The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes security analysis tools including npm audit, pip-audit, safety, and bandit through the Bash tool. These operations are limited to specific security auditing commands and are consistent with the skill's stated purpose.
[PROMPT_INJECTION]: The skill includes instructions to the agent to enforce security gates and follow specific audit methodologies. These instructions (e.g., 'Anti-Hallucination Guidelines') are designed to improve the accuracy and reliability of the security report rather than bypassing agent safety protocols.
[DATA_EXFILTRATION]: No patterns of unauthorized data transmission were found. The skill includes a WebSearch capability, which is standard for researching security vulnerabilities (CVEs) during an audit.
[CREDENTIALS_UNSAFE]: No hardcoded secrets are present. The skill provides Grep patterns for the agent to use when searching for potential credential leaks in the target codebase.

forge-security