posttrainer
Warn
Audited by Socket on Jul 10, 2026
1 alert found:
SecuritySecuritySKILL.md
MEDIUMSecurityMEDIUM
SKILL.md
SUSPICIOUS. The skill's training workflow broadly matches its stated purpose, but it asks users to send HF/WandB credentials to third-party remote worker code, relies on a mutable personal-account container image, and includes transitive skill installation via remote curl. These are meaningful supply-chain and credential-forwarding risks even though the overall use case is plausible.
Confidence: 86%Severity: 76%
Audit Metadata