Every orbit,
woven into one fabric.
TerraFabric is the meta-repository for Earth observation — open and commercial, from raw photons to finished products. Searchable by people, operable end-to-end by AI agents, and wired into OVERMATCH as its EO backbone.
From photons to products — one query away.
Every collection is placed in the chain by processing level. Go upstream for phase histories, SLCs and photon clouds; downstream for analysis-ready reflectance, mosaics, detections and reanalyses.
Raw sensor data
4L0 – L1B · SLC · photons · point clouds · CPHD
- Umbra Very-High-Resolution SAR
- USGS 3DEP Airborne LiDAR Point Clouds
- ICESat-2 ATL03 Geolocated Photons
- Sentinel-1 Single Look Complex
Pre-processed / ARD
24L1C – L2 · calibrated · orthorectified · surface reflectance
- WorldView Legion 30 cm Tasking
- Pléiades Neo 30 cm
- Vantor VHR Archive (WorldView-2/3, GeoEye-1)
- BlackSky Gen-3 Rapid Revisit
Data products
14L3 – L4 · mosaics · land cover · detections · reanalysis
- Planet Global Monthly Basemaps
- ESA WorldCover 10 m
- Dynamic World Near-Real-Time Land Cover
- ALOS/ALOS-2 PALSAR Annual Mosaic
Optical to RF. Photons to phase.
One vocabulary across electro-optical, SAR, thermal, hyperspectral, LiDAR, atmospheric, RF and AIS — each a concept in the ontology, so agents reason about “all-weather” or “night-capable” instead of product names.
Built for machines that buy data.
Every page has a machine twin. A formal ontology, an MCP server, a JSON-Schema tool manifest and principal-issued spending mandates let an agent go from question to delivered pixels without a human in the loop — and stop for one when it should.
OWL ontology aligned to SOSA, DCAT, PROV-O, GeoSPARQL, ODRL and schema.org. Every entity dereferences to JSON-LD and Turtle.
15 tools from search to tasking over the Model Context Protocol, plus the same tools as REST functions.
Per-order limits, budgets and licence scopes. Over-mandate orders park for human approval.
Every response carries explicit next actions, so agents never guess the workflow.
// 1. Discover search_collections({ q: "ship detection", all_weather: true, max_gsd: 1 }) → capella-sar · iceye-sar · umbra-sar // 2. Understand describe_term({ term: "tf:allWeather" }) → xsd:boolean · domain tf:Collection // 3. Price get_quote({ collection: "umbra-sar", aoi: [56.0,26.3,56.2,26.5], mode: "tasking", priority: "priority" }) → $26,580 · valid 24h · ETA 11h // 4. Buy (bounded by mandate) place_order({ quote_id: "q_7f3…" }) → awaiting_approval · exceeds per-order limit $5,000 next: POST /v1/orders/o_91c…/approve (principal)
The EO backbone for next-generation C2.
OVERMATCH states the information need; TerraFabric answers with a ranked, costed collection plan across every open archive and commercial constellation, tasks the best sensor, and streams availability back — as JSON, GeoJSON, JSON-LD or Cursor-on-Target.
Open the C2 console- 01RequirementAOI, window, GSD, all-weather, night, latency, budget
- 02Coverage assessmentRanked sources · time-to-first-data · cost · layered plan
- 03TaskingPriority classes from standard to flash, tracked end-to-end
- 04Feed & CoTServer-sent availability events and TAK-ready Cursor-on-Target