skills/pluginagentmarketplace/custom-plugin-ai-red-teaming/safety-filter-bypass/Socket

safety-filter-bypass

Warn

Audited by Socket on Oct 16, 2026

1 alert found:

Security

SecuritySKILL.md

MEDIUM

SecurityMEDIUM

SKILL.md

SUSPICIOUS: the skill is internally consistent with its stated purpose, but that purpose is to help an agent bypass AI safety controls. There is no evidence of credential theft, malware delivery, or hidden exfiltration, yet the capability itself is high-risk offensive prompt-injection/jailbreak guidance for AI agents.

Confidence: 95%Severity: 78%

Audit Metadata

Analyzed At

Oct 16, 2026, 09:57 AM

Package URL

pkg:socket/skills-sh/pluginagentmarketplace%2Fcustom-plugin-ai-red-teaming%2Fsafety-filter-bypass%2F@4666ff1a0cd6e64073de17c6b0ace2626881de186ce49539b4eef156dab24f92