Security audit

Seedance 图生视频工作室｜AI-HIVE

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed AI-HIVE video-generation helper that uploads user-selected media and may store an API key, with no evidence of hidden exfiltration or destructive behavior.

Install only if you are comfortable sending selected images, videos, or audio to AI-HIVE and potentially incurring generation costs after approval. Use authorized media, avoid pasting API keys into chats or logs, and review or remove ~/.ai-hive/config.json if you no longer want the local credential stored.

SkillSpector

Vulnerability Patterns

Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration

Findings (5)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 92% confidence
Finding: The skill advertises executable commands and local scripts that use environment variables, file I/O, networking, and shell access, yet no permissions are declared. This creates a transparency and consent gap: users or hosting frameworks may not realize the skill can access local files, send data to external services, or invoke tools like ffmpeg, increasing the chance of unintended data exposure or unsafe execution.

Tp4

High

Category: MCP Tool Poisoning
Confidence: 96% confidence
Finding: The documented purpose is a narrowly scoped Seedance image-to-video workflow, but the described behavior extends to generic AI-HIVE chat, image generation, model enumeration, account and wallet queries, media upload, and local credential setup/storage. This scope creep is dangerous because it expands access to sensitive account information and broader external actions beyond what a user would reasonably expect from the skill’s name and description, making misuse, surprise billing, and credential mishandling more likely.

Description-Behavior Mismatch

Medium

Confidence: 95% confidence
Finding: The file is presented as a narrowly scoped Seedance image-to-video skill, but it embeds a much broader generic AI Hive client with chat, image generation, model enumeration, account lookup, and upload features. This capability mismatch violates least privilege and expands what an invoking agent could do beyond the user's expected task, increasing the risk of unintended data access or policy bypass.

Context-Inappropriate Capability

Medium

Confidence: 93% confidence
Finding: The user-info endpoint retrieves account and wallet information that is unrelated to generating Seedance image-to-video outputs. In a skill context, exposing billing or account metadata creates unnecessary access to sensitive operational information and broadens the blast radius if the skill is misused.

Natural-Language Policy Violations

Medium

Confidence: 92% confidence
Finding: The default prompt hard-codes Chinese output regardless of the user's language preference, which can override user intent and reduce transparency about what the agent will produce. In a workflow that generates runnable commands and production content, this can cause user confusion, review errors, or unsafe acceptance of outputs the user cannot easily validate.

Static analysis

No suspicious patterns detected.