charcoal-fading-travel-study
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-uploaded photos, which creates an attack surface for indirect prompt injection if those images contain embedded text or adversarial patterns.
- Ingestion points: Processes 'uploaded photos' and 'source images' as defined in the
SKILL.mddelivery guidelines. - Boundary markers: The instructions mandate that the agent 'Use the source as the sole content reference' and 'never invent, redesign, beautify, or replace the main subject,' which serves as a mitigation against obeying instructions found within the data.
- Capability inventory: The skill uses image generation and editing capabilities to produce its output.
- Sanitization: The skill lacks explicit sanitization or filtering logic to detect or ignore textual instructions embedded within visual input.
Audit Metadata