ClawHub
Back to skill

Security audit

Meta Skill Writer

Security checks for vulnerabilities and agentic risk

Overview

This is a guide for writing OpenClaw skills; its trigger wording is broad, but it contains no executable code, credential use, persistence, or automatic actions.

Reasonable to install if you want a Chinese-language guide for creating skills. Before relying on it for published skills, refine generated descriptions so they are specific enough to avoid accidental activation, and review any scripts or install commands it helps create.

SkillSpector

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

High
Confidence
93% confidence
Finding
The frontmatter description is "帮我写个 skill", which is a very generic everyday phrase and can match a wide range of unrelated user requests. Overbroad triggers can cause unintended activation, making the agent apply this skill in the wrong context and potentially override more appropriate, narrower instructions.

Vague Triggers

Medium
Confidence
89% confidence
Finding
The guidance tells authors to copy direct user utterances into description fields without requiring boundaries, exclusions, or negative examples. This encourages creation of ambiguous triggers across many downstream skills, increasing accidental invocation and routing errors at scale.

Static analysis

No suspicious patterns detected.