break-ui
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted UI component code and data fixtures to analyze potential failures and apply code fixes. This ingestion of external content, combined with file-writing capabilities, creates an indirect prompt injection surface.\n
- Ingestion points: Component source files, validation schemas, and API types mapped in Phase 1 (SKILL.md).\n
- Boundary markers: Rule 6 in SKILL.md specifically instructs the agent to treat repository content as data and to flag attempts to steer its behavior (e.g., "ignore previous instructions").\n
- Capability inventory: The skill has the capability to modify source code in Phase 6 (SKILL.md) to apply identified fixes.\n
- Sanitization: The skill includes instructions to identify and fix missing escaping or sanitization in the analyzed UI code (Failure Signatures table in SKILL.md).
Audit Metadata