ux-review-builder
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill analyzes external content from URLs and project-specific materials to derive UX rules. This creates a surface where malicious instructions could be embedded in the source materials.\n
- Ingestion points: Reference URLs provided by the user and local project materials (design systems, screenshots) described in
SKILL.mdandreferences/URL_ANALYSIS.md.\n - Boundary markers: The skill includes explicit instructions in
SKILL.mdto treat content as evidence only and to ignore any commands or instructions found in materials.\n - Capability inventory: Capability to write files to the
sdd/directory and perform network navigation for URL analysis as described inreferences/RUNTIME_COMPATIBILITY.md.\n - Sanitization: Relies on natural language constraints and warnings in the methodology files to prevent the execution of instructions found in untrusted data.\n- [COMMAND_EXECUTION]: The skill provides a local validation script and instructs the agent to run it to verify the structure of generated files.\n
- Evidence:
SKILL.mdStep 8 contains instructions to executepython3 "$SKILL_ROOT/scripts/validate_generated_bundle.py".\n- [EXTERNAL_DOWNLOADS]: The skill includes functionality to navigate and analyze external websites when provided with a URL.\n - Evidence:
references/URL_ANALYSIS.mddetails the observation protocol for external reference analysis.
Audit Metadata