ClawHub
Back to skill

Security audit

water-heater-tuner

Security checks for vulnerabilities and agentic risk

Overview

This is a local water-heater decision-support skill with disclosed profile storage and no hidden network, credential, or system-persistence behavior.

Before installing, be aware that running the profile command saves household composition, tank details, setpoint, location, and optional bill data in a local heater_profile.json file. The skill gives safety-critical water-temperature advice, so verify temperatures at fixtures and use a licensed professional for gas work, valves, plumbing changes, or vulnerable-household concerns.

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
  • System Prompt LeakageDirect Leakage, Indirect Extraction, Tool-Based Exfiltration
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • 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
Findings (4)

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The documented behavior does not fully match the actual capabilities: it persists household profile data locally without clearly declaring that storage behavior, and it omits/overstates certain functional areas. In an agent setting, this is dangerous because users and orchestrators rely on the manifest/description to make trust and permission decisions; hidden persistence can expose household composition and appliance details beyond expected processing.

Content

No source excerpt is available for this finding.

Direct Prompt Extraction

High
Category
System Prompt Leakage
Confidence
85% confidence
Finding

Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.

Content

Scanner excerpt · scripts/heater_tuner.py (reported line 224)May include surrounding context.

python
else:
        print("  Off at breaker/pilot. Freeze-risk zone: drain the tank.")
        print("  Return: full pasteurize protocol (see `pasteurize`).")
    print("\nReturn rule of thumb: healthy adults flush 5 min hot before "
          "showering; kids/elders/immunocompromised — full protocol.")

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
91% confidence
Finding

The skill advertises file persistence (heater_profile.json) and likely requires file read/write, but it declares no explicit tool scope or permissions. That creates an authorization and transparency gap: an agent may access local storage without the user or platform being clearly informed, increasing the chance of unintended data retention or broader file access than necessary.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill instructs users to heat water to 60°C+ and flush every tap/shower for 5 minutes, but it does not prominently warn that this procedure creates an immediate scald hazard at fixtures. In the context of a household safety skill, omission of an explicit burn warning is especially risky because the guidance may be followed by non-expert users, including in homes with children, elders, or other vulnerable occupants.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.