ClawHub
Back to skill

Security audit

fluid-network-analysis

Security checks for vulnerabilities and agentic risk

Overview

This skill locally validates and analyzes fluid-network models as described, without hidden network access, credential handling, persistence, or unrelated authority.

Installers should be aware that the skill runs a bundled Python script on model files they provide. That is proportionate to the analysis purpose; review model assumptions and solver output before relying on engineering results.

SkillSpector

Vulnerability Patterns
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • 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)

Lp3

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding
Without declared permissions the skill's intent is opaque and cannot be validated.

Static analysis

No suspicious patterns detected.