ux-evaluate
Pass
Audited by Gen Agent Trust Hub on Apr 2, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill's architecture is restricted to the local file system with no network access, as specified in the allowed-tools configuration (Read, Write, Glob, Grep). This provides a strong technical safeguard against data exfiltration.
- [PROMPT_INJECTION]: The skill ingests untrusted HTML and Markdown data from project directories, creating a surface for indirect prompt injection. Ingestion points: Glob patterns in SKILL.md for HTML and MD files. Boundary markers: Explicit security warning in workflow step 2. Capability inventory: File reading and report writing only. Sanitization: Instruction-based filtering of external directives.
- [SAFE]: No obfuscation, persistence mechanisms, or unauthorized privilege escalation commands were detected in any of the analyzed files.
Audit Metadata