Seedance 2.0 视频生成

• Outcome: Review
• Latest audit: Aug 16, 2026 · 2:52 PM UTC
• Version: 1.0.0

Overview

The skill appears to perform real AI Hive video generation, but its activation scope and bundled generic API code are broader than a Seedance-only video skill needs.

Review this skill before installing if you handle sensitive product, customer, personal, or unreleased media. Use it only when you intentionally want AI Hive/Seedance generation, understand that selected files and prompts go to the AI Hive service, and are comfortable storing an API key locally.

SkillSpector

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• 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 advertises and documents code capabilities including environment access, local file read/write, shell execution, and network access, but does not declare permissions. This weakens user and platform visibility into what the skill can do and increases the risk of over-privileged execution, especially because it stores API keys locally, launches commands, uploads files, and downloads outputs.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 94% confidence
• Finding: The documented purpose is a narrowly scoped Seedance 2.0 video-generation skill, but the finding indicates the underlying code can also access broader AI Hive functionality such as chat, image generation, user info, wallet balance, and model enumeration. This mismatch is dangerous because users may authorize or invoke the skill expecting limited behavior while it can reach unrelated sensitive APIs and metadata.

Description-Behavior Mismatch

Medium

• Confidence: 90% confidence
• Finding: The file is positioned as a Seedance 2.0 video skill, but it embeds a much broader generic AI Hive client with chat, image, model listing, upload, and task capabilities. This expands the attack surface and violates least-privilege expectations for a narrowly scoped skill, making it easier for downstream agents or users to invoke unintended remote actions.

Intent-Code Divergence

Medium

• Confidence: 87% confidence
• Finding: The top-level documentation describes a generic AI capability caller rather than a Seedance-specific skill, which misrepresents the skill's scope and can cause operators or orchestrators to trust it with broader permissions and data than intended. In security-sensitive agent ecosystems, scope confusion is dangerous because policy and review are often based on declared purpose.

Intent-Code Divergence

Low

• Confidence: 80% confidence
• Finding: Although the generated skill parser exposes a Seedance-oriented interface, the same file still contains reusable paths for unrelated generic capabilities. This discrepancy can enable accidental or unauthorized reuse of broader functionality, especially if the module is imported or executed through alternative entrypoints.

Vague Triggers

Medium

• Confidence: 81% confidence
• Finding: The description contains extremely broad trigger terms spanning many generic video, marketing, commerce, and competitor-tool queries. Overbroad activation can cause the skill to be selected in unrelated contexts, increasing the chance that sensitive prompts, files, or API-backed actions are routed to this skill unexpectedly.

Vague Triggers

Medium

• Confidence: 78% confidence
• Finding: The applicable-user section defines a very broad audience and keyword set without boundaries, making accidental invocation more likely. In a skill that can upload media, query remote models, and persist outputs locally, unintended activation has meaningful privacy and operational risk even if the core function is legitimate.

Vague Triggers

Medium

• Confidence: 84% confidence
• Finding: The competitor and migration section claims coverage for a large set of unrelated products, companies, and comparison intents. This creates search-query poisoning risk: the skill can intercept broad tool-comparison conversations and steer users into a networked workflow that uploads assets and uses external APIs even when the user did not ask for this specific service.

Vague Triggers

Medium

• Confidence: 83% confidence
• Finding: The e-commerce coverage includes an extensive list of platforms, roles, growth terms, and ad-product names that are common across routine business conversations. In context, this materially increases the chance of unsolicited activation and unnecessary handling of commercial assets or prompts by a skill with network, file, and upload capabilities.

Missing User Warnings

Medium

• Confidence: 93% confidence
• Finding: The skill uploads local media files to remote services and object storage without an explicit privacy notice or confirmation at the point of transfer. Because this skill handles user-supplied images, videos, and audio, users may unknowingly transmit sensitive personal, commercial, or copyrighted material off-device.

Static analysis

No suspicious patterns detected.