auto-review-loop
Warn
Audited by Socket on Oct 27, 2026
1 alert found:
SecuritySecuritySKILL.md
MEDIUMSecurityMEDIUM
SKILL.md
SUSPICIOUS. The skill’s core behavior mostly matches its stated purpose as an autonomous review/improvement loop, and the named external tools are official/same-org developer tooling rather than obvious malware. However, it is still a medium/high-risk skill because it autonomously edits code, runs experiments, may access remote servers via SSH, forwards repository contents to external reviewer backends, and explicitly instructs silent execution in at least one path. This is not confirmed malware, but it grants broad real-world action capability disproportionate to a simple 'review' helper.
Confidence: 88%Severity: 71%
Audit Metadata