The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides extensive workflows for processing external data files (such as satellite imagery and vector layers) obtained from remote sources.
Ingestion points: Multiple scripts across the references/ directory use libraries like rasterio, geopandas, and fiona to read data from external files and cloud-based STAC catalogs.
Boundary markers: Code examples demonstrate direct processing of external content without explicit "ignore embedded instructions" delimiters.
Capability inventory: The skill allows the agent to execute shell commands and file operations to process large-scale geospatial datasets.
Sanitization: Input data is processed using standard scientific libraries without specific content validation or sanitization routines.
[EXTERNAL_DOWNLOADS]: The skill references and fetches geospatial data from numerous well-known scientific and government services.
Evidence: References data acquisition from the European Space Agency (ESA) Copernicus Data Space, USGS EarthExplorer, Microsoft Planetary Computer, and Google Earth Engine.
Documentation: The skill includes code for programmatic access to these repositories using official APIs and client libraries.
[PRIVILEGE_ESCALATION]: The troubleshooting documentation suggests using administrative commands for system-level dependency management.
Evidence: Mentions sudo apt-get install gdal-bin libgdal-dev in references/troubleshooting.md as a solution for resolving GDAL installation issues on Linux systems.

alterlab-geomaster