flux-best-practices
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [NO_CODE]: The skill consists entirely of Markdown files providing guidelines and instructions. It contains no executable scripts, shell commands, or platform-specific code triggers that could run at skill load or execution time.
- [INDIRECT_PROMPT_INJECTION]: The skill instructs agents on how to ingest and process external data, specifically user-supplied text prompts and image URLs. This design creates an inherent surface for indirect prompt injection (IPI), which is a standard characteristic of image generation tools and not an indication of malicious intent in this context.
- [METADATA_POISONING]: The metadata in
SKILL.mdattributes authorship to 'Black Forest Labs', while the platform context identifies 'guyronnen' as the skill author. This inconsistency is noted as metadata poisoning, though in this case, it appears to be a reference to the source of the technology (the FLUX models) rather than a deceptive attempt to hide identity.
Audit Metadata