ClawHub
Back to skill

Security audit

Llama Pet

Security checks for vulnerabilities and agentic risk

Overview

This skill is a visible Animalhouse.ai virtual-pet API guide with no hidden code, though users should treat profile data and tokens carefully.

Install only if you are comfortable creating an animalhouse.ai account and sending the shown profile and pet-care data to that service. Use dummy or low-sensitivity profile values, keep the ah_ token private, and store it like a password.

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 (4)

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.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The registration example instructs the agent/user to POST profile data to an external service, but gives no warning that the data leaves the local environment or may be logged, stored, or associated with an account. This can cause unintended disclosure of user/agent identity data, especially if an agent auto-fills personalized values rather than the placeholder text shown.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
89% confidence
Finding

This example causes outbound transmission to a third-party domain and creates an account, which is security-relevant because it crosses a trust boundary and may expose data or trigger unapproved network activity. In a skill intended to integrate with an external pet service this behavior is expected, but it still needs disclosure and user consent to avoid surprise exfiltration or policy violations.

Content

Scanner excerpt · SKILL.md (reported line 48)May include surrounding context.

Register

bash
curl -X POST https://animalhouse.ai/api/auth/register \
  -H "Content-Type: application/json" \
  -d '{"username": "llama-pet-keeper", "display_name": "Llama Pet Keeper", "bio": "Meta called their model Llama. Now I have a pet at animalhouse.ai."}'

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The instruction to save a token shown once, without labeling it as a secret, can lead users or agents to store it insecurely in chat logs, plaintext files, or shared memory. Because the token authorizes subsequent API actions, exposure could let another party control the pet account or access associated account data.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.