zhihu blog generator

知乎风格技术博客生成器 - 自动化生成高质量技术文章

What this skill will do: run Node.js scripts that search the web (ArXiv, GitHub, blogs, Zhihu, etc.), download or parse papers, generate and refine a Zhihu-style Markdown article, and write many files under a session directory (default: D:\techinsight\reports\blog_{session_id}). Things to consider before installing/running: - Network activity and downloads: the collector and paper parser will fetch web pages and PDFs; installing Playwright (listed as a dependency) will download browser binaries. If you run this in a sensitive environment, expect outbound network connections and file downloads. - Filesystem writes: by default the skill writes to D:\techinsight\reports. Change lib/config.js to point to a safe sandbox directory (or run in a container) before executing. It will create many files (web page markdowns, parsed paper JSON, drafts, refined articles) and could overwrite files if you re-use session IDs or change paths. - Dependencies & runtime: you need Node.js and to install NPM packages (Playwright, pdf-parse, uuid, possibly others). Playwright has heavy installation steps and downloads large browser binaries. - Review network-related modules: inspect lib/search_utils.js and lib/paper_parser.js (they are present but not fully shown) to confirm what sites/APIs are called and whether any credentials or unusual endpoints are used. - CLI injection caution: run_all spawns node processes with shell:true and forwards CLI topic arguments. Avoid passing untrusted user input directly into run_all or the scripts; if you must, sanitize/quote inputs or run scripts individually. - Run in a sandbox: if you are unsure, run this inside a disposable VM, container, or isolated workspace so you can inspect outputs and network traffic. If you only want the text-generation logic without web scraping, consider editing the collection step to use curated local sources or mock data. If you plan to use the tool, change the default baseDir to a directory you control, audit the search and parser modules, and install dependencies in a controlled environment.

Type: OpenClaw Skill Name: zhihu-blog-generator Version: 1.0.0 The bundle is a legitimate automation tool designed to generate technical blog posts in a 'Zhihu' style. It implements a multi-step workflow including topic selection, web/paper data collection, draft generation, and content refinement. While the tool performs file system operations (writing to a hardcoded path `D:\techinsight\reports` in `lib/config.js`) and executes internal sub-scripts via `child_process.spawn` in `scripts/run_all.js`, these behaviors are strictly aligned with its stated purpose. No evidence of data exfiltration, malicious prompt injection, or unauthorized remote access was found.