ClawHub
Back to skill

Security audit

Seedream 5.0 Lite 图片编辑

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent AI image-editing helper that uses a fixed external service and local API key setup with disclosed, purpose-aligned behavior.

Install only if you are comfortable sending the selected images and prompts to AI Hive and storing an AI Hive API key locally. Use images you are authorized to edit, and review generated outputs carefully before using them for people, news, brands, evidence, or product claims.

SkillSpector

Vulnerability Patterns
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (2)

Lp3

Medium
Category
MCP Least Privilege
Confidence
86% confidence
Finding
The skill advertises no declared permissions, yet the documented commands and described helper behavior imply shell execution, filesystem reads/writes, environment access, and network communication. This creates a transparency and consent gap: a caller may invoke the skill expecting a narrow image-edit capability while it can also persist credentials locally and contact external services. In a skill ecosystem, undeclared capabilities materially increase risk because users and policy layers cannot accurately gate or review what the skill can do.

Tp4

High
Category
MCP Tool Poisoning
Confidence
92% confidence
Finding
The documented behavior exceeds the stated purpose: beyond constrained image editing, it supports task lookup by arbitrary task ID, interactive API key setup with browser opening and credential storage, media upload primitives, and general prompt-based generation. That mismatch is dangerous because reviewers and users may authorize the skill for a narrowly scoped editing workflow while it actually enables broader data handling and persistent side effects. The risk is higher here because the skill processes user-supplied images and stores credentials in the user's home directory, increasing confidentiality and misuse concerns.

Static analysis

No suspicious patterns detected.