Build a STAC Wildfire Defensible Space Boundary Auditor — implementation

## Build
Evaluate commercial facilities in the wildland-urban interface by parsing high-resolution drone STAC Items and simplified vegetation density records. Identify unmitigated fuel buffer encroachment to generate actionable mitigation work plans.

## User
Wildfire Loss Prevention Engineer

## Workflow
1. Ingest property parcel footprint and query STAC API for recent post-clearing aerial asset links.
2. Calculate radial defensible zones at thirty-foot and one-hundred-foot perimeters around structures.
3. Evaluate normalized vegetation density records inside defensible buffers to identify brush canopy encroachment.
4. Synthesize structured vegetation abatement work orders with exact geographic coordinates.

## Interface
Interactive spatial viewer showing building buffers, STAC asset previews, and highlighted fuel hazards.

## Data
Query public STAC catalogs with local GeoJSON polygon fixtures representing wildland-urban buildings. Normalized vegetation inputs are explicitly labeled as simplified synthetic app records, not complete official schema files.
Starter examples below are synthetic. They are not downloaded records.

## Source
[STAC specification](https://raw.githubusercontent.com/radiantearth/stac-spec/master/README.md)
Parse STAC Item bounding boxes, asset types, and GeoJSON geometry structures from the specification to locate relevant aerial assets.

## Tools
agno, pandas, shapely.
Agno provides agentic tool orchestration to coordinate spatial queries, evaluate simplified vegetation records, and manage work-order generation.
Hosted reasoning model supporting schema-constrained JSON extraction.

## Deliverables
Submit runnable source, pinned dependencies, setup commands, a working interface, example outputs, and a recorded end-to-end demo. Include tests and an action log.

## Acceptance
- Buffers structure geometries by specified buffer distances accurately.
- Correctly identifies asset links for high-resolution imagery in STAC Item.
- Generates vegetation removal tasks when fuel coverage exceeds 20% in Zone 1.

## Starter Examples
```json
[
  {
    "input": {
      "facility_id": "RESORT-CAL-44",
      "geometry": {
        "type": "Point",
        "coordinates": [
          -120.1,
          39.2
        ]
      },
      "stac_item_metadata": {
        "id": "AIR_NEVADA_20230601",
        "assets": {
          "ortho": {
            "href": "s3://imagery/resort_2023.tif"
          }
        }
      },
      "fuel_density_zone1_percent": 34.5
    },
    "expected_output": {
      "facility_id": "RESORT-CAL-44",
      "zone1_compliance": false,
      "vegetation_hazard_level": "CRITICAL",
      "abatement_actions": [
        "CLEAR_TREE_CANOPY_WITHIN_30FT"
      ],
      "inspection_status": "FLAGGED_FOR_AUDIT"
    }
  },
  {
    "input": {
      "facility_id": "RESORT-CAL-45",
      "geometry": {
        "type": "Point",
        "coordinates": [
          -120.15,
          39.22
        ]
      },
      "stac_item_metadata": {
        "id": "AIR_NEVADA_20230601",
        "assets": {
          "ortho": {
            "href": "s3://imagery/resort_2023.tif"
          }
        }
      },
      "fuel_density_zone1_percent": 8
    },
    "expected_output": {
      "facility_id": "RESORT-CAL-45",
      "zone1_compliance": true,
      "vegetation_hazard_level": "LOW",
      "abatement_actions": [],
      "inspection_status": "VERIFIED_COMPLIANT"
    }
  }
]
```

## Safety
Vegetation mitigation reports do not mandate legal clear-cutting without local forestry permits.
Use mock external writes. Require human review before any real deployment.