ClawHub
Back to skill

Security audit

New Agent Onboarding

Security checks across malware telemetry and agentic risk

Overview

This is a documentation-only agent onboarding/training skill, but it teaches agents to maintain local memory and user-profile files.

Install this only if you want agents to follow a local training manual that reads and updates memory/profile files. Before using it for sub-agents, confirm which agent folders may be changed and whether USER.md should be copied, because that can duplicate private user context across local agent workspaces.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • 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 (1)

Description-Behavior Mismatch

Medium
Confidence
92% confidence
Finding
The skill is framed as a training/manual document, but it instructs the agent to create, copy, and update operational state files for other agents such as IDENTITY.md, SOUL.md, MEMORY.md, HEARTBEAT.md, and USER.md. This creates a capability mismatch: a user or orchestrator invoking what appears to be documentation can trigger persistent cross-agent state changes, which increases the risk of unauthorized configuration drift, privacy leakage, and covert prompt-state manipulation.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.