ClawHub
Back to skill

Security audit

What To Eat

Security checks for vulnerabilities and agentic risk

Overview

This is a simple food recommendation skill with broad but food-related activation wording and no code, persistence, credentials, or system access.

Installing this skill should be low risk. The main practical consideration is that it may activate for broad food-choice conversations, but it only provides meal recommendations and does not appear to access data, run code, or change the environment.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (2)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill metadata advertises very broad trigger phrases such as general food suggestions and 'needs help deciding on food', which can overlap with ordinary conversation and cause the skill to activate when the user did not explicitly intend to use it. Over-broad routing can hijack benign requests, suppress more appropriate skills, or increase prompt-surface exposure by invoking this skill in loosely related contexts.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The usage guidance says to act when the user asks '吃什么' but provides only a broad positive trigger and no limiting conditions, disambiguation rules, or exclusions. This ambiguity makes unintended activation more likely in common conversational contexts, which can lead to inappropriate skill selection and unreliable agent behavior.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.