ClawHub

AI Code Review

Security checks across malware telemetry and agentic risk

Overview

This is a scoped ClawHub maintainer review skill that can use GitHub and proof-publishing workflows, but the behavior is disclosed and aligned with that purpose.

Install this only if you want an agent to help with ClawHub maintainer PR/issue workflows. Be aware it can guide the agent to use your GitHub authority for comments, proof publication, labels, and close decisions, so review generated comments and target PRs before allowing write actions.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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
Findings (2)

Description-Behavior Mismatch

Medium
Confidence
95% confidence
Finding
This is a true scope-expansion issue: a skill presented as a code-review tool also instructs the agent to stage, commit, and push repository changes when CI fails on its own PR. That turns a review workflow into a code-modification and publication workflow, increasing the chance of unintended or unauthorized changes being made under the guise of review automation.

Vague Triggers

Medium
Confidence
84% confidence
Finding
The activation phrases are broad enough to match many ordinary requests such as 'review this' or 'check the code,' which can cause the skill to trigger in contexts the user did not specifically intend. Because this skill includes shell commands and repository-affecting guidance, overbroad activation raises the risk of unnecessary command execution or workflow takeover.

VirusTotal

60/60 vendors flagged this skill as clean.

View on VirusTotal