ClawHub
Back to skill

Security audit

Ceo Notify Agents

Security checks across malware telemetry and agentic risk

Overview

This skill matches its stated notification purpose, but it writes persistent cross-agent memory through an under-scoped shell command that needs careful review before use.

Review before installing. Use only trusted messages and known simple agent names, do not include secrets, and prefer a revised version that validates target names, escapes message content, declares shell/filesystem/OpenClaw requirements, avoids hardcoded personal paths, and asks before writing persistent shared memory.

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 (1)

Missing User Warnings

Medium
Confidence
95% confidence
Finding
This skill persists arbitrary user-provided messages into shared log files for other agents without clearly disclosing the persistence, retention, or cross-agent visibility. That creates a real confidentiality and prompt-injection risk because sensitive data or manipulative instructions can be written into a shared memory channel that other agents may later consume as trusted context.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.