The Agent Skills Directory

[PROMPT_INJECTION]: Benign prompt injection examples detected in references/ai-agent-security.md and references/owasp-checklists.md are educational reference data for the agent to use during security audits. They are not instructions to subvert the agent's behavior.
[COMMAND_EXECUTION]: The skill provides pre-configured commands to run its internal security auditing scripts. No arbitrary command execution from untrusted sources or unsafe argument interpolation was detected.
[EXTERNAL_DOWNLOADS]: As stated in the scripts/requirements.txt and verified in the source code, the toolset uses zero external dependencies, relying exclusively on the Python standard library to eliminate supply chain risks.
[DATA_EXFILTRATION]: All data processing and report generation are strictly local. No network operations or exfiltration patterns were identified in the script implementations.

007