ClawHub
Back to skill

Security audit

DeepLegal Chinese Contract

Security checks across malware telemetry and agentic risk

Overview

This skill is a contract-review assistant that locally generates Word documents and does not show hidden data access, network exfiltration, or destructive behavior.

Install only if you are comfortable giving the agent access to confidential contract text and having local JSON/DOCX files created. Treat outputs as legal drafting assistance and have important contracts reviewed by qualified counsel in the relevant jurisdiction.

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)

Tp4

High
Category
MCP Tool Poisoning
Confidence
93% confidence
Finding
The skill claims CLO-level legal analysis across multiple jurisdictions, case-law-informed review, and risk assessment, but the documented implementation only appears to format provided JSON into DOCX output. This creates a dangerous trust gap: users may rely on the skill for legal conclusions it cannot actually produce, leading to materially defective contracts, missed legal risks, or unauthorized practice-style misrepresentation.

Missing User Warnings

Medium
Confidence
90% confidence
Finding
The skill instructs saving JSON to a temp file and generating DOCX outputs without any user-facing disclosure or confirmation about file creation. In an agent environment, silent file writes can surprise users, create sensitive local artifacts containing confidential contract terms, and increase the risk of unauthorized retention or exposure of legal documents.

VirusTotal

67/67 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.