flux-best-practices
Warn
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: MEDIUMMETADATA_POISONINGINDIRECT_PROMPT_INJECTION
Full Analysis
- [METADATA_POISONING]: The
SKILL.mdandAGENTS.mdfiles identify the author as "Black Forest Labs", while the actual author is "joeplouis". This misrepresentation can mislead users regarding the skill's official status and safety. - [INDIRECT_PROMPT_INJECTION]: The skill facilitates the processing of untrusted data from external sources.
- Ingestion points: External image URLs provided to the
input_imageparameter inrules/i2i-prompting.mdandrules/multi-reference-editing.md. - Boundary markers: No delimiters or warnings are specified to protect against instructions embedded in the external data.
- Capability inventory: The skill interacts with image generation APIs (referenced in
SKILL.md) that perform network requests to fetch external assets. - Sanitization: No instructions are provided for the validation or sanitization of input URLs or structured JSON data in
rules/json-structured-prompting.md.
Audit Metadata