prompt-injection-test
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [SAFE]: The skill provides a structured methodology for performing security assessments and red-teaming on LLM applications, based on the Arcanum PI Taxonomy and OWASP standards.
- [SAFE]: All mentions of attack intents and techniques, such as jailbreaking, system prompt leakage, and various obfuscation layers, are presented as test cases for defensive validation rather than instructions to the agent.
- [SAFE]: The skill references a well-known security taxonomy repository on GitHub for research purposes.
- [NO_CODE]: No scripts, binaries, or automated shell commands are present within the skill files.
Audit Metadata