ClawHub
Back to skill

Security audit

iaiops-clinical

Security checks across malware telemetry and agentic risk

Overview

The skill is mostly coherent for clinical facility monitoring, but it mixes read-only claims with high-impact building-control write capability in a hospital context.

Review this carefully before installing in any clinical or OT environment. The skill appears intended for safety monitoring and analysis, but its documentation should be corrected so users know whether any write capability is actually available and how it is technically blocked, approved, logged, and scoped away from life-safety systems.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Description-Behavior Mismatch

High
Confidence
96% confidence
Finding
The manifest explicitly says the edition is read-only, but the skill later exposes `bacnet_write_property`, a real write primitive for building-control points. In a clinical facility context, this mismatch can cause an agent or operator to trust the skill as non-invasive and then issue state-changing commands against HVAC or medical-gas-related systems, creating patient-safety and operational risk.

Intent-Code Divergence

Medium
Confidence
86% confidence
Finding
The documentation presents Modbus as read-only, but the version matrix claims support for Modbus write function codes (5/6/15/16). Even if write tools are not listed for normal use, this inconsistency can mislead users, downstream agents, or integrators about the actual safety boundary and may enable unsafe assumptions about non-mutating access in an OT environment.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.