multi-agent-deep-research
Security checks across malware telemetry and agentic risk
Overview
This is a structured research workflow skill that saves research ledgers in the user’s workspace and shows no hidden execution, credential handling, or destructive behavior.
Install this if you want OpenClaw to run research projects with local ledgers and auditable claims. Be aware it intentionally writes project state and drafts into your workspace, and only use the optional multi-agent workflow when you are comfortable with separate agents writing to clearly scoped files.
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.