A field guide to AI agents that read the Earth

What it does This is the companion repository to a formal survey on Intelligent Remote Sensing Agents. It collects and categorizes over 100 papers, datasets, and benchmarks that apply LLM-based agents to satellite and aerial imagery. Each entry is tagged by both its real-world domain—such as ecological monitoring, urban governance, or emergency response—and by its agentic architecture, including memory, tool use, planning, and collaboration.

The interesting bit Instead of a flat bibliography, the maintainers use a dual-badge system: every paper gets an application tag and an agent-design tag. This makes it possible to quickly locate, say, a multi-agent climate system or a tool-using geological explorer, and a roadmap diagram visualizes the field’s evolution from classical remote sensing to autonomous agents.

Key highlights

• 100+ papers across seven application domains, from marine supervision to precision agriculture
• Dual-tagging with purple application badges and green design badges (e.g., “External Memory,” “Execution via Programmatic Actions”)
• Dedicated sections for datasets and benchmarks, including recent 2026 entries like OpenEarth-Bench and GeoMMBench
• Active curation with pull requests welcome and direct links to arXiv, publications, and code repos
• Associated with a formal survey paper with a provided BibTeX citation

Caveats

• This is a curated reading list and survey companion, not a runnable agent framework or codebase
• The content is under a CC BY-NC 4.0 license, which limits commercial reuse

Verdict Worth bookmarking if you are researching or building at the intersection of geospatial AI and agentic systems; skip it if you are looking for a drop-in remote sensing library to import.