Geoagent
A multimodal AI agent for geospatial data analysis and interactive visualization
Last verified:
What is Geoagent?
GeoAgent is an open-source AI agent framework designed for geospatial data analysis and visualization. It provides a shared integration layer for the geospatial Python and QGIS ecosystem, allowing libraries like leafmap, aniamap, geoai, geemap, STAC, and NASA Earthdata to expose their tools to large language models through one consistent interface. GeoAgent uses a 4-agent LangGraph pipeline consisting of a Planner that parses natural language into structured parameters, a Data Agent that searches STAC catalogs and retrieves geospatial data, an Analysis Agent that computes indices and statistics with transparent code generation, and a Visualization Agent that renders results on interactive leafmap MapLibre maps.
Key features include a natural language interface for geospatial workflows, multi-LLM support (OpenAI, Anthropic, Google Gemini, Ollama), multi-catalog STAC search across Earth Search, Planetary Computer, USGS, and NASA CMR, code transparency showing generated Python code at each step, Jupyter-native integration with interactive MapLibre maps via leafmap, DuckDB spatial SQL for GeoParquet and Overture Maps, raster analysis with xarray/rioxarray/rasterio, and vector operations with geopandas. It also includes a Solara-based web UI chat interface with a persistent interactive map.
GeoAgent is designed for geospatial analysts, data scientists, GIS professionals, researchers working with Earth Observation data, and anyone using Python or QGIS for geospatial analysis. It supports QGIS through the OpenGeoAgent plugin which adds a dockable project-aware AI assistant, and works in Jupyter notebooks, Python scripts, and browser-based MapLibre applications. The tool is particularly useful for automating GIS workflows, generating maps, analyzing satellite data, and running complex hydrological models using natural language commands.
Geoagent pricing
Pricing model: Freemium
GeoAgent is completely free and open-source under the MIT License. The core package is installable via pip install geoagent with no cost. All features including the 4-agent pipeline, STAC search, QGIS plugin, web UI, and integrations with leafmap, NASA Earthdata, and Google Earth Engine are free. Optional dependencies for different LLM providers have separate pip extras (geoagent[llm], geoagent[ollama], geoagent[ui], geoagent[all]) but the software itself has no paid tiers. Users only pay for external LLM API usage (OpenAI, Anthropic, Google Gemini) if they choose those providers - local Ollama models require no API key or payment.
Geoagent pros
- Natural language interface for geospatial workflows eliminates complex coding
- Open-source MIT license with no licensing fees
- Four specialized agents (Planner, Data, Analysis, Visualization) for structured workflow
- Multi-LLM support including OpenAI, Anthropic, Google Gemini, and local Ollama
- Multi-catalog STAC search across Earth Search, Planetary Computer, USGS, NASA CMR
- Code transparency shows generated Python code at every step
- Jupyter-native with interactive MapLibre maps via leafmap integration
- QGIS plugin (OpenGeoAgent) adds dockable AI assistant to QGIS
- DuckDB spatial SQL support for GeoParquet and Overture Maps
- Raster analysis with xarray, rioxarray, and rasterio
- Vector operations with geopandas
- Solara-based web UI with persistent interactive chat map
- Voice command support for hands-free interaction
- Multimodal support for image paste and screenshot attachments
- Safety features with confirmation-gated tools for destructive operations
- Fast mode for lower-latency map control with reduced tool surface
- NASA Earthdata and OPERA product integration built-in
- Google Earth Engine integration available
- Browser-based MapLibre app support with WebSocket backend
- Active development with 350 GitHub stars and 16 releases
Geoagent cons
- Requires API key for most LLM providers (OpenAI, Anthropic, Gemini)
- QGIS plugin requires QGIS to be separately installed system-wide
- Geospatial packages are optional extras requiring additional installation
- Some providers like Anthropic/Gemini require pip install extras
- Local Ollama requires separate Ollama installation and model pulling
- Confirmation gates may slow down trusted workflows unless auto-approved
- Complex geospatial operations may generate expensive API calls
- Browser backend requires FastAPI WebSocket setup for MapLibre apps
- Natural language OPERA chat disabled inside QGIS currently
- LLM call dominates latency especially for local models
Frequently asked questions about Geoagent
What is GeoAgent?
GeoAgent is an open-source AI agent framework for geospatial data analysis and visualization. It provides a shared integration layer for geospatial Python packages, live map widgets, and QGIS plugins, allowing multiple libraries to expose their tools to large language models through one consistent interface. It uses a 4-agent LangGraph pipeline: Planner, Data Agent, Analysis Agent, and Visualization Agent.
Which LLM providers does GeoAgent support?
GeoAgent supports OpenAI (default model gpt-4.1), Anthropic (claude-sonnet-4-5-20250929), Google Gemini (gemini-2.5-flash), and Ollama for local models (llama3.1). It also supports ChatGPT/Codex OAuth, LiteLLM, vLLM, and Amazon Bedrock. The first available provider is auto-detected from environment variables in order: OpenAI → Anthropic → Google → Ollama.
Do I need an API key to use GeoAgent?
You need at least one LLM provider configured to use GeoAgent. OpenAI, Anthropic, and Google Gemini require API keys set as environment variables (OPENAI_API_KEY, ANTHROPIC_API_KEY, GOOGLE_API_KEY). However, Ollama (local LLMs) requires no API key - just install Ollama, pull a model, and install the geoagent[ollama] extra.
What is the OpenGeoAgent QGIS plugin?
OpenGeoAgent is the QGIS plugin interface for GeoAgent. It adds a dockable, project-aware AI assistant to QGIS that connects to GeoAgent's capabilities. Features include provider/model controls, streaming responses, image paste and screenshot attachments, screenshot capture from map canvas, confirmation-gated PyQGIS script execution, and the ability to inspect projects, navigate the map, load data, run workflows, and style layers from natural language.
What STAC catalogs can GeoAgent search?
GeoAgent supports multi-catalog STAC search across Earth Search, Planetary Computer, USGS, and NASA CMR (Climate Model Registry). The Data Agent searches these catalogs to retrieve geospatial data based on your natural language query parameters like location, date range, and data type.
What is GeoAgent's code transparency feature?
Code transparency means GeoAgent shows the generated Python code at each step of the workflow. After running a query, you can access result.code to see the exact Python code that was generated and executed. This allows users to understand, verify, and learn from the agent's approach, and to reuse the code for their own scripts.
Is GeoAgent suitable for production use?
GeoAgent is open-source MIT-licensed and actively maintained with 350 GitHub stars and 16 releases (latest v1.8.1 in May 2026). It includes safety features like confirmation-gated tools for destructive operations, supports enterprise LLM providers, and has been used for NASA Earthdata workflows. However, as with any AI agent, you should review generated code and use confirmation callbacks for production workflows involving expensive API calls or data modifications.