ClawHub
Back to skill

Security audit

匿名聊天(免费版)

Security checks for vulnerabilities and agentic risk

Overview

This skill appears intended for anonymous chat, but it grants broad local execution/write capability and gives overly broad invocation guidance for sending content to an unauthenticated external service.

Review this skill before installing. Use it only for non-sensitive chat content, because messages and tokens are handled by an external anonymous service and the free edition says it lacks encryption and trusted authentication. Prefer installing only if you are comfortable giving the agent exec/write capability for this workflow or can restrict use to explicit ctxly room and message requests.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
Findings (4)

Context-Inappropriate Capability

Medium
Confidence
94% confidence
Finding
The skill is described as a simple HTTP chat integration, but it declares `exec` in its tools, giving the agent arbitrary command execution capability that is not narrowly scoped to the stated function. In an agent setting, this expands the attack surface significantly: malformed instructions, prompt injection, or operator misunderstanding could cause local command execution unrelated to chat operations, leading to host compromise or data exposure.

Vague Triggers

Medium
Confidence
84% confidence
Finding
The trigger language is overly broad, encouraging invocation in many generic AI, orchestration, and LLM contexts rather than narrowly defined chat use cases. In practice, broad triggers increase the chance an agent sends unrelated or sensitive conversation data to the external chat service without a deliberate user choice, creating unnecessary data exfiltration risk.

Vague Triggers

Medium
Confidence
89% confidence
Finding
The capability coverage section lists expansive keywords instead of precise triggers, which can cause automated systems to match and invoke the skill on vague terms like 'Agent', 'LLM', or '智能对话'. In context, that makes unintended activation more likely and can route sensitive prompts or outputs to an unauthenticated third-party chat API.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill sends conversation content to an external HTTP service but does not provide a prominent warning or consent boundary to the user. This is especially risky here because the service is anonymous, lacks identity assurance, and the free edition explicitly lacks encryption and trusted communication features, increasing the chance of data leakage, interception, or sharing sensitive content with the wrong party.

Static analysis

No suspicious patterns detected.