STAC GIS
STAC GIS is an open-source, AI-powered geospatial platform that bridges high-resolution environmental data cubes with conversational AI. It is a STAC-compliant Geographic Information System (GIS) that lets you query, analyze, and visualize geospatial assets using natural language — with all AI execution running locally, so your data never has to leave your servers.
The platform is made of three independent, self-contained components:
| Component | What it is | Repository |
|---|---|---|
| Core | The STAC-compliant platform: FastAPI backend, React frontend, Postgres/pgvector, MinIO storage, and the local AI chat | stac-gis/core |
| Geo-AI | stacgis-ai + stacgis-ai-react — a reusable harness for building geospatial AI agents (ReAct loop, MCP tools, FastAPI streaming, React chat UI) | stac-gis/geo-ai |
| Catalog | A Python toolkit (stac-catalog) that generates STAC Collections & Items from CSV/Google Sheets, loads them into pgSTAC, and serves them (STAC FastAPI + TiTiler + STAC Browser) | stac-gis/catalog |
Each component can be used standalone or combined into a full platform: Core stores and serves your geodata through a STAC API, Geo-AI powers the conversational agent that reasons over it, and the Catalog turns your raster metadata (CSV or Google Sheets) into a STAC catalog and serves it for browsing.
🗺️ Where to go next
- Platform overview — how the three components fit together, with the full architecture diagram.
- Core — Getting started — install and run the STAC GIS platform (dev and production).
- Geo-AI — Quick start — run the agent harness with Docker Compose or a manual setup.
- Catalog — Getting started — generate a STAC catalog from CSV or Google Sheets and serve it with one command.
📂 Component repositories
stac-gis/
├── core/ # STAC GIS Core — backend, frontend, and full-stack compose
├── geo-ai/ # Geo-AI — stacgis-ai (Python) + stacgis-ai-react (npm)
└── catalog/ # Catalog — stac-catalog Python STAC generator + serving stack
💡 Why STAC GIS?
- 🗺️ STAC-native — data models follow the SpatioTemporal Asset Catalog specification, so your assets stay interoperable and portable.
- 🤖 Local-first AI — the agent loop runs against locally hosted LLMs (Ollama) and MCP tools; no data leaves your infrastructure.
- 🧩 Composable — use the core platform, drop the agent harness into your own FastAPI/React apps, or just self-host a catalog for an existing STAC API.
- 🐋 Self-hosted by design — every component ships with Docker support and runs on commodity hardware.