WeChat Work Doc Fetcher

Fetch and convert WeChat Work developer docs pages into clean Markdown files for use in Obsidian, handling SPA content and required authentication.

Key points to consider before installing/using: - The tool needs an authenticated session cookie to fetch protected pages. The SKILL.md's wording that Playwright 'gets session cookies automatically' is misleading — the script uses Playwright only to extract doc_id and does not transfer browser cookies into the requests.Session. You will usually need to supply cookies via --cookies or by editing COOKIES_RAW. Treat those cookies like passwords: only paste them into the script on machines you trust, and consider revoking the session after use. - Playwright requires installing a headless Chromium (~150 MB). Install it only if you accept that download and run browser automation locally. - The script only contacts developer.work.weixin.qq.com (no other remote endpoints). You can verify network calls by reviewing the code (fetch_doc uses a single POST to the site) or by running the script in a network-monitored/isolated environment. - If you want the advertised 'automatic' behavior (no manual cookie paste), you or the author would need to modify the script to extract cookies from Playwright and transfer them into the requests.Session before calling the API; as-is, the documentation overpromises. - If you are uncomfortable pasting session cookies into a script, use the manual fallback to get doc_id and then query the API using a browser-exported curl only on an environment you control, or ask the author to add Playwright cookie transfer or OAuth support. Run the script in an isolated environment (container/VM) if possible.

Type: OpenClaw Skill Name: wecom-doc-fetcher Version: 1.0.0 The `wx_doc_fetch.py` script contains significant vulnerabilities. It directly uses user-provided input for the output file path (`args.output`) without sanitization, leading to a file path traversal vulnerability (e.g., writing to `../../../../etc/passwd`). Additionally, the script fetches content from user-provided URLs (`args.url`) using `requests` and `playwright` without domain validation, creating a Server-Side Request Forgery (SSRF) vulnerability. Both vulnerabilities could be exploited by a malicious user through prompt injection against the OpenClaw agent, instructing it to use the skill with harmful arguments.