ClawHub
Back to skill

Security audit

公众号内容提取

Security checks across malware telemetry and agentic risk

Overview

This skill appears intended to fetch WeChat articles, but it sends credentials and article URLs to external services without enough clear user-facing disclosure.

Review before installing if the公众号名称, date range, article URLs, and AppID/SecureKey are sensitive. Prefer environment variables over command-line secrets, use limited or revocable credentials, and assume article URLs may be visible to both wxpub.aibana.art and anything-md.doocs.org during use.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (6)

subprocess module call

Medium
Category
Dangerous Code Execution
Content
def fetch_and_save(url, output_dir):
    try:
        r = subprocess.run(
            [
                "curl", "-s", "--max-time", "60",
                "-X", "POST", "https://anything-md.doocs.org/",
Confidence
93% confidence
Finding
r = subprocess.run( [ "curl", "-s", "--max-time", "60", "-X", "POST", "https://anything-md.doocs.org/", "-H", "Content-Type: applica

Description-Behavior Mismatch

Medium
Confidence
97% confidence
Finding
The implementation contradicts the skill description by forwarding article URLs to an external conversion service instead of only fetching and saving content locally. This is dangerous because it silently expands data exposure and gives a third party visibility into what the user is retrieving, which may include sensitive research interests or private/internal URLs if the tool is misused.

Missing User Warnings

Medium
Confidence
84% confidence
Finding
The README encourages passing AppID and SecretKey on the command line, which can expose credentials through shell history, process listings, logs, or terminal recordings. While environment variables are somewhat better, the documentation omits credential-handling warnings and safer practices, increasing the chance of accidental secret disclosure.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill sends user-supplied公众号名称, date range, and especially AppID/SecureKey to an external service (`wxpub.aibana.art`) and local scripts that likely call that service, but it does not clearly warn the user that these inputs and credentials will be transmitted off-platform. This creates privacy and credential-handling risk, particularly because users may not expect secrets passed on the command line to be exposed to subprocess listings, logs, or third-party processing.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The code downloads remote content and also sends article URLs to an external service without any explicit warning, consent flow, or domain restrictions. In an agent skill context, this is more dangerous because users may assume the agent performs local retrieval only, while the implementation causes network egress to multiple third parties and could be abused to fetch unexpected or sensitive resources.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The script accepts an app_id and secure_key, packages them into a JSON request body, and sends them to a third-party remote service. Although the transport uses HTTPS, the skill gives no user-facing warning that credentials will be disclosed to an external domain, which is a real security and privacy issue because users may assume secrets are only used locally or with an expected first-party service. In this skill context, the risk is elevated because the skill is explicitly designed to fetch WeChat public-account data and users may provide live API credentials to accomplish that task.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.