Walmart 电商视频生成与编辑

• Outcome: Pass
• Latest audit: Aug 17, 2026 · 6:29 AM UTC
• Version: 1.0.0

Overview

The skill is a coherent AI Hive video-generation wrapper for Walmart product videos, with expected remote media processing and local API-key configuration but no hidden exfiltration or destructive behavior found.

Install only if you are comfortable sending the selected product images, videos, audio, and prompts to AI Hive for processing. Prefer environment-variable or per-command API keys if you do not want a credential stored in ~/.ai-hive/config.json, and avoid using confidential marketplace assets unless your AI Hive data-handling obligations permit it.

SkillSpector

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• 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

Findings (7)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 90% confidence
• Finding: The skill advertises only a Walmart video-generation workflow, but the runtime commands invoke a Python script with shell, file, environment, and network capabilities without any declared permissions or visible constraints in the skill manifest. This creates a trust gap: users and reviewers cannot accurately assess what the skill can access, increasing the risk of unintended file access, credential exposure, or outbound data transmission if the backing script is broader than the stated purpose.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 95% confidence
• Finding: The described purpose is narrowly scoped to Walmart product video creation, but the detected behavior includes general-purpose AI Hive functions such as chat, image generation, account/wallet retrieval, model catalog access, and configuration management. This mismatch is dangerous because it can conceal materially broader access and operations than a user expects, enabling unnecessary exposure of account data, abuse of paid resources, or invocation of unrelated model capabilities.

Description-Behavior Mismatch

Medium

• Confidence: 93% confidence
• Finding: The file is presented as a Walmart-specific video generation/editing skill, but the implementation is a broad AI Hive client supporting chat, model listing, image generation, uploads, and generic task handling. This scope expansion violates least-functionality and increases the attack surface, making it easier for a caller to use the skill for unintended remote AI operations unrelated to the declared purpose.

Context-Inappropriate Capability

Medium

• Confidence: 91% confidence
• Finding: The skill exposes a generic text chat endpoint even though the advertised purpose is Walmart product video generation/editing. An agent granted this skill could be induced to perform unrelated LLM interactions, increasing data exfiltration and prompt-abuse risk beyond the expected permission boundary.

Context-Inappropriate Capability

Medium

• Confidence: 89% confidence
• Finding: The generic image-generation endpoint is out of scope for a skill marketed as video generation/editing for Walmart content. This overbroad capability lets callers invoke additional paid remote AI operations and process user prompts/media in ways the user may not expect from the declared skill.

Missing User Warnings

Medium

• Confidence: 87% confidence
• Finding: The init flow persists the API key to a local config file without an explicit warning that credentials will be stored on disk. Although the file mode is tightened to 0600, silent long-term storage can surprise users and increases the risk of credential exposure through backups, endpoint compromise, or accidental file sharing.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The upload workflow sends arbitrary user-provided media to remote API infrastructure and object storage without a clear privacy or data-handling warning. In this skill context, users may upload proprietary product assets, internal ad creatives, or sensitive media, so silent remote transfer materially raises confidentiality and compliance risk.

VirusTotal

64/64 vendors flagged this skill as clean.

Static analysis

No suspicious patterns detected.