The Agent Skills Directory

[SAFE]: The skill's components are consistent with its stated purpose of performing security audits. The provided files consist of documentation templates and a basic file-processing script.
[PROMPT_INJECTION]: No direct prompt injection or instructions to bypass safety guidelines were found in the skill's instructions or metadata.
[DATA_EXFILTRATION]: No hardcoded credentials, sensitive file path access, or unauthorized network operations were detected. The scripts/security_scan.sh script (which contains Python code) performs local file metadata checks and JSON validation without external transmission.
[COMMAND_EXECUTION]: Although the skill specifies Bash(cmd:*) in its allowed tools, there is no evidence of arbitrary or dangerous command execution logic within the provided scripts.
[PROMPT_INJECTION]: The skill processes external files and code for auditing purposes, which constitutes a potential surface for indirect prompt injection. However, no specific exploitable logic was identified.
Ingestion points: User-provided source code, configurations, and infrastructure data (referenced in SKILL.md).
Boundary markers: None explicitly defined in the provided templates to delimit untrusted content.
Capability inventory: Bash, Write, Edit, Grep, Glob (per SKILL.md frontmatter).
Sanitization: No sanitization or validation of the content being audited is performed in the template script (scripts/security_scan.sh).

performing-security-audits