Back to skill

Security audit

Poetry Master - Chinese Classical Poetry Review

Security checks for vulnerabilities and agentic risk

Overview

This is a Chinese poetry critique skill with disclosed, purpose-aligned online metrical checking, but users should know that using the optional checker can send poem text to sou-yun.cn and requires an unpinned Playwright install.

Install this skill only if you are comfortable with online metrical validation. For unpublished or private poems, prefer the included local reference files or manual review instead of running the optional Playwright checker, because that checker submits text to sou-yun.cn. If you use the script, install dependencies in an isolated environment and consider pinning Playwright versions yourself.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (1)

T08 · Insecure Dependencies

Warning
Location
scripts/check_ge_lu.js:16
Finding

Unpinned Playwright Dependency and Browser Installation

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (6)

Tp4

High
Category
MCP Tool Poisoning
Confidence
91% confidence
Finding

The skill’s declared scope is a broad poetry critique and polishing assistant, but its instructions explicitly direct external website use and automated browser-based validation against sou-yun.cn. This creates an integrity and safety gap: users and orchestrators may route requests expecting a self-contained literary assistant, while the skill may trigger network access, leak user-submitted content to third parties, or fail outside its advertised coverage. In a skill-routing environment, this mismatch can cause unintended data exposure and unreliable behavior.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill description and persona are entirely specified in Chinese and direct a specific stylistic response mode, but nowhere offers users a language or locale choice. Under the policy, forcing a specific language without opt-in is a natural-language policy violation unless the locale constraint is clearly documented and justified.

Content

No source excerpt is available for this finding.

Rp1

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The script automatically sends user-provided poem text to sou-yun.cn via browser automation without any explicit consent, warning, or data-handling notice. Even if poems seem low sensitivity, users may submit unpublished work or personal content, causing unintended disclosure to a third-party site and potential retention in server logs.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

This path sends user-supplied ci text and optional cipai metadata to an external website through Playwright automation without informing the user or obtaining consent. In the context of a poetry skill, users may reasonably assume local processing, so silent third-party transfer creates a privacy and trust risk.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

The natural-language comments and usage guidance are presented only in Chinese, and the runtime help/output messages are likewise Chinese-only. Under the language/locale policy, forcing a single language without opt-in or documented justification can be a policy violation unless the regional constraint is explicitly stated.

Content

No source excerpt is available for this finding.