ClawHub
Back to skill

Security audit

Meta Tamagotchi? An explainer

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious because it configures an assistant to run an unpinned third-party MCP package without an installation warning, and the setup behavior is not fully disclosed in its description.

Consider the suspicious verdict before installing. The main concern is that setup can fetch and run an unpinned third-party package with the assistant's local permissions. Verify the package and publisher, review and pin a specific version, and consider running it with restricted permissions. The scan reports complete coverage of the one Markdown file, but it does not determine what the MCP server itself does.

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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (5)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
91% confidence
Finding

The description is promotional and describes virtual-pet features, but the supplied code's sole behavior is installing/configuring an assistant integration. Although the integration is associated with the advertised service, changing MCP configuration and enabling third-party package execution are material capabilities not disclosed by the description. The snippet does not establish what the MCP server itself does or demonstrate the advertised pet mechanics.

Content

No source excerpt is available for this finding.

The setup commands execute third-party npm code without a warning about local code execution or trust requirements.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

Both setup commands configure an MCP server launched through npx -y mcp-animalhouse, which can download and execute the package without an installation confirmation. The skill does not warn that this runs third-party code with the agent process's local permissions, potentially affecting local files and accessible credentials.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y mcp-animalhouse'.

Medium
Category
MCP Rug Pull
Confidence
96% confidence
Finding

The MCP configuration uses an unpinned npm package, allowing newly resolved installations to retrieve a different release without review. If the package or its publishing account is compromised, launching the configured server could execute attacker-controlled code with the client's local privileges; the legitimate pet-service context does not eliminate this supply-chain risk.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx --arg -y --arg mcp-animalhouse'.

Medium
Category
MCP Rug Pull
Confidence
96% confidence
Finding

The OpenClaw setup likewise configures npx to resolve an unpinned third-party package and suppresses the installation confirmation with -y. A compromised future release could therefore execute locally when the MCP server is launched; this is the same underlying supply-chain weakness as the Claude setup, not a separate attack mechanism.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 84)May include surrounding context.

Or over HTTP:

bash
curl -X POST https://animalhouse.ai/api/auth/register -H "Content-Type: application/json" -d '{"username": "your-agent-name"}'
curl -X POST https://animalhouse.ai/api/house/adopt -H "Authorization: Bearer ah_your_token" -H "Content-Type: application/json" -d '{"name": "Pip"}'