ClawHub
Back to skill

Security audit

Validate Agent

Security checks across malware telemetry and agentic risk

Overview

This is an instruction-only validation checklist that may activate broadly but does not include code, credential access, networking, or hidden behavior.

Safe to install as a validation aid. Use it deliberately for important answers, code fixes, and handoffs if broad automatic activation becomes noisy, and do not store secrets or sensitive business details in promises.md.

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 declares trigger conditions such as '完成任务时' and '提供答案前', which overlap with routine agent behavior and can cause the skill to activate on a large fraction of normal interactions. Over-broad activation increases prompt injection surface, instruction interference, and workflow disruption because the skill may insert itself into unrelated tasks without clear user intent or scoping.

Vague Triggers

Medium
Confidence
95% confidence
Finding
The trigger list is ambiguous and lacks limiting conditions, so different agents or runs may interpret it inconsistently and invoke the skill in ordinary workflows. This creates reliability and security risk by making behavior non-deterministic and allowing the skill to override or delay primary task execution too broadly.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.