ClawHub
Back to skill

Security audit

Philosophical Three Questions

Security checks across malware telemetry and agentic risk

Overview

This is mostly a navigation reasoning skill, but it also tells the agent to persist memories and later modify or create skills without clear user approval.

Use the navigation framework only if you are comfortable with persistent episode logs. Before installing, require the agent to ask before writing memory, and do not allow it to edit this skill or create new skills unless you explicitly review and approve those changes.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Context-Inappropriate Capability

Medium
Confidence
96% confidence
Finding
The skill expands from a navigation reasoning aid into instructions for persistent memory writes and even self-modification of skills. This is dangerous because it authorizes stateful data storage and code/documentation changes without clear approval boundaries, enabling unintended persistence, prompt drift, or unsafe evolution of agent behavior across sessions.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill instructs the agent to write episode summaries to a persistent file under the user's home directory without notice or consent. Persistent storage can leak sensitive environment details, accumulate behavioral traces, and create cross-session data retention that users did not expect.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.