ClawHub
Back to skill

Security audit

Switch Modes

Security checks across malware telemetry and agentic risk

Overview

The skill appears to do what it claims, but it can persistently change OpenClaw’s default model from ambiguous short words, which may affect other sessions and costs.

Review before installing if you do not want a skill that can edit ~/.openclaw/openclaw.json and change the default model beyond one chat. Prefer explicit phrases like "eco mode" or "/modes status", check the active mode after switching, and be careful with SMART or MAX because a persistent model change can affect cost and 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 (3)

Vague Triggers

Medium
Confidence
78% confidence
Finding
The reference says the agent scans for mode keywords but does not define exact trigger phrases or boundaries. In a skill that changes a global model setting, ambiguous activation can cause unintended switches from ordinary conversation text, leading to surprising behavior, cost changes, or degraded task quality across sessions.

Missing User Warnings

Medium
Confidence
86% confidence
Finding
The documentation notes that the skill modifies `~/.openclaw/openclaw.json`, but the cross-session and global impact is not emphasized early where users decide to use it. Because the change affects all OpenClaw sessions, a user may unknowingly alter other workflows, causing unexpected model usage, higher costs, or reduced security review quality.

Vague Triggers

Medium
Confidence
95% confidence
Finding
The skill uses very broad standalone trigger words like "eco", "smart", "balanced", and "max", which are common in normal conversation and can cause accidental activation. Because this skill modifies persistent local configuration files and changes the active model, unintended triggering can silently alter agent behavior, cost, or capability without the user clearly intending a mode switch.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.