multi-agents-orchestration
Security checks across malware telemetry and agentic risk
Overview
This is a documentation-only skill for configuring OpenClaw multi-agent orchestration, with disclosed examples for background agents and optional Discord bots.
Install this only if you intend to configure multi-agent OpenClaw workflows. When copying examples, replace wildcard agent permissions with named allowlists, keep Discord bot tokens scoped to the required guilds/channels, leave bot-to-bot mentions disabled unless needed, and add your own approval or logging expectations for background sub-agent work.
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
66/66 vendors flagged this skill as clean.