ClawHub
Back to skill

Security audit

天猫 电商图片生成与编辑

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed AI Hive image-generation helper for Tmall ecommerce visuals, with normal API-key, upload, network, and download behavior for that purpose.

Install only if you are comfortable providing an AI Hive API key, uploading selected product or brand reference files to AI Hive, and storing the key locally under ~/.ai-hive/config.json. Avoid using --base-url unless you trust the destination, because the API key is sent to that service.

SkillSpector

Vulnerability Patterns
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Lp3

Medium
Category
MCP Least Privilege
Confidence
90% confidence
Finding
The skill invokes a Python script with shell execution, package installation, filesystem access, environment-variable usage, and network calls, but does not declare permissions. This creates a transparency and least-privilege problem: users and reviewers cannot accurately assess what the skill can access, and hidden capabilities can be abused if the underlying script is modified or compromised.

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding
The stated purpose is limited to Tmall ecommerce image generation/editing, but the referenced implementation apparently supports broader functions such as text chat, video generation, user-info and wallet queries, model enumeration, browser-based API-key setup, uploads, and task/result handling. This mismatch is dangerous because it expands the attack surface well beyond user expectations and can expose sensitive account data, credentials, or unrelated media through capabilities not justified by the skill's declared scope.

Description-Behavior Mismatch

Medium
Confidence
90% confidence
Finding
The advertised skill is narrowly framed as a Tmall image-generation/editing tool, but the exposed `task` and `upload` subcommands allow broader interaction with the backend than the declared purpose suggests. This expands the skill's effective capability surface, enabling arbitrary media staging and task inspection that could be used to access or manipulate non-skill-specific workflows if the surrounding platform assumes tighter scoping.

Static analysis

No suspicious patterns detected.