ClawHub
Back to skill

Security audit

Code Reviewer

Security checks across malware telemetry and agentic risk

Overview

This is a read-only code review skill whose repository and PR access is disclosed and aligned with its purpose.

Install this only where you are comfortable having an assistant read code diffs and nearby changed-code context. Avoid using it on repositories containing secrets or sensitive code that should not be sent to an AI model, and be aware that saying "review" inside a git repo is intended to review uncommitted changes.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (4)

Vague Triggers

Medium
Confidence
95% confidence
Finding
The trigger phrase "review" is extremely generic and likely to collide with ordinary user conversation, code review requests outside the intended workflow, or other installed skills. In an agent environment, broad triggers can cause unintended activation, leading the skill to access repository diffs or analyze code when the user did not explicitly intend to invoke this specific skill.

Vague Triggers

High
Confidence
89% confidence
Finding
The trigger set includes very broad phrases like 'review' and 'check my changes', which can cause unintended activation in ordinary conversation or in contexts unrelated to a deliberate code-review request. Because the skill has network access and permission to run commands such as git and shell utilities, accidental activation can lead to unnecessary inspection of local repositories or remote PR content and increased exposure of sensitive code context.

Vague Triggers

Medium
Confidence
82% confidence
Finding
The top-level usage section repeats broad invocation examples without clearly constraining them at the point of use, even though a later activation guard adds nuance. In practice, early broad trigger guidance increases the chance that an orchestrator or agent framework invokes the skill on weak signals, causing over-collection of repository data or unnecessary network fetches for PRs.

Vague Triggers

Medium
Confidence
84% confidence
Finding
The evaluation suite promotes a very broad natural-language trigger ('review my changes' asserted via generic 'review' phrasing), which can encourage over-activation in ambiguous contexts. In an agent system, overly permissive triggers increase the chance the skill runs on unintended input, potentially causing unnecessary repository inspection or tool use when the user did not mean to invoke code review.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.