Tianshu Short Script
Security checks across malware telemetry and agentic risk
Overview
The skills are mostly coherent, but the bundled autoreview helper gives nested review agents full unsandboxed authority by default, which deserves manual review before use.
Install only if you trust the publisher and need these ClawHub/Convex operational workflows. Before using autoreview, prefer the documented '--no-yolo' or AUTOREVIEW_YOLO=0 path unless you intentionally want a nested reviewer running with full local authority. Use the moderation skills only with explicit targets, reasons, and an authenticated account whose permissions you intend to exercise.
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
64/64 vendors flagged this skill as clean.