ClawHub
Back to skill

Security audit

Z.ai 图片理解

Security checks for vulnerabilities and agentic risk

Overview

This skill transparently uses Z.ai's remote vision API to analyze user-provided images, with no hidden persistence or unrelated behavior found.

Install this only if you are comfortable sending selected images, prompts, and resulting analysis to Z.ai/Open BigModel. Avoid using it on IDs, medical records, confidential screenshots, or private documents unless you have permission and understand the provider's data handling terms; avoid saving full raw responses or reasoning output unless needed.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (7)

Lp3

Medium
Category
MCP Least Privilege
Confidence
95% confidence
Finding
The skill describes capabilities that access environment variables, local files, write output files, and make network requests, but it does not declare permissions or present those capabilities transparently. This increases the risk of over-privileged or unexpected behavior, especially because local image paths and API-key-based outbound requests are supported.

Missing User Warnings

Medium
Confidence
97% confidence
Finding
The skill routes user images and prompts to an external third-party API, but the description does not clearly warn users that their data leaves the local environment. This can cause unintended disclosure of sensitive visual content, OCR text, documents, or metadata to an external service without informed consent.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The skill is explicitly designed to transmit user-supplied images and prompts to a third-party API, but the documentation does not warn about privacy, consent, retention, or sensitive-data handling. In a vision-analysis skill, images may contain faces, IDs, medical documents, screens, or other sensitive content, so silent external sharing creates real privacy and compliance risk.

Missing User Warnings

Medium
Confidence
87% confidence
Finding
The script supports reading a local image file, base64-encoding it, and transmitting both the image contents and user prompt to a third-party remote API, but the CLI flow does not clearly warn or require explicit consent at the point of transmission. In an agent-skill context, users may assume local analysis, so sensitive screenshots, documents, or photos could be exfiltrated off-host unexpectedly.

Ssd 3

Medium
Confidence
98% confidence
Finding
The documented return format includes `reasoning_content`, i.e., the model's full chain-of-thought or internal reasoning. Returning this can expose sensitive details derived from the image or user prompt, leak unnecessary internal processing, and propagate confidential information beyond the intended answer surface.

External Transmission

Medium
Category
Data Exfiltration
Content
"Content-Type": "application/json"
}

response = requests.post(url, json=payload, headers=headers, timeout=60)
result = response.json()

if response.status_code == 200:
Confidence
97% confidence
Finding
This example performs an outbound POST request carrying user-provided image references and prompt text to an external service. In the context of an image-understanding skill, external transmission is core functionality, but it is still security-relevant because it can expose sensitive user data to a third party if used without clear disclosure, minimization, and validation controls.

External Transmission

Medium
Category
Data Exfiltration
Content
tool_choice: "auto"
};

const response = await fetch("https://open.bigmodel.cn/api/paas/v4/chat/completions", {
  method: "POST",
  headers: {
    "Authorization": `Bearer ${apiKey}`,
Confidence
97% confidence
Finding
The Node.js example likewise sends image and prompt content to an external API endpoint. While expected for this skill, the combination of third-party transmission and potentially sensitive image contents makes this a real privacy/security concern unless users are clearly informed and implementation safeguards are present.

Static analysis

No suspicious patterns detected.