travel-city-walk-by-1oyim0r
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to access and analyze media evidence from external URLs, specifically goodcase.ai and x.com, to derive cinematic traits for prompt generation.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a potential attack surface where instructions could be embedded in the external media content it is directed to analyze.
- Ingestion points: External media URLs and metadata found in references/cases.md.
- Boundary markers: The skill implements a 'reference contract' (Preserve, Replace, Avoid) and a verification checklist that constrains the agent to structural traits, reducing the risk of following adversarial instructions found in external data.
- Capability inventory: Utilizes browsing, image, and video analysis tools to process remote content.
- Sanitization: Relies on a multi-step workflow that requires explicit selection of observed traits before generating artifacts.
Audit Metadata