gpt-image-2
Warn
Audited by Socket on May 10, 2026
1 alert found:
AnomalyAnomalySKILL.md
LOWAnomalyLOW
SKILL.md
SUSPICIOUS: the skill is coherent as a Frevana image wrapper, but it is not a direct OpenAI integration despite framing around OpenAI's gpt-image-2. User prompts, local files, downloaded remote images, and credentials flow through Frevana's backend and storage layer, so trust and data exposure are broader than a first-party OpenAI skill.
Confidence: 84%Severity: 62%
Audit Metadata