ClawHub
Back to skill

Security audit

Bash

Security checks across malware telemetry and agentic risk

Overview

This is a Bash scripting guidance skill with disclosed local preference storage and no hidden execution, exfiltration, or persistence behavior.

Installers should understand that this skill can influence how an agent writes shell scripts, including scripts that delete files, call APIs, use sudo, or run unattended jobs. Those topics are expected for Bash hardening, and the artifact generally frames them with safety checks, dry-runs, validation, and explicit confirmation. The only persistent local state described is user preference configuration under ~/Clawic/data/bash/config.yaml.

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 (1)

Intent-Code Divergence

Medium
Confidence
91% confidence
Finding
The documentation states that `tput` styling 'degrades to empty strings', but the sample code uses command substitution without guarding failures. In real scripts, `bold=$(tput bold)` and similar assignments can fail when `TERM` is unset/invalid or terminfo is unavailable, and under strict error handling this may abort the script instead of degrading safely, creating reliability and availability issues in CI, cron, and nonstandard terminal environments.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.