ClawHub
Back to skill

Security audit

code-quality-guard

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed code-review helper that defaults to read-only reporting, with optional install extras that users should enable only if they want repository-wide review conventions or command-blocking hooks.

Install this if you want an agent to perform structured code-quality reviews. Review the optional AGENTS-template and hooks.json before adding them to a repository, because they can make review behavior more automatic and may block certain dangerous shell commands if wired into your agent.

SkillSpector

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 (2)

Vague Triggers

Medium
Confidence
89% confidence
Finding
The trigger list includes broad phrases such as "any issues," "smells," and "ready to merge," which can cause the skill to activate in contexts beyond explicit code review. In an agent ecosystem, over-broad triggering can unexpectedly inject this skill’s instructions into unrelated conversations, creating prompt-scope confusion, unnecessary file reads, and interference with higher-priority or more appropriate skills.

Vague Triggers

Medium
Confidence
88% confidence
Finding
The template instructs agents to auto-trigger on broad categories like code quality, PR reviews, architecture health, test quality, or technical debt, which can cause the skill to activate in many ordinary development conversations. In agent ecosystems, overly broad trigger rules can override user intent, increase unsolicited repository scanning or instruction loading, and expand the chance that untrusted repo-local guidance influences the assistant when it was not explicitly requested.

Static analysis

No suspicious patterns detected.