Skip to main content
Glama
fablackner

earth-data-mcp

by fablackner
OverviewSchemaSourceRelated ServersDiscussionsAdmin
JavaScript
Remote

earth-data-mcp

An MCP server that gives AI agents live Earth data: earthquakes, volcanoes, weather and warnings, air quality, sea state, natural events, disaster alerts, space weather, and atmospheric CO2.

All upstream sources are public and unauthenticated, so the server needs no API keys or configuration.

Tools

Tool

Purpose

search_earthquakes

USGS earthquakes by magnitude, time window, and location radius.

get_earthquake

Full detail for one USGS event.

search_volcanic_activity

Current weekly volcanic activity report, filterable by volcano or country.

get_weather

Current conditions and daily forecast for a place or coordinates.

get_lightning_risk

Hourly thunderstorm/lightning risk.

search_weather_alerts

Official warnings (US NWS, or MeteoAlarm for Europe).

get_air_quality

AQI, pollutants, UV, and pollen (Europe).

get_marine_conditions

Waves, swell, sea temperature, and currents offshore.

search_natural_events

NASA EONET events: wildfires, storms, icebergs, dust, …

search_disasters

GDACS disaster alerts with humanitarian impact levels.

get_space_weather

NOAA storm scales, Kp, flares, and aurora outlook.

get_co2_record

Mauna Loa atmospheric CO2: latest month, yearly change, last 24 months.

Related MCP server: quake-mcp

Usage

bun install
bun run start      # stdio
bun run start:http # HTTP on $PORT (default 3000) at /mcp

Add it to Claude Code:

claude mcp add earth-data -- bun /absolute/path/to/earth-data-mcp/src/index.js

Deploying

The HTTP mode is stateless, so it runs on any serverless host. The repo is set up for Vercel:

bunx vercel deploy --prod    # → https://<project>.vercel.app/mcp

Other options: Cloudflare Workers or Deno Deploy (with their own entry point), or Render / Fly.io running bun run start:http. Static hosts like GitHub Pages won't work.

The deployed endpoint is public. That is fine here since it only proxies read-only public data.

Eval

eval/ checks whether an agent picks the right tool, passes sensible arguments, and reports answers that match the live upstream data.

ANTHROPIC_API_KEY=... bun run eval
MODEL=claude-opus-4-8 EFFORT=high RUNS=3 bun run eval
MOCK=1 bun run eval    # scripted agent, no API calls

Scoring is deterministic (no LLM judge), and ground truth is fetched at run time. The command exits non-zero on failure.

Data sources

License

MIT

This server cannot be deployed

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Maintenance

ActivityMaintained
ResponsivenessNo issues

Resources

Related MCP Connectors

Related MCP Servers

  • wems-mcp-server

    heliosarchitect
    A
    license
    Not graded
    quality
    D
    maintenance
    A powerful MCP server that connects AI assistants to authoritative natural hazard data sources, enabling monitoring of earthquakes, tsunamis, volcanoes, and solar events with configurable alerts and webhooks.
    1
    MIT

  • quake-mcp

    adamcwade
    A
    license
    A
    quality
    C
    maintenance
    MCP server wrapping the USGS Earthquake Hazards API, enabling AI assistants to search the global earthquake catalog, look up event details, count quakes, find 'Did You Feel It' reports, and read realtime feeds.
    5
    MIT

  • geo-agent

    KavyaAgarwal2001
    F
    license
    A
    quality
    D
    maintenance
    Exposes live earth-science data (e.g., earthquakes from USGS) as tools for MCP-compatible AI agents, enabling them to fetch and reason about real-time geophysical events.
    1
    -