audit-safety
Pass
Audited by Gen Agent Trust Hub on Oct 23, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches audit checklists and definitions of related skills from the author's GitHub repository (github.com/Melvynx/ai-maxxing-skills). This is used to maintain up-to-date auditing criteria.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted application code and configurations to perform security audits, which creates an attack surface for instructions embedded in the analyzed data.
- Ingestion points: Web application source code, URLs, and directory structures provided as arguments (e.g., Next.js or TanStack Start projects).
- Boundary markers: The agent is instructed to explicitly state the inferred scope and report findings only with concrete paths (file:line), which helps maintain focus on the audit task.
- Capability inventory: Primarily read-only access to files; uses curl to fetch its own reference files from a specific GitHub path; can read other skills from the same repository.
- Sanitization: No specific sanitization or filtering of the analyzed code is mentioned; the skill relies on the agent's ability to distinguish between data being audited and instructions.
Audit Metadata