pentestcode-ai-pentest-agent

Warn

Audited by Socket on Sep 18, 2026

1 alert found:

Security
SecurityMEDIUM
SKILL.md

SUSPICIOUS. The offensive capabilities match the skill's stated pentesting purpose, but the footprint is still high risk: it installs a remote script via curl|bash from an apparently different publisher identity, stores discovered credentials, and enables autonomous exploitation workflows including scope bypass. This looks coherent as a pentest skill, not overt credential theft, but it is dangerous and should not be treated as benign.

Confidence: 83%Severity: 88%
Audit Metadata
Analyzed At
Sep 18, 2026, 05:11 PM
Package URL
pkg:socket/skills-sh/reason-machines%2Fsecurity-skills%2Fpentestcode-ai-pentest-agent%2F@aa327e50bf0049a1596905d893e42014aab6ed9ae7586513e5959981eb6b665e