Agent Collaboration Framework

Security checks across malware telemetry and agentic risk

Overview

This is an instruction-only multi-agent collaboration skill with disclosed session routing and file handoff behavior that matches its stated purpose.

Install this only if you want OpenClaw agents to route work and share context with each other. Review any openclaw.json and AGENTS.md changes before applying them, keep allowed agents narrow, avoid secrets in HANDOFF files, and monitor or close persistent sessions when the collaboration task is finished.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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
Findings (2)

Vague Triggers

Medium
Confidence
92% confidence
Finding
The skill advertises very broad trigger phrases such as '跨部门', '协同', and '让XX做', which can match ordinary user conversation and cause the collaboration framework to activate unexpectedly. Because this skill can spawn sessions, route tasks across agents, and influence file-based handoff behavior, accidental invocation expands the operational surface and may trigger unintended agent actions.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The HANDOFF mechanism instructs agents to write files into other workspaces and delete them after reading, but it provides no user-facing warning, consent boundary, or validation controls around those modifications. In a multi-agent setting this is risky because arbitrary content can be persisted, consumed as trusted inter-agent input, or destroyed, creating opportunities for prompt injection persistence, tampering, loss of auditability, and unintended data modification.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal