Agent Orchestrate
Security checks across malware telemetry and agentic risk
Overview
This is a documentation-only OpenClaw skill for coordinating sub-agents, with expected cautions around sharing sensitive data between spawned agents.
Install if you want quick-reference patterns for OpenClaw multi-agent workflows. Avoid delegating secrets, credentials, personal data, or regulated material unless necessary, and scope what each spawned agent receives before using fan-out or synthesis patterns.
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.