ClawHub
Back to skill

Security audit

Docparser

Security checks across malware telemetry and agentic risk

Overview

The skill appears purpose-aligned, but its safety text understates that some import or reprocessing actions can change remote state.

Install only if you intend the agent to operate this connected import/indexing service. Before allowing URL imports, reparse, reintegrate, or enqueue-style actions, ask the agent to show the exact action and payload and confirm that you want the remote service state changed.

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
95% confidence
Finding
The safety guidance is inaccurate because some untagged actions such as reparse/reintegrate and URL-based import can trigger processing or enqueue work, which changes remote system state. This can mislead an agent into treating state-changing operations as safe reads and executing them without user confirmation, causing unintended imports, reprocessing, or downstream integrations.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.