ClawHub
Back to skill

Security audit

自定义多agent

Security checks across malware telemetry and agentic risk

Overview

This skill is a user-invoked writing-agent template generator with scoped, disclosed persistence for reusable local templates.

Safe to install if you want reusable writing-agent templates. Review any generated SOUL.md changes before applying them, and only save or keep custom templates you trust because they can override built-in templates and shape future agent behavior.

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 (2)

Intent-Code Divergence

Medium
Confidence
91% confidence
Finding
The checklist and file-location guidance state that SOUL.md has been 'created/updated' and reference backup paths, which contradicts the skill metadata saying it only generates configurations and does not modify user files. This can mislead an agent or user into believing file writes are expected behavior, increasing the risk of unintended overwrite of a primary workspace configuration file.

Missing User Warnings

Medium
Confidence
84% confidence
Finding
The skill describes automatically saving generated content to a local path under ~/.openclaw/workspace/agent-templates/ without a strong, repeated user-consent warning at the point of action. Silent or insufficiently disclosed local file creation can lead to unexpected persistence of untrusted prompts or configurations that later influence agent behavior.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.