System Inspection

• Outcome: Review
• Latest audit: May 29, 2026 at 9:36 PM UTC
• Version:
  1.0.0

Overview

This is mostly a system health-check SOP, but it has automatic operational side effects that users should review before installing.

Install only if you are comfortable with a scheduled ops skill checking OpenClaw logs/status, probing configured channels, potentially restarting the gateway during troubleshooting, and saving inspection summaries into both personal and shared memory. Prefer manual runs first, require human approval for restarts and test messages, and change the memory writes to sanitized, opt-in audit records in paths you control.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

Description-Behavior Mismatch

Medium

• Confidence: 95% confidence
• Finding: The skill is presented as a low-disturbance inspection SOP, but it mandates writing inspection results into both a personal and a shared memory store after every run. This creates persistent side effects unrelated to mere observation, can expose operational details to other agents or users, and increases the chance of unwanted data retention or cross-agent data pollution.

Context-Inappropriate Capability

Medium

• Confidence: 92% confidence
• Finding: The mandatory writes to `~/agents/dalong/workspace/memory.md` and `memory/YYYY-MM-DD.md` are not justified by the stated purpose of periodic system inspection. Because these writes affect personal and shared state, they can leak system status, normalize unnecessary persistence, and create a hidden modification channel in what users would reasonably expect to be a read-mostly healthcheck skill.

Description-Behavior Mismatch

Medium

• Confidence: 90% confidence
• Finding: The SOP claims to be low-disturbance, yet it instructs operators to actively probe channels by sending low-disturbance test messages. Even limited active messaging can notify real users, trigger automations, or create audit noise, so the documented behavior is more invasive than the description suggests.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The skill requires writing inspection results into personal and shared memory files without warning that it modifies user-controlled or shared data. Hidden writes are risky because they can overwrite prior context, introduce sensitive operational information into broadly accessible stores, and surprise users who expect a diagnostic routine to be non-destructive.

VirusTotal

65/65 vendors flagged this skill as clean.