The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests potentially untrusted cloud configuration data, creating a surface for indirect prompt injection attacks.\n- Ingestion points: The scripts/cloud_posture_check.py script reads and parses JSON configuration files provided as input to the tool.\n- Boundary markers: No specific delimiters or instructions are implemented in scripts/cloud_posture_check.py to distinguish data from potential instructions within the JSON content.\n- Capability inventory: The skill provides instructions in SKILL.md for the agent to execute cloud CLI commands and perform security assessments based on the configuration findings.\n- Sanitization: The script scripts/cloud_posture_check.py lacks validation or sanitization of string values within the JSON files for natural language instructions that could influence agent behavior.

cloud-security