aihw-mcp

# Or install permanently
pip install aihw-mcp

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "aihw": { "command": "uvx", "args": ["--upgrade", "aihw-mcp"] }
  }
}

Why --upgrade? uvx aihw-mcp (without the flag) uses whatever wheel is cached and never adopts new PyPI releases on its own. --upgrade makes uvx check PyPI on each launch and pull a newer release if one exists. To verify which version is currently serving you, look at the server_version field on any DataResponse.

Claude Code / Cursor

claude mcp add aihw --command uvx --args -- --upgrade aihw-mcp

Auto-updating data

Beyond the wheel-level --upgrade, the server has a second auto-update path inside the data layer: when AIHW refreshes GRIM with another year of deaths data or publishes a new MORT release, aihw-mcp resolves the new resource URL via data.gov.au's CKAN API at fetch time and uses the freshest match. Hard-coded YAML URLs are the safe fallback if discovery fails. You do not need to wait for a new wheel release to get new yearly data — just delete ~/.aihw-mcp/cache.db to force a refresh, or wait for the 7-day TTL to expire.

What it exposes

Six tools, all plain-English in, structured out:

Every response is the same shape — dataset_id, dataset_name, query, period, unit, row_count, records, aihw_url, attribution, server_version — across every curated dataset.

Time-series datasets (GRIM, MORT, ACIM, Health Expenditure, Youth Justice) accept start_period and end_period on get_data — e.g. start_period="2000", end_period="2010" narrows GRIM to that decade. latest() returns the most-recent observation per measure, sorted by the dataset's declared period dimension (not by source row order). Error messages include fuzzy "did you mean?" suggestions when you typo a filter or measure name.

Curated datasets (6 in v0.1)

Adding a new dataset is a single YAML drop into src/aihw_mcp/data/curated/ — see CONTRIBUTING.md.

Example queries (paste into Claude)

Cross-source compatibility. The state filter on HEALTH_EXPENDITURE, PUBLIC_HOSPITALS, YOUTH_JUSTICE_DETENTION, and other location-aware datasets accepts canonical state codes ("NSW"), full names ("New South Wales"), case-insensitive variants ("nsw"), ISO 3166-2 ("AU-NSW"), and 4-digit postcodes ("2000" → NSW). Powered by aus-identity — the same input format works across abs-mcp, ato-mcp, apra-mcp, aihw-mcp, and asic-mcp.

Public-health research: "For GRIM_DEATHS, give me the deaths and age-standardised rate for 'Diabetes' for Persons every year from 1980 to the latest, so I can chart the trajectory."

Health-tech / regional analysis: "Using MORT_GEOGRAPHY, list the 10 SA3 regions with the highest age-standardised mortality rate for Persons in the most recent year."

Oncology: "From CANCER_INCIDENCE_MORTALITY, give me Breast cancer incidence in Females across the 50–54 age band for every available year, plus the same age band's mortality."

Health-policy: "From HEALTH_EXPENDITURE, what was the real spend on 'Public hospitals' in NSW in 2022-23, broken down by broad source (Government vs Non-government)?"

Criminal-justice tech: "Using YOUTH_JUSTICE_DETENTION, compare the average nightly youth-detention population in NSW vs VIC in 'Jun qtr 2017', for both Indigenous and Total."

Hospital-tech / market intel: "From PUBLIC_HOSPITALS, list every 'Principal referral' hospital with their state and Local Hospital Network. Then count how many there are per state."

Each prompt resolves to one or two get_data / top_n calls. The response includes the source URL so the agent can cite it back.

Architecture

Same shape as the sister packages — client → cache → parsing → shaping → server:

• client.py wraps httpx with a SQLite-backed disk cache (per-resource TTL).
• parsing.py reads CSV (via pandas) and XLSX (via openpyxl/pandas). Header rows + sheet names live in the curated YAML so future format quirks are a YAML edit, not a code change.
• curated.py loads dataset specs from data/curated/*.yaml — each one declares its dimensions, measures, dimension value enums, source/download URLs, format, and parse layout.
• shaping.py transforms the parsed DataFrame into DataResponse (records / series / csv).
• server.py is the FastMCP entrypoint — six tools, full input validation with helpful "Try X" hints on error.

Cache lives under ~/.aihw-mcp/cache.db. Most AIHW datasets refresh once a year; the TTLs are tuned for that cadence.

Attribution

Data sourced from the Australian Institute of Health and Welfare (AIHW) via data.gov.au. Licensed under Creative Commons Attribution 3.0 Australia (CC BY 3.0 AU). The MCP server is MIT-licensed; the data carries the upstream CC-BY 3.0 AU licence, which is echoed in every response's attribution field.

Sister MCPs (Australian Public Data portfolio)

Want all 9 sources behind one REST API? The hosted gateway at ausdata.io adds cross-source joins, full history, webhooks, and HMAC-signed responses on top of these MCPs — free tier (500 calls/mo, no card).