Agent Recruitment

Security checks across malware telemetry and agentic risk

Overview

This is a disclosed agent-administration skill, but it can create persistent agent workspaces and route every message from a Feishu group to a bot without enough approval or rollback detail.

Install only if you want this skill to perform admin-style OpenClaw agent creation and Feishu route setup. Before using it, require explicit approval for each group route, prefer @mention-only routing unless all participants expect full-message processing, review generated files under ~/.openclaw/agents, and document how to disable routes and delete retained memory.

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

Findings (1)

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The skill explicitly instructs the agent to create directories and configuration files under a local path and to write routing rules that can redirect messages, but it does so without requiring explicit user confirmation, preview, or safety checks before making those changes. In an agent-execution context, this can cause unauthorized local file modification and broad message-routing changes, including auto-responding to all messages in a group, which meaningfully expands the blast radius beyond a simple documentation issue.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal