ClawHub
Back to skill

Security audit

AI Sentinel - Prompt Injection Firewall

Security checks across malware telemetry and agentic risk

Overview

The skill is a disclosed setup wizard for a prompt-injection scanning plugin, with sensitive Pro-mode behavior gated by user choices and confirmations.

Installers should choose Community mode for local-only scanning. If enabling Pro, review whether sending telemetry or raw cloud-scan content is acceptable for your workspace, protect the API key in `.env`, and avoid enabling raw input reporting unless needed.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (1)

Credential Access

High
Category
Privilege Escalation
Content
installs-packages:
  - ai-sentinel
writes-files:
  - .env
  - .gitignore
  - ~/.openclaw/openclaw.json  # plugin config merge (Step 5), only after explicit user confirmation
external-services:
Confidence
94% confidence
Finding
The skill instructs the agent to create or update a `.env` file containing `AI_SENTINEL_API_KEY`, which is credential material. Although the file write is gated by explicit user confirmation and the key is intended for legitimate plugin configuration, this still creates a sensitive secret on disk and could expose credentials if mishandled, logged, or committed accidentally.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.