ClawHub

Build Session

Security checks across malware telemetry and agentic risk

Overview

This skill is transparent about helping autonomous build sessions, but it can make an agent commit and push workspace changes without enough built-in review controls.

Install only if you intend to run autonomous or scheduled build sessions. Use an isolated repo or branch, inspect `git status` and diffs before any commit, avoid `git add -A` unless reviewed, run secret checks, and require explicit approval before `git push`.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Vague Triggers

Medium
Confidence
87% confidence
Finding
The skill is framed for 'any autonomous time' and other common session types without strong scope limits or approval gates, making it easy for an agent to apply it in routine contexts where unsupervised action may be inappropriate. Because the framework then encourages selecting and shipping work, broad applicability increases the chance of unintended autonomous changes beyond the user's intent.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill instructs running git add, commit, and push as a default code workflow, but it does not warn that these commands can stage unrelated files, create permanent history, and transmit code or secrets to a remote repository. In an autonomous agent context, this can cause unintended data exfiltration, disclosure of sensitive changes, or unauthorized repository modifications.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal