Cookbook

Current RBA cash rate

curl -H 'Authorization: Bearer ak_your_key' \
  https://api.ausdata.io/v1/series/AU.CASHRATE/latest

{
  "data": { "series_id": "AU.CASHRATE", "period": "...", "value": ..., "units": "Percent" },
  "meta": {
    "endpoint": "/v1/series/AU.CASHRATE/latest",
    "period": { "start": "...", "end": "..." },
    "row_count": 1,
    "retrieved_at": "...",
    "sources": [ { "name": "Reserve Bank of Australia", "url": "...", "attribution": "..." } ],
    "stale": false
  },
  "links": { "csv": null }
}

Inflation-adjust a wage series (real wages)

curl -H 'Authorization: Bearer ak_your_key' \
  'https://api.ausdata.io/v1/real-wages?start_period=2015-Q1&end_period=2024-Q4'

CPI history as a CSV for Excel or pandas

Append ?format=csv to any series request and you get a CSV body instead of JSON. The same URL drops straight into pandas.

curl -H 'Authorization: Bearer ak_your_key' \
  'https://api.ausdata.io/v1/series/AU.CPI.YOY?format=csv&limit=1000' -o cpi.csv

Load a series straight into pandas

No download step. pd.read_csv reads the ?format=csv URL directly when you pass the key in the header.

import pandas as pd

url = "https://api.ausdata.io/v1/series/AU.CPI.YOY?format=csv&limit=1000"
headers = {"Authorization": "Bearer ak_your_key"}

df = pd.read_csv(url, storage_options=headers)
print(df.tail())

Prefer a typed client? The Python SDK returns rows you can hand to a DataFrame:

# pip install ausdata-sdk
from ausdata import Client
import pandas as pd

client = Client(api_key="ak_your_key")
rows = client.series("AU.CPI.YOY", limit=1000)
df = pd.DataFrame(rows)
print(df.tail())

Compare two series on one chart (overlay)

Mint a read-only embed token at /v1/account/embed-token, then drop the chart anywhere. No live key in your HTML.

<img src="https://api.ausdata.io/v1/series/multi/chart.svg?ids=AU.CPI.YOY,AU.WAGES.YOY&key=emb_your_embed_token">

Search across all nine sources

curl -H 'Authorization: Bearer ak_your_key' \
  'https://api.ausdata.io/v1/search-datasets?q=unemployment'

{
  "data": [ { "source": "...", "dataset_id": "...", "name": "...", "score": ... } ],
  "meta": {
    "endpoint": "/v1/search-datasets",
    "query": { "q": "unemployment" },
    "row_count": ...,
    "retrieved_at": "...",
    "sources": [ ... ],
    "stale": false
  },
  "links": { "csv": null }
}

Connect Claude (or any agent) via MCP

The MCP server wraps the same API. For Claude Desktop, add this block to your claude_desktop_config.json, drop in your key, and restart Claude.

{
  "mcpServers": {
    "ausdata": {
      "command": "npx",
      "args": ["-y", "ausdata-mcp"],
      "env": {
        "AUSDATA_API_KEY": "ak_your_key"
      }
    }
  }
}

No global install needed, npx fetches the server on first run. Once connected, Claude can call every series, transform and search endpoint directly.