ClawHub
Back to skill

Security audit

AIOS 自改进 Agent

Security checks across malware telemetry and agentic risk

Overview

This skill locally records an agent's mistakes and reusable lessons, with clear limits intended to keep data inside the current workspace.

Install this only if you want the agent to keep local workspace learning notes about errors, corrections, and reusable practices. Review `.learnings/` periodically, especially in sensitive projects, because the skill is designed to persist sanitized summaries even though it tells the agent not to store secrets or cross-user data.

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 (1)

Vague Triggers

Medium
Confidence
91% confidence
Finding
The default prompt is broad enough to trigger on common requests about remembering corrections, failures, or reusable experience, which can cause the skill to be invoked implicitly in situations the user did not clearly intend. In a skill that reads and writes workspace learning data, over-broad invocation increases the chance of unnecessary data persistence, accidental capture of sensitive context, or unreviewed modification of `.learnings/` state.

VirusTotal

52/52 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.