ClawHub
Back to skill

Security audit

Skill Factory

Security checks across malware telemetry and agentic risk

Overview

The skill does what it says, but its one-step publish command can upload local skill contents under an authenticated account and uses unsafe shell command construction.

Install only if you are comfortable auditing generated skills before publication. Before running publish, review the package contents for secrets or private logic, verify the ClawHub account and destination, and avoid untrusted skill directories or SKILL.md metadata until the shell-string execSync calls are replaced with argument-array execution and a confirmation or dry-run step is added.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Lp3

Medium
Category
MCP Least Privilege
Confidence
88% confidence
Finding
The skill instructs use of environment-dependent external tooling (`npx clawhub@latest login`) and a publish pipeline, but the skill metadata shown in this file declares no permissions or capability boundaries. That mismatch makes it easier for an agent or user to invoke credentialed or environment-sensitive operations without clear review, increasing the chance of unintended secret exposure or unauthorized publishing.

Missing User Warnings

Medium
Confidence
97% confidence
Finding
The publish workflow explicitly packages and publishes artifacts to ClawHub, but the skill does not present a clear warning that this is an external side effect with potentially irreversible consequences. In an agent setting, this can lead to accidental disclosure of private code, prompts, or embedded secrets by turning a local scaffold into a publicly distributed artifact.

Missing User Warnings

Medium
Confidence
83% confidence
Finding
The publish command performs an external network upload to ClawHub via npx without an explicit confirmation prompt or clear warning at execution time that local skill contents will be transmitted off-host. In a skill-building context, generated files may contain sensitive prompts, code, tokens, or proprietary logic, so accidental publication can cause data leakage.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

Detected: suspicious.dangerous_exec

Shell command execution detected (child_process).

Critical
Code
suspicious.dangerous_exec
Location
scripts/factory.js:154