ClawHub
Back to skill

Security audit

Seedream 5 图片工作室｜AI-HIVE

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed AI-HIVE image-generation workflow with expected network, upload, download, and API-key use, but users should understand the local credential storage and possible billing before running it.

Install only if you intend to use AI-HIVE for Seedream 5 image work. Before running generation commands, confirm the prompt, uploaded files, routing mode, batch size, and expected cost. Treat ~/.ai-hive/config.json as a local secret store and remove or rotate the API key if the machine is shared or the key is no longer needed.

SkillSpector

Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
86% confidence
Finding
The skill includes operational capabilities to access environment variables, read/write files, use the network, and invoke shell commands, but it does not declare permissions or constrain those powers in the manifest. This creates a trust-boundary problem: users and reviewers cannot accurately assess what the skill may access or execute, increasing the risk of unintended data exposure, filesystem modification, or command execution if supporting scripts are invoked.

Tp4

High
Category
MCP Tool Poisoning
Confidence
91% confidence
Finding
The skill is presented as an image-studio workflow, but static analysis indicates additional behaviors including text/video model access, account and wallet retrieval, generic model enumeration, and interactive browser-based API key setup. That mismatch is dangerous because it can mislead operators about the true attack surface and enable access to sensitive account data or broader service actions beyond the user’s expected scope.

Vague Triggers

Medium
Confidence
93% confidence
Finding
The skill allows implicit invocation with no activation boundaries, so it can be selected automatically even when the user did not clearly request image-generation workflow automation or AI-HIVE operations. In this skill’s context, that increases the chance of unintended execution paths involving external model queries, uploads, pricing, polling, or downloads, which can cause privacy, cost, and policy-compliance issues.

Natural-Language Policy Violations

Medium
Confidence
84% confidence
Finding
The default prompt hard-codes Chinese output regardless of the user’s language preference, which can degrade transparency, cause misunderstandings, and interfere with informed review of generated commands or workflow steps. In a skill that may produce runnable AI-HIVE commands and operational instructions, language forcing can increase the risk that users approve actions they do not fully understand.

Static analysis

No suspicious patterns detected.