dogfood

Pass

Audited by Gen Agent Trust Hub on May 15, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Recommends updating the agent-device tool using NPM. This is a standard dependency for the skill's functionality.
  • [PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection by processing mobile app UI snapshots. 1. Ingestion points: UI snapshots and screenshots (SKILL.md). 2. Boundary markers: Absent. 3. Capability inventory: Bash tool for agent-device commands. 4. Sanitization: Absent.
Audit Metadata
Risk Level
SAFE
Analyzed
May 15, 2026, 01:12 AM
Security Audit — agent-trust-hub — dogfood