ClawHub
Back to skill

Security audit

Ozon-爆款跟卖与选品大师 | 青虎AI

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed Ozon product-research helper that uses a Qinghu API with user-provided authorization and explicit confirmation before calls.

Before installing, understand that this skill can call Qinghu's authenticated API using a token you provide or an environment variable, and those calls may consume Qinghu points after confirmation. It does not appear to install code or alter your system, but users should only approve API calls they actually want run.

A.I.G

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Vague Triggers

Medium
Confidence
93% confidence
Finding
The skill description says Ozon product-analysis requests 'must trigger,' which is an overly broad activation rule that can cause the agent to invoke this skill for generic commerce or research queries without sufficient narrowing. That increases the chance of unnecessary external API use, unintended cost-incurring calls, and routing user requests into a specialized workflow that may not actually fit the user's intent.

Vague Triggers

Low
Confidence
85% confidence
Finding
The trigger examples are broad and lack exclusion criteria, so common requests like 'what is selling well' or 'can this item be followed' may match even when the user has not clearly requested Ozon-specific analysis. In this skill, broad examples are especially relevant because invocation can lead to authenticated third-party API calls and possible paid usage, making accidental activation more harmful than a purely local skill.

Static analysis

No suspicious patterns detected.