ClawHub
Back to skill

Security audit

clawhub-publish-flow

Security checks across malware telemetry and agentic risk

Overview

This is a real ClawHub publishing helper, but it can upload an entire local skill folder to a public registry using the user's saved ClawHub token without script-level preview or safeguards.

Use this only for deliberate ClawHub releases from a clean skill directory. Before running it, inspect the exact files that will be uploaded, remove secrets, .env files, logs, build output, hidden local config, and test data, and confirm the target slug/version with ClawHub first.

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)

Tp4

High
Category
MCP Tool Poisoning
Confidence
83% confidence
Finding
The skill claims safety-critical release checks such as inspecting remote state before update and validating release context, but the described implementation does not enforce those checks and appears to rely on a generic upload path. In a public publishing workflow using an already-authenticated local session, this mismatch can lead to accidental overwrites, incorrect version publication, or release of unintended content because operators may trust controls that are not actually implemented.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.