Valkenburg
demonstration AOI.
A 4.6 km² block over the historic town of Valkenburg aan de Geul in South Limburg, Netherlands. Hilly terrain, river valley, dense urban centre, surrounding agriculture and forest. Six layers, all sourced from open Dutch government data and OpenStreetMap, exposed as a STAC catalog with per-asset pricing.
# skill file — read this first curl https://api.megalithlabs.ai/llms.txt
# STAC 1.0.0 — programmatic access curl https://api.megalithlabs.ai/discover
| Layer | Type | Format | Sold as | Assets / features | From |
|---|---|---|---|---|---|
| ortho | Orthomosaic — RGB aerial photo | GeoTIFF, 0.25 m GSD | Tile (8×8 grid) | 64 tiles | $0.01 / tile |
| dsm | Digital Surface Model — first-return LIDAR | GeoTIFF, float32 m, 0.5 m GSD | Tile (8×8 grid) | 64 tiles | $0.03 / tile |
| dtm | Digital Terrain Model — bare-earth LIDAR | GeoTIFF, float32 m, 0.5 m GSD | Tile (8×8 grid) | 64 tiles | $0.03 / tile |
| buildings | Building footprints with derived height | GeoJSON polygons | Whole AOI | 3,891 features | $2.00 / AOI |
| power_lines | Overhead lines and pylons | GeoJSON LineStrings / Points | Whole AOI | 0 features (no overhead transmission in AOI) |
Free |
| vegetation | Vegetation patches with canopy height | GeoJSON polygons | Whole AOI | 117 features | $1.00 / AOI |
Ask your agent anything, or try one of the example queries below. Each query is designed to exercise a different subset of the catalog. Copy the query text and paste it into a fresh agent session.
A property developer's pre-acquisition agent triages parcels before a surveyor visits. It needs to confirm what's on the site, gauge slope and drainage, and check immediate built context — particularly in dense historic towns where a tiny parcel can have decisive constraints from neighbours.
We're a property developer triaging parcels before sending a surveyor. Assess an 80m × 80m candidate near Valkenburg's historic centre (~50.866°N, 5.835°E). Confirm current land use from imagery, gauge slope for drainage and foundation costs, and flag any constraints from neighbouring buildings. Recommend proceed to site visit or pass. ←COPY
The agent buys the single tile of ortho, DSM and DTM covering the parcel ($0.07), plus the buildings vector for the immediate built context ($2.00). Total ≈ $2.07. If the parcel clips into a neighbouring tile, the agent buys those additional tiles and proceeds.
A loss adjuster's pre-renewal agent flags policies along river corridors for elevated flood risk. After the catastrophic 2021 Geul floods, insurers and water boards routinely re-screen the building stock along the valley each season.
We're an insurer re-screening policies along the Geul valley after the 2021 floods. Identify all buildings within 100m of the river and rank by floor elevation (DTM at footprint) relative to a Q100 flood level. Flag the ten highest-risk addresses. ←COPY
The agent buys the buildings vector ($2.00) and the DTM tiles intersecting the river corridor (roughly 8 tiles, $0.24). Total ≈ $2.24. DTM rather than DSM, because the agent reads ground elevation at each footprint — DSM would read the roof.
A solar installer's lead-generation agent scores roofs across a target town and triages which households to approach. It needs roof orientation, slope, area, and shading from neighbours and vegetation.
We're a solar installer generating leads in Valkenburg. For each building, compute a roof-suitability score combining usable area, mean pitch, southerly exposure, and shading from adjacent structures and trees. Return the top 20 candidates with estimated kWp capacity. ←COPY
The agent buys the buildings vector ($2.00), all 64 DSM tiles ($1.92), and the vegetation vector ($1.00). Total ≈ $4.92. The DSM resolves roof shape per building at 0.5 m resolution. The agent runs a virtual sun-arc over each footprint to compute annual radiation and outputs a ranked list.
The orthomosaic is from PDOK Beeldmateriaal Open, the open Dutch national aerial-imagery programme operated by Het Kadaster — captured 2024, 0.25 m native resolution. CC-BY.
The DSM and DTM are from AHN (Actueel Hoogtebestand Nederland), the Dutch national LIDAR programme operated by Rijkswaterstaat and the water boards — 0.5 m raster derived from > 10 returns / m² point density. CC-BY.
The vector layers — buildings, power lines, vegetation — are extracted from OpenStreetMap under the Open Database Licence (ODbL). Building heights are computed by sampling DSM−DTM at each footprint centroid.