ClawHub
Back to skill

Security audit

Agent Loop Engineering

Security checks across malware telemetry and agentic risk

Overview

This skill is a coherent coding-loop controller with local state files and optional runner/checker scripts, but users should only run its automation runner with commands they trust.

Install only if you want a coding agent to maintain project-local Docs/ state and run bounded development loops. Review any use of scripts/agent-loop-runner.ps1 carefully: provide only a trusted LoopCommand, keep MaxLoops low, and do not use it for secrets, production systems, destructive Git operations, or unattended privileged automation.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • 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
Findings (2)

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The script executes the user-supplied LoopCommand through `powershell.exe -Command` with no validation, restriction, or confirmation. In an agent skill context, this creates a direct arbitrary command-execution primitive in the workspace, so any caller or upstream component that can influence `LoopCommand` can run PowerShell code with the runner's privileges.

Missing User Warnings

Low
Confidence
82% confidence
Finding
The checker is launched as a subprocess with `-ExecutionPolicy Bypass`, which weakens PowerShell's built-in policy guardrails and permits execution of a script even in environments intended to restrict it. While the checker path is somewhat constrained, using bypass normalizes policy circumvention and increases risk if the checker file or its resolution path is compromised.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.