The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to run Python logic using python3 -c. It instructs the agent to dynamically construct these commands by replacing placeholders (such as COLLECTION, ITEM_ID, and MINX) with user-supplied data.
[PROMPT_INJECTION]: The instructions lack input validation or sanitization requirements for variables interpolated into shell commands. This makes the skill susceptible to injection attacks where a user could provide crafted input (e.g., using quotes and semicolons) to break out of the Python string context and execute arbitrary code on the host system.
[EXTERNAL_DOWNLOADS]: The skill downloads satellite imagery from Microsoft Planetary Computer, a well-known service. The files are saved locally to the ./stac_data/ directory.
[REMOTE_CODE_EXECUTION]: The skill suggests the installation of the geoai-py Python package. This is an external dependency that is not part of the standard library and is fetched from an external registry.
[DATA_EXPOSURE]: The skill processes external metadata from the STAC API, which is then used as input for subsequent command executions, creating a surface for indirect prompt injection.
Ingestion points: Satellite metadata (item IDs, asset lists) retrieved from the Microsoft Planetary Computer API in Step 3 and Step 4.
Boundary markers: No boundary markers or instructions to ignore embedded commands are present in the processing logic.
Capability inventory: Access to the Bash tool and the ability to execute arbitrary Python code.
Sanitization: The skill does not perform any sanitization or escaping of the metadata retrieved from the API before using it in future execution steps.

search-stac