travel-city-walk
Pass
Audited by Gen Agent Trust Hub on Nov 16, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill provides numerous URLs to external media hosted on
goodcase.aiandmedia.goodcase.ai, as well as references to social media platforms (x.com,tiktok.com). The workflow requires the agent to 'Inspect the anchor's finished media with an available image, video, or browser tool' to ensure the generated artifacts maintain visual fidelity. These downloads are central to the skill's primary purpose of grounded video prompting. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data by instructing the agent to analyze the content of external images, videos, and webpages. This creates an attack surface for indirect prompt injection. However, the skill implements a 'Required evidence gate' and a 'Reference contract' (Preserve/Replace/Avoid) which acts as a boundary to separate external influences from the agent's core instructions. Additionally, the skill explicitly warns the agent to distinguish observed evidence from its own recommendations.
Audit Metadata