Security audit
Skill Reader
Security checks across malware telemetry and agentic risk
Overview
The artifacts are mostly coherent ClawHub/Convex helper skills, but one review helper defaults to launching a nested Codex review with approval and sandbox bypass enabled.
Review the autoreview helper before using it. Prefer running it with `--no-yolo` or `AUTOREVIEW_YOLO=0`, and only run moderation, PR commenting, publishing, or Convex deployment commands when you intend those external changes.
SkillSpector
By NVIDIA
Vulnerability Patterns
- Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
- Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
- Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
- Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
- Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
VirusTotal
42/42 vendors flagged this skill as clean.
Static analysis
No suspicious patterns detected.