multi-agent-image
Warn
Audited by Socket on Sep 21, 2026
1 alert found:
AnomalyAnomalySKILL.md
LOWAnomalyLOW
SKILL.md
SUSPICIOUS. The skill’s general capabilities fit image generation, and its package installs are ordinary PyPI dependencies, but the stated credential and network flow is inconsistent with direct OpenAI usage: an OPENAI_API_KEY is sent to apimart-compatible third-party endpoints instead of clearly to official OpenAI APIs. That proxy-style routing is the main security concern and makes the skill medium risk despite otherwise coherent purpose.
Confidence: 87%Severity: 64%
Audit Metadata