ClawHub
Back to skill

Security audit

AI内容创作工作站

Security checks for vulnerabilities and agentic risk

Overview

This content-creation skill is mostly coherent, but it describes automated public posting and broad command/file/API authority without clear approval gates.

Review this before installing if you might process private drafts, proprietary scripts, or unpublished campaign material. Only use it with API keys and platform accounts you are comfortable exposing to this workflow, and require manual confirmation before any generated content is posted publicly or sent to third-party model, image, video, or TTS services.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (5)

Description-Behavior Mismatch

Low
Confidence
77% confidence
Finding
The skill advertises generic file handling, API integration, and command execution beyond its stated content-creation purpose while also declaring powerful tools like exec and write. That broadening increases the chance the agent is induced to perform unnecessary filesystem or shell actions unrelated to the user's intent, expanding the attack surface for misuse.

Vague Triggers

Medium
Confidence
87% confidence
Finding
Broad trigger phrases such as generic references to 'content creation' or 'workflow' can cause the skill to activate on ordinary conversation without clear user intent. In a skill with exec, write, and external-integration capabilities, ambiguous activation can lead to unintended actions, data transmission, or tool use.

Missing User Warnings

Medium
Confidence
91% confidence
Finding
The skill exposes read, write, glob, grep, and exec capabilities but does not prominently warn users up front that it may modify files or run commands. Missing this disclosure is dangerous because users may invoke the skill for benign content help while unknowingly granting a workflow authority to alter local state or execute shell operations.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The documentation includes API-calling behavior and a requests.post example, but the skill description does not clearly warn that user content may be sent to external services. This creates a real risk of unintentional data exfiltration, especially in a workflow that handles content drafts, uploaded material, and possibly proprietary text.

External Transmission

Medium
Category
Data Exfiltration
Content
import requests
api_url = "http://localhost:8080/api/skills/plug-ai-content-creation-workstation"
headers = {"Authorization": "Bearer YOUR_API_KEY"}
response = requests.post(api_url, json={
    "input": "示例输入数据",
    "options": {"mode": "advanced"}
})
Confidence
90% confidence
Finding
The example explicitly shows sending input data and an authorization header to an HTTP API endpoint, which is an external transmission path. In context, this skill processes user-provided content and may handle sensitive drafts or business material, so undocumented or automatic transmission increases confidentiality risk even if the endpoint is localhost in the example.

Static analysis

No suspicious patterns detected.