ClawHub
Back to skill

Security audit

Antibody Humanizer

Security checks across malware telemetry and agentic risk

Overview

This appears to be a local antibody-sequence analysis skill with no evidence of hidden data sharing, but its patent-analysis documentation is inconsistent and should be treated cautiously.

Install only if you are comfortable processing potentially proprietary antibody sequences in the current workspace. Do not rely on the patent-landscape example unless the publisher supplies and documents the missing patent-checking component, including what databases are contacted and what sequence data would be sent.

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 (1)

Intent-Code Divergence

Medium
Confidence
79% confidence
Finding
The documentation is internally inconsistent: it claims the skill does not check existing patent claims, yet it provides a workflow invoking `scripts/patent_check.py` against patent databases. This discrepancy can mislead users about network access, external data sharing, and legal/compliance-sensitive processing of proprietary antibody sequences, increasing the chance of unintended disclosure or unsafe operational assumptions.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.