ClawHub
Back to skill

Security audit

多肤色多体型模特套图｜AI-HIVE

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed AI-HIVE image-generation workflow with expected API key setup, upload, polling, and download behavior for its e-commerce model-image purpose.

Before installing, understand that this skill can send selected images and prompts to AI-HIVE, may create a local API-key config file, and can download generated files locally. Use it only with media you are authorized to upload, confirm paid generation parameters before running, and avoid storing or sharing API keys in project files or logs.

SkillSpector

Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
88% confidence
Finding
The skill exposes operational capabilities including environment access, file read/write, network access, and shell execution without declaring permissions or scoping them in the manifest. This weakens reviewability and user consent, and if the bundled scripts are invoked, they could access local files, environment secrets, or perform external requests beyond what a user expects from the skill description.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding
The documented purpose is narrowly framed around diverse model catalog image generation, but the behavior reportedly includes broader account queries, model inspection, general chat, video generation, generic upload, browser-based API-key configuration, and local brief-file creation. This mismatch is dangerous because users and reviewers may authorize the skill for a limited imaging workflow while it actually enables broader data access, credential handling, and external interactions that expand the attack surface.

Natural-Language Policy Violations

Medium
Confidence
86% confidence
Finding
The default prompt hard-codes Chinese output and workflow style without checking the user's language preference or explicit consent. This can override user intent, cause prompt steering in downstream interactions, and reduce transparency about why the assistant is responding in a specific language or format.

Vague Triggers

Medium
Confidence
89% confidence
Finding
Enabling implicit invocation without tightly scoped trigger conditions can cause the skill to activate unexpectedly on loosely related requests. Because this skill is capable of producing executable workflows and AI-HIVE commands, accidental activation may steer user sessions, trigger external-action planning, or bypass clearer user confirmation boundaries.

Static analysis

No suspicious patterns detected.