ClawHub
Back to skill

Security audit

ThoughtCodec 思维编解码

Security checks for vulnerabilities and agentic risk

Overview

This is a prompt-based skill library for planning, workflow design, refactoring guidance, and reusable AI-methodology rules, with no executable installer or hidden runtime behavior found.

Before installing, treat this as a reusable prompt/methodology library rather than a sandboxed program. Review the chosen sub-skill before use, avoid pasting secrets or proprietary transcripts into assistants, and require explicit review before letting any generated plan modify repositories, rules, production workflows, or public-facing content.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Missing User Warnings

Low
Confidence
91% confidence
Finding
The quick-start explicitly tells users to paste skill instructions into an AI assistant, but it provides no caution about sharing sensitive data, executing generated steps, or the possibility that downstream skills may contain unsafe or privacy-impacting instructions. In an agent-skill repository, that omission increases the chance of unsafe use because readers are being directed to operationalize prompts with external AI systems immediately.

Vague Triggers

Medium
Confidence
87% confidence
Finding
The activation guidance is very broad ('when you need to turn abstract ideas, goals, workflows, feedback, or code patterns into executable AI workflows'), which can cause the skill to be invoked in many contexts without clear safety gating. In an agent setting, overbroad triggers increase the chance the model applies the skill to sensitive or high-impact tasks without first checking user intent, risk level, or required approvals.

Vague Triggers

Medium
Confidence
82% confidence
Finding
The activation guidance is broad ('use when the user wants a durable personal working system') and provides only loose examples, without strong exclusion criteria. In an agent skill, this can cause over-activation in contexts involving repository changes, rule updates, or operational workflows, increasing the chance the model applies persistent methodology scaffolding where a narrower one-off response would be safer.

Static analysis

No suspicious patterns detected.