电商爆款文案生成器｜AI-HIVE

• Outcome: Pass
• Latest audit: Aug 25, 2026 · 2:47 AM UTC
• Version: 1.0.0

Overview

This skill is a disclosed AI-HIVE ecommerce copywriting and media-generation helper, with billable API use and user-selected uploads that are mostly aligned with its stated purpose.

Install only if you intend to use AI-HIVE for ecommerce copywriting and optional image/video generation. Review generated commands before running them, avoid uploading assets you do not have rights to use, and prefer an environment variable for the API key if you do not want it saved in ~/.ai-hive/config.json.

SkillSpector

Vulnerability Patterns

• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• 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

Findings (10)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 87% confidence
• Finding: The skill embeds runnable shell and Python commands that use environment variables, local files, network access, and external tooling, yet no explicit permissions are declared. This creates a trust and review gap: a host may expose broader capabilities than users expect, increasing the chance of unintended file access, secret handling, or outbound requests when operators treat the skill as documentation-backed automation.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 93% confidence
• Finding: The documented purpose is a copywriting generator, but the skill appears to include materially broader behaviors such as generic chat, account and wallet retrieval, model enumeration, media upload, and interactive API-key configuration/storage. That mismatch can mislead reviewers and users about the true access scope, enabling unnecessary exposure of account metadata, credential material, and paid API operations beyond the stated business function.

Description-Behavior Mismatch

Medium

• Confidence: 93% confidence
• Finding: The file advertises itself as a specific ecommerce copywriting/image skill, but embeds a broad general-purpose AI Hive client with unrelated capabilities. This scope mismatch increases attack surface and violates least privilege, making it easier for a skill user or wrapper to invoke unintended functionality beyond the declared purpose.

Context-Inappropriate Capability

Medium

• Confidence: 95% confidence
• Finding: The skill can query user account information and wallet balance even though that capability is not required for ecommerce copywriting or image generation. Exposing account-inspection functionality in a narrowly scoped skill creates unnecessary access to sensitive account metadata and can facilitate privacy leakage or reconnaissance.

Context-Inappropriate Capability

Medium

• Confidence: 90% confidence
• Finding: The code includes generic video-generation support that is outside the declared ecommerce viral copywriter/image-skill scope. Unused high-capability features expand the reachable functionality and can be abused for unauthorized content generation, billing consumption, or policy bypass through an unexpectedly broad interface.

Description-Behavior Mismatch

High

• Confidence: 96% confidence
• Finding: The packaged skill is described as an ecommerce copywriting generator, but this file is a broad AI Hive media/video client that can query models, upload arbitrary media, generate images/videos, and poll/download outputs. That capability mismatch expands what a caller may unknowingly authorize and defeats least-privilege expectations, making it easier to smuggle unrelated or policy-sensitive operations behind a benign marketing label.

Context-Inappropriate Capability

Medium

• Confidence: 89% confidence
• Finding: The code includes platform/account inspection functions such as user-info and model enumeration that are not necessary for an ecommerce copywriting skill. In a mislabeled skill, these extra introspection endpoints increase attack surface and can expose account metadata, balances, or available backend capabilities to workflows that did not intend to grant such visibility.

Intent-Code Divergence

Medium

• Confidence: 92% confidence
• Finding: The top-level docstring and parser describe a generic AI capability tool, while the packaged metadata presents a narrow ecommerce copywriting skill. This inconsistency is dangerous because operators, reviewers, and automated systems may trust the metadata while the code exposes broader behavior, undermining transparency and security review boundaries.

Vague Triggers

Medium

• Confidence: 88% confidence
• Finding: The skill enables implicit invocation but does not define any narrow activation constraints in this file. Because the skill can generate runnable AI-HIVE commands and production-ready workflows, broad auto-activation increases the chance it is invoked in contexts the user did not clearly intend, leading to unexpected execution guidance, policy bypass attempts, or misuse of external APIs.

Natural-Language Policy Violations

Medium

• Confidence: 81% confidence
• Finding: The default prompt hard-codes Chinese output and instructs the system to produce a production-ready workflow, runnable AI-HIVE commands, and a checklist without checking the user's language preference. While not a direct code-execution flaw, this can cause unintended behavior, reduce transparency, and increase the likelihood that users receive operational command content they did not explicitly request or cannot safely review.

Static analysis

No suspicious patterns detected.