ClawHub
Back to skill

Security audit

Gen Design

Security checks across malware telemetry and agentic risk

Overview

This skill is a documented system-design generator that writes a design document and may offer a related validation step, with no evidence of hidden code, credential access, exfiltration, or destructive behavior.

Before installing, know that the skill will create a design document under docs/design and may proceed to a validation step unless you clearly tell it to pause. Use it in repositories where creating documentation files is expected, and explicitly opt out of validation if you only want the design document generated.

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 (3)

Description-Behavior Mismatch

Medium
Confidence
89% confidence
Finding
The skill is scoped as a design-generation tool, but it also instructs the agent to automatically invoke a second skill (`validate`) when the user replies ambiguously or does not explicitly pause. This creates a cross-skill action without clear, fresh user consent, which can violate least-astonishment and enable unintended execution chains in agentic environments.

Description-Behavior Mismatch

Medium
Confidence
90% confidence
Finding
The skill declares a narrow boundary of only generating a system design document, but later instructs the agent to automatically continue into `/validate --mode=dataflow` when the user says "yes" or does not explicitly pause. This creates scope expansion and unintended tool chaining, which can cause actions beyond the user's original request and weaken user consent and routing guarantees.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The skill mandates creating directories and writing a generated design document to disk, including in empty repositories, without first warning the user or obtaining confirmation for filesystem modification. In an agent context, silent writes can alter project state, create unexpected artifacts, or be abused to persist unwanted content in the workspace.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.