ClawHub
Back to skill

Security audit

Nous Safety

Security checks across malware telemetry and agentic risk

Overview

This skill is a coherent agent safety layer, but its installer can delete the configured install directory without confirmation if an update fails.

Review the installer before running it. Do not set NOUS_INSTALL_DIR to a broad or valuable directory, back up any existing ~/.nous installation before updates, and prefer installing from a pinned trusted commit in a virtual environment. Start in shadow mode, inspect the local logs and any data sent to your LLM provider, then enable blocking only after tuning the rules.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • 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)

Intent-Code Divergence

Medium
Confidence
89% confidence
Finding
The documentation claims the quick-start runs in 'shadow mode — observe only, no blocking', but the example calls `evaluate_request()` and presents a `BLOCK` outcome without showing any explicit shadow-mode configuration. In a safety-control skill, this mismatch can cause operators to deploy it under the false assumption that it is non-enforcing, leading to unintended blocking or workflow disruption in production.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The script removes the existing installation directory with `rm -rf "$INSTALL_DIR"` after a failed `git pull`, without prompting the user or validating that the path is safe to delete beyond shell quoting. This can cause unintended data loss if the directory contains local modifications, user data, or if `NOUS_INSTALL_DIR` points to an unexpected location; in an install script, destructive fallback behavior increases operational risk.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.