address-disaster-risk-assessor

Produce a source-attributed disaster risk packet for a Metro Manila address using validated public hazard and geospatial sources.

SKILL.md

GitHub

Address Disaster Risk Assessor

Use this skill when the user wants to evaluate disaster exposure for a specific address or set of coordinates.

Scope

First geography: Metro Manila / NCR.

First risk classes:

  • flood hazard and flood-prone history;
  • river and waterway proximity;
  • liquefaction exposure;
  • active fault proximity;
  • landslide relevance;
  • storm-surge relevance where applicable.

Weak or later risk classes:

  • waste-management quality;
  • drainage maintenance quality;
  • garbage-collection reliability;
  • local complaint history.

Only include weak-service signals when a clean source or defensible proxy is validated. Do not infer them from vibes or one-off anecdotes.

Inputs

Preferred:

  • address string;
  • optional latitude and longitude;
  • optional property/use type.

If coordinates are not available, geocode first and record confidence. If geocoding confidence is low, stop and ask for a more precise address or coordinates before scoring.

Workflow

  1. Normalize the input location.
  2. Identify city and barangay when possible.
  3. Load or query priority hazard sources from data/disaster-risk/source-priorities.json.
  4. Check point-in-polygon exposure for hazard zones.
  5. Check nearest-neighbor distance to flood-prone points, rivers, waterways, and active-fault features where available.
  6. Assign per-component confidence based on source quality, freshness, spatial precision, and address confidence.
  7. Generate a risk packet with sources, limitations, and non-certification language.

Output Format

Return a concise report with these sections:

  1. Location Normalization
  2. Overall Risk Summary
  3. Flood And Water Risk
  4. Earthquake, Fault, And Liquefaction Risk
  5. Landslide / Storm Surge Relevance
  6. Waste Or Drainage Service Signals, only if evidence exists
  7. Data Gaps And Confidence
  8. Sources
  9. Disclaimer

Also emit a structured JSON object when producing files:

{
  "input_address": "",
  "normalized_address": "",
  "coordinates": { "lat": null, "lon": null },
  "geocode_confidence": "",
  "admin_context": {
    "city": "",
    "barangay": "",
    "psgc": ""
  },
  "overall_risk_label": "",
  "risk_components": [],
  "data_gaps": [],
  "limitations": [],
  "sources": []
}

Risk Labels

Use:

  • high
  • medium
  • low
  • unknown
  • not_applicable

Do not collapse unknowns into low risk.

Required Caveat

This output is a public-data screening report, not an engineering assessment, insurance determination, legal opinion, or official government hazard certification.

Priority Sources

Primary automation stack:

  • MGBPublic Flood MapServer for flood susceptibility.
  • MMDA Flood-Prone Areas FeatureServer for local flood-prone point history.
  • MMDA River Line FeatureServer for waterway proximity.
  • PHIVOLCS Liquefaction MapServer for liquefaction exposure.
  • PHIVOLCS ActiveFault MapServer for nearest active fault distance.
  • Project NOAH hazard maps through BetterGov for flood, landslide, debris-flow, and storm-surge zones after PMTiles schema inspection.
  • HDX Philippines administrative boundaries for NCR/city/barangay joins.
  • OSM / Geofabrik waterways, roads, and facility proxies.

Secondary or manual sources:

  • MGBPublic RainInducedLandslide MapServer when landslide relevance exists.
  • PAGASA Pasig-Marikina-Tullahan FFWS for live flood-warning context only.
  • HazardHunterPH / GeoRiskPH for official manual cross-check.
  • PHIVOLCS FaultFinder for official manual fault-distance cross-check.
  • PHIVOLCS GroundShaking and Tsunami MapServers after more point validation.
  • LGU DRRMO/CDRA and waste-service pages only as manual context or weak proxies.

Verification

Before committing changes to this workflow:

npm run validate
git diff --check