Security audit

网文拆文工具-免费版

Security checks for vulnerabilities and agentic risk

Overview

The skill’s novel-analysis purpose is coherent, but it asks for broad command execution and loosely scoped file/network behavior that users should review before installing.

Review this skill before installing. Use it only in a workspace where you are comfortable letting the agent read manuscript files, write report files, and run local commands. Avoid callback_url for private or copyrighted text unless you fully trust the destination, and choose output paths carefully to prevent unintended overwrites.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

Context-Inappropriate Capability

Medium

Confidence: 90% confidence
Finding: The skill declares `exec` alongside `read` and `write` even though the documented purpose is text analysis and report generation. In an agent environment, unnecessary shell execution materially expands the attack surface: crafted prompts or malformed parameters could cause arbitrary local command execution, file access, or chaining with network-capable system tools.

Vague Triggers

Medium

Confidence: 82% confidence
Finding: The statement that users can trigger the skill through natural-language instructions is overly broad and lacks clear activation constraints. In agent systems, vague trigger rules increase the chance of unintended invocation on unrelated content, which becomes more dangerous here because the skill also has `write` and `exec` capabilities.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The report export feature explicitly writes Markdown output but does not warn users that this results in local filesystem writes. In an agent context, undisclosed write side effects can surprise users, overwrite existing files, or be abused through path manipulation if output paths are user-controlled.

Missing User Warnings

Medium

Confidence: 86% confidence
Finding: The optional `callback_url` parameter implies asynchronous delivery of results over the network, but the skill does not warn that analysis data may be transmitted externally. This can lead to unintended data exfiltration, especially if users provide copyrighted or sensitive manuscript content for analysis.

Static analysis

No suspicious patterns detected.