FA v2 guidelines for the DisasterWorker and processing USGS, NASA EONET, and GDACS disaster intelligence data.
The disaster-tracking skill is centered around Phase 15 processing of real-time multi-source disaster Intelligence. It primarily relies on DisasterWorker to ingest, deduplicate, and normalize warnings/alerts from natural disasters globally, but specifically focusing on implications for supply chains crossing the Indian Subcontinent and associated maritime chokepoints.
xml.etree.ElementTree with proper namespace handling.process() function captures all XML ParseErrors and translates them to WorkerError(provider="GDACS").The DisasterWorker must output a standardized DisasterEvent schema.
class DisasterEvent(BaseModel):
schema_version: int = 1
source: Literal["usgs", "eonet", "gdacs"]
event_id: str
event_type: str
severity: str
coordinates: tuple[float, float]
timestamp: datetime
impact_radius_km: Optional[float] = None
affected_countries: list[str] = Field(default_factory=list)
raw_data: dict
DisasterWorker. Build real SQLite temp tables if you implement localized caching.@breaker(timeout=60, failures=3) decorator as USGS/EONET feeds can occasionally hang.KnowledgeGraphAgent, but the worker should ensure unique event_id generation for identical events reported by different APIs using SHA-256 hashing if necessary.