ClawHub
Back to skill

Security audit

novel-master test

Security checks for vulnerabilities and agentic risk

Overview

This is a file-backed novel-writing helper that creates local writing project state but shows no hidden network, credential, or destructive behavior.

Install this if you want a structured, persistent novel-writing workspace. Before using it, confirm the project name and expect files to be created or updated under ~/.qclaw/workspace/novels; for quick one-off story help, avoid invoking the project-management workflow.

SkillSpector

Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • 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
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
90% confidence
Finding
The skill instructs creation and ongoing maintenance of files under a local workspace, including state, trackers, and chapter artifacts, yet no permissions are declared. That mismatch can cause the agent to perform filesystem reads/writes without clear user-visible authorization boundaries, increasing the risk of unintended data modification or leakage if the skill is auto-invoked.

Tp4

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding
The stated purpose is a writing skill, but the content defines substantial operational behavior: creating local project structures, persisting state, maintaining role-process snapshots, and running gate/audit workflows over local files. This broader behavior can surprise users and any orchestration layer, leading to unintended persistent storage and tool use beyond what the description suggests.

Missing User Warnings

Low
Confidence
85% confidence
Finding
The README instructs users to run initialization and state-management scripts that create and update persistent files under a workspace path, but it does not clearly warn that these actions will write to disk and maintain ongoing state. In an agent environment, undisclosed filesystem changes can surprise users, affect existing data, or normalize implicit side effects without explicit consent.

Vague Triggers

Medium
Confidence
82% confidence
Finding
The trigger scope covers very common requests such as writing stories, chapters, suspense, or long-form fiction, making accidental activation likely. In combination with the skill's file-management and stateful workflow, broad triggering can cause unexpected filesystem operations or style/language enforcement when a user only wanted lightweight creative assistance.

Static analysis

No suspicious patterns detected.