The Agent Skills Directory

[PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection through data ingestion in the contestation processing script.\n- Ingestion points: The scripts/process.py script accepts untrusted user input via the --grounds and --name parameters.\n- Boundary markers: The script does not implement specific delimiters or instructions to ignore embedded commands within the user-provided input strings in its output.\n- Capability inventory: The skill provides local data processing and JSON generation; it does not contain network, file-system write, or other high-privilege operations in its scripts.\n- Sanitization: Input is serialized using standard json.dumps(), which ensures structural integrity but does not filter the semantic content of the strings for instruction injection.\n- [COMMAND_EXECUTION]: The skill utilizes a local Python script (scripts/process.py) to automate decision assessment and contestation processing workflows.

automated-decision-rights