ClawHub
Back to skill

Security audit

Ai Agent Helper Free

Security checks for vulnerabilities and agentic risk

Overview

This skill is mostly an AI prompt-design guide, but it asks for broad file and command permissions that are not clearly scoped to that purpose.

Install only if you are comfortable giving this skill broad local file/search/write and command-execution authority. Use it for prompt and ReAct design guidance, and require explicit confirmation before letting it write files, run commands, call APIs, process bulk files, or use callback URLs/API keys.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Intent-Code Divergence

High
Confidence
95% confidence
Finding
The skill is presented earlier as a design/helper skill for prompts and ReAct patterns, but later claims operational capabilities such as file processing, API integration, and command execution. This mismatch can cause an agent or user to invoke a skill with write/exec expectations that were not transparently disclosed up front, increasing the risk of unsafe system-impacting actions.

Vague Triggers

Medium
Confidence
85% confidence
Finding
The invocation scope is so broad that the skill could be auto-selected for many generic AI-related requests outside its narrow safe design purpose. Overbroad matching increases the chance that a tool-enabled skill with read/write/exec permissions is activated in inappropriate contexts, leading to unnecessary exposure to prompt injection or unintended actions.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The markdown advertises file writing, API integration, and command execution without a prominent upfront warning that the skill may affect the local system or external services. In a tool-enabled agent environment, hiding system-impacting behavior in later sections reduces informed consent and increases the likelihood of unsafe execution on user machines or connected resources.

Static analysis

No suspicious patterns detected.