Model Route Guard

Security checks across malware telemetry and agentic risk

Overview

This skill is a routing repair guide, but its fix can permanently change OpenClaw model routing and delete an agent override without enough safeguards.

Review before installing. Use the audit portion freely, but before running the fix, back up both OpenClaw config files, confirm that coding.dashscope.aliyuncs.com is the right endpoint for your account and region, and manually review whether deleting the agent models.json would remove intentional custom routing.

SkillSpector

By NVIDIA

Vulnerability Patterns

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
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (2)

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The skill instructs the user to overwrite the global configuration file and forcibly delete an agent-specific models.json override without any explicit warning, backup step, or confirmation gate. This is dangerous because it can irreversibly remove valid user customization, break unrelated agents, and cause configuration loss in a troubleshooting workflow where users may copy-paste commands directly.

Natural-Language Policy Violations

Medium

Confidence: 87% confidence
Finding: The skill hard-codes a single Bailian provider endpoint and labels it as the correct one without validating the user's region, deployment, or provider requirements. In a routing-repair skill, this can misroute traffic, break working configurations, or direct requests to an unintended service endpoint, especially when the user's existing setup was intentionally customized.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal