wjs-promoting-skills
Audited by Socket on Sep 18, 2026
2 alerts found:
SecurityAnomalySUSPICIOUS. The skill’s capabilities mostly match its stated marketing purpose, and data flows go to official Anthropic/X endpoints rather than an obvious proxy. However, it enables autonomous public posting on a schedule and relies on an external CLI (`xurl`) that handles X credentials, making the overall risk medium-high even without clear malware indicators.
No clear malware, exfiltration, credential theft, or destructive payload is present. The script has input-validation weaknesses: an untrusted skill name can produce path traversal and malformed sed substitution, and the headless Claude invocation grants potentially broad repository-related Bash capabilities. Restrict SKILL to a safe basename, escape it before sed substitution, and narrow or remove the git tool permissions.