security-review
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill contains a large library of security payloads, such as SQL injection strings, command injection bypasses, and XSS payloads. These are correctly documented as test cases and audit material for the user to evaluate target applications, rather than malicious actions performed by the skill itself.
- [EXTERNAL_DOWNLOADS]: The
dependency-audit-guide.mddocumentation includes instructions for installing security tools such as Syft and Grype using official installation scripts from Anchore's GitHub repository. Anchore is a well-recognized security software vendor, and these references are appropriate for a security review context. - [INDIRECT_PROMPT_INJECTION]: By design, the skill uses tools like
WebFetchand file readers to ingest data from external targets for security analysis. While this creates a surface for indirect prompt injection, the skill's systematic methodology and structured reporting template provide a clear instructional context that mitigates the risk of the agent deviating from its audit task. - [SAFE]: Static detection hits for dangerous functions like
eval()andexec()were analyzed and found to be false positives. These functions are documented in thesecure-code-review-checklist-guide.mdas examples of insecure programming patterns that auditors should look for during source code reviews.
Audit Metadata