kardashev-mcp
MCP server for Kardashev Labs US grid data. Gives agents live LMP, carbon intensity, load, curtailment, interconnection queues, ERCOT large loads, and the public RT−DA spread forecast track record. No API key required.
Data comes from the public API at data.kardashevlabs.org.
Install
npx -y kardashev-mcp
Cursor (~/.cursor/mcp.json)
{
"mcpServers": {
"kardashev": {
"command": "npx",
"args": ["-y", "kardashev-mcp"]
}
}
}
Claude Code
claude mcp add kardashev -- npx -y kardashev-mcp
Tools
| Tool | What it returns |
|---|---|
list_isos | Covered ISOs/RTOs and live ingest status |
get_fuel_mix | Generation by fuel type (MW) |
get_carbon | Carbon intensity (lbs CO₂/MWh) |
get_lmp | Locational marginal prices (RT/DA) |
get_load | Actual grid load (MW) |
get_load_forecast | Near-term load forecast |
get_curtailment | Solar/wind curtailment |
get_queue | Interconnection queue projects |
get_large_load | ERCOT large-load interconnection snapshot |
get_spread_forecast | Live ERCOT RT−DA spread forecast (issued before delivery) |
get_forecast_track_record | Scored public track record for the spread models |
Row-returning tools default to limit=50 (hard max 500) so agents stay within context budgets.
Optional env
| Variable | Default | Purpose |
|---|---|---|
KARDASHEV_BASE_URL | https://data.kardashevlabs.org | Override API base (local data plane) |
KARDASHEV_TIMEOUT_MS | 30000 | Per-request timeout |
Develop
npm install
npm run build
npm start # stdio
npm run inspect # MCP Inspector
Links
- Website: kardashevlabs.org
- Forecast track record: kardashevlabs.org/forecast
- API docs: data.kardashevlabs.org/docs
- Python client:
kardashev - JS/TS client:
kardashev
License
MIT