ClawHub
Back to skill

Security audit

本地 Ollama 文生图工具。使用 x/z-image-turbo 模型生成图片，分辨率 1024x1024，每张图约耗时 90 秒。当用户说"生成图片"、"文生图"、"画一张图"、或提交中文提示词生成图像时使用。

Security checks for vulnerabilities and agentic risk

Overview

This skill is a local Ollama text-to-image helper whose network use and file output match its stated purpose.

Install only if you intend to use a local Ollama image model. Do not include secrets in prompts, confirm your local Ollama service and x/z-image-turbo model are trusted, and expect generated images to be written to the images directory or the output path you provide.

A.I.G

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (7)

Lp3

Medium
Category
MCP Least Privilege
Confidence
89% confidence
Finding
The skill invokes a local Ollama HTTP service, which is a network-capable action, but the manifest does not declare any tool scope or permission boundary. This creates a mismatch between documented and actual capabilities, increasing the risk of unintended network access and reducing users' ability to evaluate what the skill can do.

Vague Triggers

Medium
Confidence
93% confidence
Finding
The trigger phrases include very generic requests such as '生成图片', '文生图', and '画一张图', which are common user intents and may cause the skill to activate in situations the user did not specifically intend for this tool. Overbroad activation can lead to unexpected execution, unnecessary local service calls, and unreviewed content generation requests.

Natural-Language Policy Violations

Medium
Confidence
96% confidence
Finding
The manifest description repeatedly specifies using Chinese prompts, including activation on submission of Chinese prompt text and command examples with "你的中文提示词". This imposes a language constraint in the skill's natural-language instructions without presenting it as optional or justified as a region-specific requirement.

Natural-Language Policy Violations

Medium
Confidence
97% confidence
Finding
The docstring explicitly requires the prompt to be in Chinese ("中文提示词"), and the CLI help at L72 repeats the same restriction. This is a natural-language policy concern because the skill imposes a specific language requirement without any visible opt-in, fallback, or justification that the tool is region-specific.

External Transmission

Medium
Category
Data Exfiltration
Content
start_time = time.perf_counter()

    response = requests.post(
        "http://localhost:11434/api/generate",
        json={
            "model": "x/z-image-turbo",
Confidence
70% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

External Transmission

Medium
Category
Data Exfiltration
Content
start_time = time.perf_counter()

    response = requests.post(
        "http://localhost:11434/api/generate",
        json={
            "model": "x/z-image-turbo",
Confidence
80% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Internal Network Request

Medium
Category
Server-Side Request Forgery
Content
start_time = time.perf_counter()

    response = requests.post(
        "http://localhost:11434/api/generate",
        json={
            "model": "x/z-image-turbo",
Confidence
70% confidence
Finding
Code issues a request to a loopback, link-local, or private-range host. This can reach internal services not meant to be exposed and is a common SSRF pivot.

Static analysis

No suspicious patterns detected.