The Agent Skills Directory

[SAFE]: The skill implements a security auditing workflow for reviewing source code and git commits against OWASP and CWE patterns. All operations align with its stated purpose as a security analysis tool.
[SAFE]: Deterministic detectors flagged the presence of eval in reference/wstg-testing.context. This is a false positive; the code snippet is a documentation example illustrating an insecure coding pattern to be identified during a review, rather than executable code within the skill itself.
[PROMPT_INJECTION]: This skill possesses an inherent surface for indirect prompt injection due to its primary function of processing untrusted data.
Ingestion points: Source files, directories, documentation, and git commits are loaded into the agent's context as specified in the workflow in SKILL.md.
Boundary markers: The workflow does not utilize specific delimiters or XML-based markers to isolate the ingested code being analyzed from the agent's instructions.
Capability inventory: The skill performs complex multi-file reads, generates findings with remediation steps, and can chain execution to related security skills like secdevai-oci-image-security.
Sanitization: There is no automatic sanitization or escaping of the input files before they are processed for analysis.
Mitigation: The author includes a proactive security note in SKILL.md (Step 2) stating that documentation files are included in the review specifically because they 'can contain... AI agent instructions', alerting the agent to maintain its objective role when encountering adversarial content.
[SAFE]: External context files are referenced from the author's own organizational infrastructure (RedHatProductSecurity/secdevai), representing standard vendor behavior for distributing security rule sets.

secdevai-review