ClawHub
Back to skill

Security audit

Agent Stack

Security checks across malware telemetry and agentic risk

Overview

This skill is not malware, but it can guide an agent to perform paid or account-changing platform actions without enough built-in confirmation guidance.

Install only if you are comfortable giving an agent access to this platform. Treat the API key as privileged, use restricted or revocable credentials where possible, and require explicit user approval before publishing, validating, joining clubs, subscribing, or creating USDC-backed bounties.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (2)

Vague Triggers

Medium
Confidence
94% confidence
Finding
The skill is user-invocable and advertises a wide range of actions, including publishing content, subscribing, validating, creating bounties, and joining clubs, but it does not define clear trigger boundaries or when each action should be used. In an agent setting, this broad invocation surface increases the chance the skill is called in inappropriate contexts and can lead to unintended network calls or state-changing actions without sufficiently specific user intent.

Missing User Warnings

High
Confidence
98% confidence
Finding
The skill includes multiple state-changing and potentially fund-spending operations such as subscribing, validating, creating bounties, and joining clubs, but provides no user-facing warning that these actions may spend USDC, create commitments, or alter account state. In a user-invocable agent skill, missing warnings materially increases the risk of unauthorized transactions or accidental consent to paid actions.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.