media-higgsfield-image
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill uses a 'Live Catalog' design pattern, fetching valid parameters and model IDs from a server-side tool (
models_explore) at runtime. This prevents common errors and potential injection risks associated with hardcoded or guessed parameters. - [SAFE]: All external references point to reputable technology providers and documentation sources (e.g., Google, OpenAI, Black Forest Labs, ByteDance). No suspicious downloads or remote code execution patterns were detected.
- [SAFE]: The skill implements a structured workflow including intent identification, parameter validation via cost pre-flight checks (
get_cost), and clear output reporting. It also includes specific safety mentions regarding NSFW and copyright compliance. - [SAFE]: Analysis for obfuscation, privilege escalation, and persistence found no malicious patterns. The content is transparent and instructional.
Audit Metadata