codex-session-patcher
Warn
Audited by Socket on Sep 14, 2026
1 alert found:
SecuritySecuritySKILL.md
MEDIUMSecurityMEDIUM
SKILL.md
SUSPICIOUS. The skill's stated goal—cleaning refusals and injecting pentest/CTF prompts to reduce future refusals—is inherently about weakening safety behavior, and its footprint includes modifying local AI session/config files plus optional transmission of session data and API keys to configurable third-party endpoints. The main concrete security issue is install trust: the repo source does not match the named publisher, so users are asked to execute unverified third-party code via git clone, pip, and npm.
Confidence: 88%Severity: 72%
Audit Metadata