ClawHub
Back to skill

Security audit

x0x

Security checks across malware telemetry and agentic risk

Overview

This skill is a broad peer-to-peer agent networking tool with powerful but clearly documented controls such as remote exec, port forwarding, persistent keys, and self-update.

Install this only if you want a persistent peer-to-peer agent networking daemon. Keep the API token private, trust and pin peers carefully, leave remote exec disabled unless you have a narrow allowlist, review any installer/update before applying it, and use autostart only when you want the daemon to run after login or boot.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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 (3)

Context-Inappropriate Capability

Medium
Confidence
92% confidence
Finding
The skill documents a remote execution feature (`/exec/run`) that allows commands to be run on another agent's machine. Even though the text says it is disabled by default and gated by trust and ACLs, remote exec is a high-risk capability that materially exceeds simple messaging/networking and can lead to arbitrary code execution on peers if misconfigured, over-trusted, or exposed through downstream automation.

Context-Inappropriate Capability

Low
Confidence
88% confidence
Finding
The self-update capability can download, verify, and install new binaries, which modifies local executables and creates a software supply-chain and integrity risk. Even if release verification exists, an update/install primitive is security-sensitive and not strictly necessary for the core networking function described by the skill.

Missing User Warnings

Medium
Confidence
85% confidence
Finding
The self-update section documents applying updates via API/CLI but does not foreground that this action will change local installed binaries on disk. In agent contexts, insufficient warning around self-modifying installation behavior can cause unintended system changes and make socially engineered or accidental update triggers more dangerous.

VirusTotal

53/53 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.