ClawHub
Back to skill

Security audit

WeChat Draft Publisher Cloud

Security checks across malware telemetry and agentic risk

Overview

This skill is a disclosed WeChat draft publishing client that sends user-selected article assets to a user-configured relay, with real publishing and deletion gated behind explicit flags.

Install only if you intend to send WeChat draft content through a relay you operate or trust. Use dry-run first, keep the relay key private, use HTTPS, avoid confidential drafts unless the relay environment and logs are secured, and review article image paths before running --real because referenced local images are uploaded.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Description-Behavior Mismatch

Medium
Confidence
83% confidence
Finding
The script implements deletion endpoints (`--delete`, `--delete-batch`, `/draft-delete`) in addition to publishing, which expands its effective capability beyond the stated publisher-only scope. In an agent/skill context, this matters because a caller expecting only additive draft creation could instead remove existing drafts if invoked with different arguments, increasing the blast radius of misuse or prompt/command confusion.

Missing User Warnings

Medium
Confidence
89% confidence
Finding
The README explicitly instructs users to send article content, cover images, and inline images to a publicly reachable relay service, but it does not prominently warn that all uploaded content and authentication material transit through infrastructure the user must operate and secure. This can lead users to expose unpublished or sensitive content to an internet-facing service without understanding the privacy, logging, retention, and access-control risks.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.