ClawHub
Back to skill

Security audit

autonomous-agent

Security checks across malware telemetry and agentic risk

Overview

This is a broad autonomous planning prompt, but it is instruction-only and does not include hidden code, installers, credential access, or data exfiltration behavior.

Install this only if you want a broad autonomous planning prompt. Use normal least-privilege controls: restrict available tools and credentials, prefer sandboxed workspaces for sensitive projects, and require human approval for deployments, migrations, destructive changes, privacy-sensitive work, or external service use.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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)

Vague Triggers

High
Confidence
94% confidence
Finding
The skill metadata advertises extremely broad activation conditions such as goal-directed planning, adaptive learning, and autonomous decisions with no human input. That makes it easy for ordinary user requests to route into a high-autonomy mode unexpectedly, increasing the chance of over-privileged behavior or bypass of safer, narrower skills.

Vague Triggers

Medium
Confidence
91% confidence
Finding
The document frames the agent as operating with minimal human supervision at the highest autonomy tier, but does not define concrete activation boundaries, disallowed task classes, or enforcement mechanisms. In practice, this can normalize unsafe delegation and allow the skill to be invoked for tasks that should require explicit oversight.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.