break-ui
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill contains a strong defensive directive instructing the agent to treat all project files as data and specifically to ignore or flag any 'ignore previous instructions' patterns encountered during analysis. This effectively mitigates risks from processing untrusted codebases.
- [SAFE]: All test data provided in the catalog, including emails and URLs, correctly uses non-routable domains such as 'example.com' or '.test', preventing accidental data exfiltration or unintended network requests.
- [SAFE]: The skill's operating posture emphasizes data-driven testing through existing component boundaries (props, fixtures, mocks) rather than modifying core logic or styles, maintaining a clear separation of concerns.
Audit Metadata