研究方法决策助手
Security checks across malware telemetry and agentic risk
Overview
The inspected skill artifacts are purpose-aligned and disclosed, with no evidence of hidden exfiltration, persistence, or destructive behavior outside their stated workflows.
Install only if you want these developer and maintainer workflows available to your agent. The ClawHub moderation and review helpers can run powerful commands using local auth and repository access, so keep them limited to trusted users and review commands before approving staff-impacting actions.
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
65/65 vendors flagged this skill as clean.