natural-japanese

Warn

Audited by Socket on Oct 24, 2026

1 alert found:

Anomaly
AnomalyLOW
scripts/semantic.py

The fragment appears to be a legitimate experimental semantic-analysis CLI and contains no direct evidence of malware, credential theft, destructive behavior, or data exfiltration. The material security concern is supply-chain/code-execution risk from unrestricted model selection combined with trust_remote_code=True and automatic remote model downloads. Restrict model identifiers to a trusted allowlist, disable trust_remote_code unless required, and pin or verify model artifacts. The local sys.path modification also warrants integrity protection for the bundled textcore module.

Confidence: 97%Severity: 62%
Audit Metadata
Analyzed At
Oct 24, 2026, 05:29 PM
Package URL
pkg:socket/skills-sh/rebuildup%2Fwriting-skills%2Fnatural-japanese%2F@9d86e219ebf98760bd99929e8c1f17922efb7cea5a5b94d3bc0dc6f0c00552be