京东 电商图片生成与编辑

• Outcome: Pass
• Latest audit: Aug 17, 2026 · 3:07 AM UTC
• Version: 1.0.0

Overview

This skill is a coherent AI Hive image-generation helper for JD ecommerce assets, with user-directed API key setup, image upload, task polling, and result download.

Install only if you are comfortable sending selected product/reference images and prompts to AI Hive and storing an AI Hive API key locally. Avoid passing private files as --image or --file, and review ~/.ai-hive/config.json if you later want to remove the saved key.

SkillSpector

Vulnerability Patterns

• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration

Findings (6)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 87% confidence
• Finding: The skill exposes operational capabilities including shell, network, environment access, and file read/write without declaring permissions or constraining their use. That creates a transparency and trust problem: users may invoke a seemingly simple image-generation skill while it can also install packages, access local files, and make external requests, increasing the chance of unintended data exposure or system-side effects.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 96% confidence
• Finding: The documented purpose is narrow—JD e-commerce image generation/editing—but the observed behavior includes broad additional functions such as chat/text calls, video generation, generic media upload, wallet/user-info queries, model enumeration, and interactive browser/API-key setup. This mismatch is dangerous because it can mislead users about what data is processed and what external services/actions are triggered, enabling overcollection, secret handling, or unintended account/API interactions outside the advertised scope.

Description-Behavior Mismatch

Medium

• Confidence: 92% confidence
• Finding: The file is presented as a JD ecommerce image-generation/editing skill, but it also includes a generic text chat client and related CLI flows. This unnecessary capability broadens the attack surface and can enable unintended data egress or off-scope use through the same trusted skill entrypoint.

Description-Behavior Mismatch

Medium

• Confidence: 95% confidence
• Finding: The implementation contains full video-generation support even though the declared skill scope is image generation/editing for JD ecommerce assets. This is dangerous because users and orchestrators may trust the skill's narrower description while the code can perform broader remote actions and consume additional resources.

Context-Inappropriate Capability

Medium

• Confidence: 90% confidence
• Finding: The skill can query user account information and wallet balance, which is unrelated to generating or editing ecommerce images. Off-scope access to account metadata increases privacy and billing exposure and violates least-privilege expectations for a narrowly described media skill.

Context-Inappropriate Capability

Low

• Confidence: 80% confidence
• Finding: Listing and searching all available models is broader than necessary for a fixed-purpose JD image skill that already hardcodes a model in the skill wrapper. While not immediately critical, it exposes extra platform capability and facilitates repurposing the tool beyond its declared function.

Static analysis

No suspicious patterns detected.