ClawHub
Back to skill

Security audit

Juejin Skills

Security checks across malware telemetry and agentic risk

Overview

The skill is not clearly malicious, but its account-based Juejin publishing features are documented with inconsistent and under-scoped instructions that could lead to unintended file use or posting.

Install only if you intend to let an agent use your Juejin account, read local Markdown files, write downloaded articles/images under an output folder, and send article content to Juejin. Use draft-only mode by default, require a preview of the source file, title, tags, category, and visibility before any publish action, and verify where cookies are stored and how to revoke them.

SkillSpector

By NVIDIA
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 (6)

Intent-Code Divergence

High
Confidence
97% confidence
Finding
The documented output schema describes a grading/audit result unrelated to the skill’s stated Juejin functions such as ranking retrieval, article publishing, and downloading. This kind of schema mismatch can cause an agent to mis-handle results, make incorrect follow-up decisions, or invoke the skill under false assumptions, which is especially risky for a skill that can read/write files and publish content externally.

Intent-Code Divergence

High
Confidence
97% confidence
Finding
The input parameters shown ('content', 'strict_level') do not align with the described Juejin actions, which should require parameters like category, article URL, file path, publish flags, or tag selections. This discrepancy can mislead an agent into supplying the wrong data, potentially triggering unintended file access or content publication workflows.

Vague Triggers

Medium
Confidence
88% confidence
Finding
The description includes broad, inconsistent invocation guidance such as database and SQL usage that is unrelated to the Juejin skill. Overbroad guidance increases the chance an agent will invoke this skill in inappropriate contexts, which matters here because the skill has exec, read, write, browser login, cookie handling, and external publishing behaviors.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The quick-start and usage sections do not prominently warn that the skill may read local Markdown files, write downloaded content to disk, authenticate with browser-obtained cookies, and publish content to an external service. Missing warnings increase the risk of unsafe invocation by users or agents who may not realize these side effects until after sensitive actions occur.

External Transmission

Medium
Category
Data Exfiltration
Content
| 文章发布 | 通过掘金 API 创建草稿并发布，支持设置分类、标签、摘要、封面图 |
| 草稿管理 | 支持保存为草稿而不立即发布 |

* 创建草稿：`POST https://api.juejin.cn/content_api/v1/article_draft/create`
* 发布文章：`POST https://api.juejin.cn/content_api/v1/article/publish`
* 获取标签：`POST https://api.juejin.cn/tag_api/v1/query_category_tags`
Confidence
78% confidence
Finding
https://api.juejin.cn/

External Transmission

Medium
Category
Data Exfiltration
Content
| 草稿管理 | 支持保存为草稿而不立即发布 |

* 创建草稿：`POST https://api.juejin.cn/content_api/v1/article_draft/create`
* 发布文章：`POST https://api.juejin.cn/content_api/v1/article/publish`
* 获取标签：`POST https://api.juejin.cn/tag_api/v1/query_category_tags`

**鉴权方式**：Cookie 鉴权（通过 Playwright 浏览器登录获取）
Confidence
89% confidence
Finding
https://api.juejin.cn/

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.