mtm-image2
Warn
Audited by Socket on Jul 15, 2026
1 alert found:
AnomalyAnomalySKILL.md
LOWAnomalyLOW
SKILL.md
SUSPICIOUS. The core capability—sending a prompt to a remote image API and saving a PNG—matches the stated purpose, and the stdin design reduces injection risk. But the nonstandard `MTMAI_IMAGE2_KEY`, missing endpoint disclosure, and evidence pointing to a third-party `sub2`/gateway ecosystem make the credential and data flow inconsistent with a clearly direct OpenAI integration. Without the referenced docs or script contents, the skill cannot be verified as benign end-to-end.
Confidence: 82%Severity: 66%
Audit Metadata