ClawHub
Back to skill

Security audit

Auto Prodcution

Security checks across malware telemetry and agentic risk

Overview

This skill is an automated project-fixing workflow that can modify many files and create commits without a clear user approval checkpoint.

Install only if you intentionally want an agent to take broad control of improving a repository. Use it in a clean branch or disposable workspace, review diffs before trusting commits, and avoid running it on repositories with unrelated local changes or sensitive project tooling.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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 (3)

Context-Inappropriate Capability

Medium
Confidence
97% confidence
Finding
The skill explicitly instructs the agent to run `git add -A && git commit` after each sub-fix, causing persistent repository changes without an explicit confirmation gate. This is dangerous because it can create unwanted commits, capture unrelated local changes, and make automated modification behavior harder for a user to notice or stop once the skill is triggered.

Context-Inappropriate Capability

Medium
Confidence
90% confidence
Finding
The skill directs the agent to run many broad shell-based scanners across the entire project, including security, dependency, testing, and operational checks. While these commands are common in engineering workflows, running them automatically across arbitrary repositories can expose sensitive data in outputs, invoke untrusted project tooling, consume significant resources, and expand the blast radius of a simple skill trigger.

Missing User Warnings

High
Confidence
98% confidence
Finding
The skill mandates immediate code modifications, repeated self-directed iteration, score updates, and git commits without any user warning or consent checkpoint about repository writes. In context, this is especially dangerous because the workflow is designed to continue automatically until thresholds are met, so a single invocation can lead to large-scale, autonomous changes in an arbitrary codebase.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.