ClawHub
Back to skill

Security audit

Ron

Security checks across malware telemetry and agentic risk

Overview

Ron is a coherent review-only skill, but it automatically retains cross-session memory from broad reviews without clear user control or redaction rules.

Install only if you are comfortable with a reviewer skill using the agent's available read access and retaining short notes across sessions. Avoid using it on work involving secrets, customer data, production incidents, private financial plans, or confidential strategy unless you disable or tightly control the memory behavior.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Vague Triggers

High
Confidence
95% confidence
Finding
The trigger phrases are extremely broad, including common requests like 'review' and 'second opinion,' which can cause unintended invocation outside the user's actual intent. In an agent system, ambiguous activation can alter behavior, tool usage, and outputs unexpectedly, creating prompt-routing and authorization boundary problems.

Vague Triggers

Medium
Confidence
88% confidence
Finding
Allowing either the user or the agent to invoke Ron, without clear boundary conditions, creates ambiguous authority and increases the chance of self-invocation during normal operation. That can lead to unrequested behavior changes, excess tool access, or recursive review flows that the user did not authorize.

Missing User Warnings

Medium
Confidence
97% confidence
Finding
The skill mandates persistent writes to memory.md after every session without any user-facing notice or consent mechanism. This creates a clear risk of silently storing sensitive user, task, or operational details across sessions, violating user expectations and potentially exposing retained data later.

Ssd 3

Medium
Confidence
98% confidence
Finding
The persistent memory instructions explicitly require retaining prior-session observations in plain language, which can easily capture sensitive information from reviews such as incidents, financial assumptions, internal architecture, or user-provided data. Because this is cross-session retention, the skill context makes it more dangerous: a reviewer role sees broad, high-value information that should not be casually persisted.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.