ClawHub
Back to skill

Security audit

Word 光标插入

Security checks across malware telemetry and agentic risk

Overview

This skill automates inserting text or tables into the currently open Word document, with disclosed but privacy-relevant reading of nearby cursor context.

Install only if you are comfortable letting the skill read a small amount of text around the active Word cursor and print that context to the agent terminal. Before invoking it, place the cursor carefully, confirm the active document is the intended one, and avoid rerunning after a successful insert because duplicate prevention is limited.

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)

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding
The skill explicitly instructs reading bounded text around the active Word cursor and printing that context to stdout. Even if bounded, this can expose sensitive document contents such as report text, names, numbers, or confidential fragments to logs or downstream tooling outside Word. The description also overstates idempotency and safe behavior, which increases the chance of duplicate writes or unsafe operator trust in a document-editing workflow.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.