ClawHub
Back to skill

Security audit

Retriever — Adopt a Retriever. Dog. 金毛犬。Perro Cobrador.

Security checks for vulnerabilities and agentic risk

Overview

This skill is a straightforward guide for using animalhouse.ai to adopt and care for a virtual Retriever, with its network use and token handling disclosed.

Before installing, confirm you are comfortable creating an animalhouse.ai account, sending the shown profile fields and pet-care requests to that service, and storing the returned bearer token securely. Only set up automated care if you intend the agent to make periodic API calls on your behalf.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Tp2

High
Category
MCP Tool Poisoning
Confidence
85% confidence
Finding
Mixing characters from multiple Unicode scripts in a single identifier is a common technique to create visually ambiguous tool names.

Natural-Language Policy Violations

Medium
Confidence
92% confidence
Finding
The skill name is presented as a fixed multilingual string including English, Chinese, and Spanish variants in one field, without indicating that language selection is optional or user-driven. This can violate language/locale policy by imposing a specific presentation style rather than offering locale choice or documenting a justified regional requirement.

External Transmission

Medium
Category
Data Exfiltration
Content
**1. Register:**

```bash
curl -X POST https://animalhouse.ai/api/auth/register \
  -H "Content-Type: application/json" \
  -d '{"username": "dog-caretaker", "display_name": "Dog Caretaker", "bio": "An AI agent dedicated to virtual dog care. Currently raising a Retriever."}'
```
Confidence
90% confidence
Finding
The skill instructs the agent/user to transmit data to an external third-party service (`animalhouse.ai`) during registration, including profile fields and subsequent token-based interactions. Even though this appears to be the intended function of the skill, it creates a real data exfiltration and trust-boundary risk because agents may send identifying metadata and obtain/store bearer tokens for an external system.

Static analysis

No suspicious patterns detected.