aemo-mcp

mcp-name: io.ausdata/aemo-mcp

Ask Claude about Australia's electricity market and get real, current numbers — 5-minute spot prices, regional demand, generation by fuel, interconnector flows, rooftop solar — not "I don't have access to that data." This MCP server gives Claude (and other MCP clients like Cursor) live access to the Australian Energy Market Operator (AEMO) NEMWEB feeds, with curated mappings for the most-asked indicators.

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).

Companion to abs-mcp (ABS macro stats), rba-mcp (RBA interest + FX rates), ato-mcp (ATO tax + ACNC charity), apra-mcp (banking + superannuation), aihw-mcp (health & welfare), asic-mcp (companies + financial advisers), and au-weather-mcp (Australian weather) — together they cover the most-asked Australian official data.

What you can ask

Once installed, your LLM can answer questions like:

Every answer comes with the interval timestamp (AEMO market time, UTC+10), units (MW, $/MWh), and a link back to the NEMWEB source. The MCP wraps NEMWEB's CSV/ZIP feeds and exposes them through 5 plain-English tools.

Install

# After publish:
uvx --upgrade aemo-mcp

# Local dev:
uv pip install -e .

Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json:

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

Why --upgrade? uvx aemo-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. Recommended for everyone except offline-first / pinned-version workflows. To verify which version is currently serving you, look at the server_version field on any DataResponse.

If you also have rba-mcp / abs-mcp installed, all servers run side-by-side. Claude disambiguates with the server prefix (aemo:get_data, rba:get_data, abs:get_data).

For local dev (pre-PyPI):

{
  "mcpServers": {
    "aemo": {
      "command": "uv",
      "args": ["run", "--directory", "/absolute/path/to/aemo-mcp", "aemo-mcp"]
    }
  }
}

Cursor

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

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

Tools

Curated datasets

The 7 datasets cover ~95% of typical NEM analytic queries:

Use list_curated() to enumerate, describe_dataset(dataset_id) to learn the filters available on each.

Regions

NSW1 (New South Wales), QLD1 (Queensland), SA1 (South Australia), TAS1 (Tasmania), VIC1 (Victoria). Western Australia (WEM) and the Northern Territory are not on the NEM and are out of scope.

Trust contract

Every DataResponse carries:

• source = "Australian Energy Market Operator"
• attribution — AEMO Copyright Permissions verbatim attribution string
• source_url — the NEMWEB folder the data came from
• retrieved_at — UTC timestamp of the fetch
• interval_start / interval_end — period covered
• stale — True if the latest interval is older than 2× the feed cadence
• server_version — the wheel that served the call

Licence + attribution

This package is MIT-licensed (see LICENSE).

The AEMO data it fetches is published under AEMO's Copyright Permissions policy: AEMO grants general permission to use AEMO Material for any purpose (commercial included) on the sole condition of accurate attribution of the relevant material and AEMO as its author. See https://aemo.com.au/privacy-and-legal-notices/copyright-permissions.

End-users redistributing data fetched via this server must credit AEMO. The canonical attribution string is on every DataResponse.attribution.

Worked examples

"What's the current NSW spot price?"

latest(dataset_id="dispatch_price", filters={"region": "NSW1"})

→ {"records": [{"period": "2026-05-14T10:05:00+10:00", "value": 87.5, "dimensions": {"region": "NSW1", "metric": "rrp"}, "unit": "$/MWh"}], ...}

"NSW spot price for the last 24 hours"

get_data(dataset_id="dispatch_price", filters={"region": "NSW1"},
         start_period="2026-05-13", end_period="2026-05-14")

"Did SA hit negative pricing in the last 24 hours?"

get_data(dataset_id="dispatch_price", filters={"region": "SA1"},
         start_period="<24h ago>")

Then the LLM filters value < 0 client-side.

"Generation by fuel type right now in QLD"

latest(dataset_id="generation_scada", filters={"region": "QLD1"})

→ DUID-level rows with fuel attribution; the LLM aggregates by fuel.

How it works

• Live-fetch only. No NEMWEB archives in the wheel. Every request goes through the cache.
• Cache TTLs tuned per cadence. 60s for 5-min feeds, 5min for 30-min feeds, 1h forecasts, 24h daily archive. Timestamped historical files are immutable in NEMWEB and cache effectively forever.
• In-flight request deduplication. Concurrent callers for the same URL share one HTTP request. Critical at 5-min cadence with many users.
• Latest-file detection is purely lexicographic on the NEMWEB directory listing — AEMO embeds the interval timestamp in every filename (PUBLIC_DISPATCHIS_YYYYMMDDHHmm_<seq>.zip), so max() is enough.
• Multi-section CSV parser handles AEMO's I,/D, row format where a single ZIP holds several tables (DISPATCHPRICE, DISPATCHREGIONSUM, DISPATCHINTERCONNECTORRES, etc.).

Development

git clone https://github.com/Bigred97/aemo-mcp
cd aemo-mcp
uv sync --extra dev
uv pip install -e .
uv run pytest -q

Run live tests (hit NEMWEB):

uv run pytest -q -m live

Zero-flake validation:

for i in $(seq 1 10); do uv run pytest -q || break; done

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).