websecurityacademy-solutions
Pass
Audited by Gen Agent Trust Hub on Jun 13, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill contains literal strings such as 'Ignore previous instructions' and 'Delete all user data'. These are flagged as prompt injection attempts; however, they are documented payloads within the 'Web LLM Attacks' section intended for security lab practice and do not target the agent's own control logic.
- [EXTERNAL_DOWNLOADS]: The skill provides instructions to clone an external repository ('github.com/ntrunr/WebSecurityAcademy.git'). This is a direct reference to the skill's primary content (lab solutions) and is considered safe within its educational context.
- [COMMAND_EXECUTION]: The skill contains numerous code snippets for SQL injection, XSS, and command injection (e.g., 'nslookup', 'sleep 10'). These are provided as static examples for manual use in controlled lab environments and are not executed automatically by the skill.
- [INDIRECT_PROMPT_INJECTION]: The inclusion of common jailbreak and injection payloads in the skill body represents a minor surface for accidental obedience by the processing LLM. However, these are clearly delimited as examples within markdown code blocks, which mitigates the risk of the agent following them as instructions.
Audit Metadata