ClawHub
Back to skill

Security audit

Quality Boost - 大模型回答质量提升器

Security checks for vulnerabilities and agentic risk

Overview

This is a prompt-only quality-improvement skill that is transparent about its optional global configuration and does not include code, credential use, network access, or hidden data handling.

Install is reasonable if you want a Chinese prompt-engineering quality checklist. Before enabling autoApply or systemPrompt prepend, review the prompt, back up or note your OpenClaw config, and test manual quick-apply use first because global configuration can change the style and flow of all future agent responses.

SkillSpector

Vulnerability Patterns
  • System Prompt LeakageDirect Leakage, Indirect Extraction, Tool-Based Exfiltration
  • 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
Findings (2)

Vague Triggers

Medium
Confidence
89% confidence
Finding
The trigger phrases are broad enough to activate on ordinary user requests about improving answer quality, which can cause the skill to apply unexpectedly outside the user's explicit intent. This creates prompt-routing risk: unrelated conversations may be steered into this skill's restrictive workflow, altering behavior, output style, or constraints without clear user consent.

Natural-Language Policy Violations

Medium
Confidence
95% confidence
Finding
This skill hard-codes Chinese-language system prompt content and instructs users to prepend it globally via OpenClaw configuration, which can force Chinese behavior across all interactions without user opt-in or locale negotiation. Because it is injected as a systemPrompt in prepend mode, it can override expected language behavior, reduce usability for non-Chinese users, and create prompt-control issues across unrelated tasks.

Static analysis

No suspicious patterns detected.