ClawHub
Back to skill

Security audit

Analytics Dashboard Free

Security checks across malware telemetry and agentic risk

Overview

This skill is a local dashboard guide that reads local mailbox, session, and task status files, with no artifact-backed evidence of hidden exfiltration or destructive behavior.

Use this only for dashboard data you are comfortable viewing in a local web UI. Keep the host set to 127.0.0.1 unless you deliberately need remote access; if you expose it on 0.0.0.0, use a strong token, HTTPS/reverse proxy controls, and network restrictions. Be aware the package does not include the referenced server script, so the agent may need to create or locate implementation code before it can run.

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 (3)

Intent-Code Divergence

Medium
Confidence
79% confidence
Finding
The skill claims data stays local and is not uploaded, yet it also defines a callback_url input for asynchronous completion notifications. If implemented, that creates an outbound communication path that could transmit metadata or results to a remote endpoint, violating user expectations and weakening the trust boundary around a supposedly local-only dashboard.

Description-Behavior Mismatch

Medium
Confidence
76% confidence
Finding
The manifest presents a lightweight local visualization tool, but the documented callback behavior introduces outbound network capability unrelated to the core purpose. That mismatch can cause the agent or user to authorize a skill under a narrower trust model than the behavior actually requires, enabling unintended data disclosure.

Missing User Warnings

Medium
Confidence
91% confidence
Finding
Advising users to bind the dashboard to 0.0.0.0 materially expands exposure from localhost to any reachable network interface, while the dashboard aggregates sensitive mailbox, session, and task data. Token-only protection is not a strong enough safeguard on its own, especially if tokens are leaked via URLs, logs, browser history, or localStorage, making unauthorized access significantly more likely.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.