ClawHub
Back to skill

Security audit

SaMD软件医疗器械专家

Security checks for vulnerabilities and agentic risk

Overview

This is a coherent SaMD regulatory guidance skill, with a bundled local audit helper but no evidence of hidden runtime access, exfiltration, persistence, or unsafe automatic behavior.

Installers should treat this as a regulatory reference skill, not legal or regulatory advice. The bundled audit script should only be run intentionally; it writes a local security_results.json in the selected skill directory but does not show network, credential, or persistence behavior.

SkillSpector

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 (1)

Tp4

High
Category
MCP Tool Poisoning
Confidence
81% confidence
Finding
A description-behavior mismatch is dangerous because users may invoke a medical-regulatory guidance skill without realizing it also inspects files, scans for sensitive content, validates links, and writes a security_results.json report. Hidden or undisclosed analysis and output generation can lead to unexpected processing of local content and metadata, undermining informed consent and trust even if the extra behavior is not overtly malicious.

Static analysis

No suspicious patterns detected.