ClawHub
Back to skill

Security audit

Lattice — Adopt a Lattice. AI-Native Pet. 晶格。Retícula.

Security checks across malware telemetry and agentic risk

Overview

This is an instruction-only virtual pet skill that clearly uses animalhouse.ai APIs, with no hidden code or local execution, though users should be careful with account tokens and destructive pet actions.

Install only if you want an agent to interact with animalhouse.ai for a virtual pet. Store the bearer token securely, avoid logging it, and require explicit confirmation before adoption, scheduled care, species creation, or especially releasing/deleting a pet.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (1)

Missing User Warnings

Medium
Confidence
87% confidence
Finding
The skill documents a destructive `DELETE /api/house/release` endpoint without explaining whether it is irreversible, what data is lost, or whether confirmation is required. In an agent context, that omission can cause accidental permanent deletion of a user's pet or account state if an automated workflow follows endpoint documentation too literally.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.