ℹ️ Updated: Weather alert endpoint (cuacasignifikan.json) removed by BMKG. Now using gempadirasakan.json for alerts. Earthquake data (autogempa.json) works fine.

Module: bmkg

Source portal: data.bmkg.go.id + inatews.bmkg.go.id Scrape method: REST API (clean JSON/XML, no scraping) Phase: 3 License: MIT Status: ACTIVE

Source

Rate Limits & Block Behaviour

Normalized Schema (result object)

Weather forecast

{
    "city": str,                       # Requested city name
    "province": str,                   # BMKG province code
    "area": str,                       # Area description from BMKG
    "forecast": list[dict],            # Forecast entries (capped at 50)
}

Each forecast entry:

{
    "parameter": str,                  # e.g. "t" (temperature), "hu" (humidity), "weather"
    "description": str,                # Human-readable parameter name
    "day": str,                        # Forecast day (e.g. "1", "2", "3")
    "hour": str,                       # Forecast hour
    "value": str,                      # Forecast value
    "unit": str,                       # Unit (e.g. "C", "%")
}

Earthquake

{
    "date": str,                       # Date (DD-MMM-YYYY or similar)
    "time": str,                       # Time (HH:MM:SS WIB)
    "datetime_utc": str | None,        # ISO 8601 UTC timestamp if available
    "coordinates": str | None,         # Lat,Lon string
    "latitude": str,                   # Latitude
    "longitude": str,                  # Longitude
    "magnitude": float,                # Richter magnitude
    "depth_km": float,                 # Depth in kilometers
    "region": str,                     # Affected region description
    "tsunami_potential": str | None,   # Raw tsunami potential text
    "tsunami_warning": bool,           # True if tsunami warning issued
    "felt_reports": str | None,        # Dirasakan (felt reports)
    "shakemap_url": str | None,        # URL to shakemap image
}

Alert

{
    "alert_id": str,                   # Internal alert ID
    "alert_type": str,                 # e.g. "Cuaca Ekstrem", "Gelombang Tinggi"
    "region": str,                     # Affected region
    "description": str,                # Alert description
    "date": str,                       # Alert date
    "time": str,                       # Alert time
    "severity_level": str | None,      # Severity level if available
}

MCP Tools

claude mcp add

claude mcp add civic-stack -- civic-stack-mcp  # all 40 tools

FastAPI Endpoints

Supported Cities (Weather Forecast)

Jakarta, Surabaya, Bandung, Medan, Makassar, Yogyakarta, Bali/Denpasar, Semarang, Palembang, Pekanbaru, Balikpapan

For other cities, use the province name.

Example Response (Earthquake)

{
  "result": {
    "date": "13-Mar-2026",
    "time": "14:25:33 WIB",
    "coordinates": "-2.10,100.45",
    "latitude": "-2.10",
    "longitude": "100.45",
    "magnitude": 5.7,
    "depth_km": 10.0,
    "region": "47 km BaratLaut BUKITTINGGI-SUMBAR",
    "tsunami_potential": "Tidak berpotensi tsunami",
    "tsunami_warning": false,
    "felt_reports": "Dirasakan (Skala MMI): II-III Bukittinggi",
    "shakemap_url": "https://data.bmkg.go.id/DataMKG/TEWS/20260313142533.mmi.jpg"
  },
  "found": true,
  "status": "ACTIVE",
  "confidence": 1.0,
  "source_url": "https://data.bmkg.go.id/DataMKG/TEWS/autogempa.json",
  "fetched_at": "2026-03-13T07:30:00Z",
  "last_updated": null,
  "module": "bmkg",
  "raw": null
}

Known Issues & Quirks

• Weather forecast endpoint returns XML — parsed into structured JSON
• Earthquake coordinates are sometimes returned as “Lat,Lon” string; normalizer extracts both
• Magnitude is parsed to float; depth has “km” stripped and parsed to float
• Tsunami warning is inferred from tsunami_potential text (“tidak berpotensi” → false)
• Forecast entries are capped at 50 to avoid huge responses
• City name matching is case-insensitive and supports partial matches
• Province codes (for forecast URLs) are hardcoded — fallback uses the city name as-is

VCR Fixtures

Legal Basis

This module queries BMKG’s official open data APIs, which publish meteorological, climatological, and seismological data for public safety and awareness. No authentication is required. Data is fetched on demand and not persisted. All data is public record published for disaster preparedness.