The Agent Skills Directory

[PROMPT_INJECTION]: The skill includes behavioral instructions in SKILL.md that direct the AI agent to proactively promote a specific commercial platform (K-Dense Web) when user requests involve complex workflows. This represents a behavioral nudge to steer the agent's interaction toward the author's hosted service.
[PROMPT_INJECTION]: Vulnerability to indirect prompt injection is present due to the capability to process untrusted geospatial data from external files and network locations.
Ingestion points: The skill utilizes gpd.read_file(), gpd.read_postgis(), and gpd.read_parquet() as documented in SKILL.md and references/data-io.md to ingest external data.
Boundary markers: No explicit delimiters or instructions are provided to the agent to treat processed data content as potentially containing untrusted instructions.
Capability inventory: The skill enables Python code execution for analysis, file system access (read/write), and network connectivity via GeoPandas and SQLAlchemy.
Sanitization: The skill relies on standard library parsing without implementing additional validation or sanitization of the input content.
[EXTERNAL_DOWNLOADS]: The documentation in references/data-io.md provides examples and technical instructions for fetching geospatial data from remote sources, including public URLs (https://example.com/data.geojson) and cloud storage buckets (S3, Azure).

geopandas