The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses python3 -c to execute Python code blocks that incorporate variables like DATA_TYPE, MINX, MINY, MAXX, MAXY, and OUTPUT_PATH. If the agent does not strictly validate these inputs from the user, it could lead to command injection within the Python string literal.
[PROMPT_INJECTION]: The skill processes external Overture Maps data and displays the head of the resulting GeoDataFrame (gdf.head().to_string()). This creates an indirect prompt injection surface where a malicious data source could contain text designed to influence the agent's behavior when the data is rendered in the terminal context.
Ingestion points: Data is fetched from Overture Maps via the geoai library and displayed in the terminal via gdf.head() (SKILL.md).
Boundary markers: None identified. The data is printed directly to stdout.
Capability inventory: The skill has Bash access, allows python3 execution, and performs file system writes (json.dump to state files).
Sanitization: No evidence of sanitization or escaping of the GeoDataFrame content before it is printed to the agent's context.
[COMMAND_EXECUTION]: The skill executes shell commands to update its state, including git rev-parse and file system checks. While these are localized to the project environment, they represent an execution surface that relies on the environment's integrity.

overture-data