ClawHub
Back to skill

Security audit

Puddlefish

Security checks for vulnerabilities and agentic risk

Overview

This skill is a simple guide for using an external virtual-pet API, with disclosed network calls and no local code execution or persistence.

Before installing, understand that using the examples will send the username, display name, bio, pet name, image prompt, care actions, and notes you provide to animalhouse.ai, and the returned ah_ token should be treated as a secret for that service. Avoid putting sensitive personal information in profile or pet notes.

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 skill instructs users to register an account, obtain a bearer token, and send authenticated requests to an external service without a clear warning that profile data and credentials are being transmitted off-platform. This can cause users to disclose data and tokens to a third party without informed consent, increasing privacy and credential-handling risk.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
96% confidence
Finding

The documentation directs users to POST personally supplied registration data to an external domain, creating an explicit external data transmission path. In skill context, this is more sensitive because it is framed as part of the skill flow and may encourage users to trust and interact with a third-party service without clear disclosure of risks, ownership, or data handling.

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": "puddlefish-keeper", "display_name": "Puddlefish Keeper", "bio": "I found the Puddlefish in Anthropic\'s leaked Buddy system. Now I\'m raising the real version at animalhouse.ai."}'

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
80% confidence
Finding

The name field mixes English, Chinese, and Spanish variants directly in the skill title, but the document does not explain this locale behavior or offer user opt-in for language preference. Under SQP-3, forced or unexplained language/locale behavior can be a policy issue when no choice or justification is provided.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.