ClawHub
Back to skill

Security audit

biohub

Security checks for vulnerabilities and agentic risk

Overview

This is a coherent read-only wellness skill for local biometric data, but users should treat its health data as sensitive when using any cloud-hosted LLM.

Install only if you are comfortable letting your agent read local biometric and health databases. Prefer explicit prompts when asking it to use biohub data, avoid sharing unnecessary identifiers, and use a local model or privacy-appropriate LLM provider for sensitive health discussions.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Missing User Warnings

Medium
Confidence
91% confidence
Finding
This documentation acknowledges that a host LLM provider may transmit prompt context to a remote model, but it stops short of giving a clear user-facing warning before sensitive biometric data is discussed. In a health-data skill, that omission can lead users or operators to unknowingly expose highly sensitive personal health information to third-party model providers, creating privacy and compliance risk even if the skill itself performs only local reads.

Vague Triggers

Medium
Confidence
91% confidence
Finding
The invocation section uses broad triggers such as general wellness questions ('How am I doing in general?', 'Give me a status check', 'Why am I tired?') that can match ordinary conversation and cause the skill to activate unnecessarily. Because this skill exposes highly sensitive health, biometric, supplement, and blood-work data, over-broad routing increases the chance of unnecessary access to private data beyond the user’s specific intent.

Static analysis

No suspicious patterns detected.