agentleFS
rookie-ricardorookie-ricardo/erduo-skills/skills/web-to-markdown/SKILL.md

web-to-markdown

Skill935 starsChanged 10 months agoMITView on GitHub

Convert a web URL into cleaned Markdown with deterministic routing. Use when Codex needs to read article-like content from links and should apply source-aware fetch strategies: default to r.jina.ai for general pages (including X/Twitter), use defuddle.md for YouTube links, and use browser-impersonated extraction for WeChat/Zhihu/Feishu pages with Mozilla Readability cleanup.

  • Installs packages

10 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

9 rules. 0 are common, 4 are distinctive, and 5 are very specific to this project.

  • 0

    Common

    Most of the 8 edited versions we read keep these.

  • 4

    Distinctive

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    if it fails, fallback to direct HTTP fetch + Readability,

    line 22
  2. Distinctive:

    if direct fetch still fails or returns shell-like content, fallback to browser extraction.

    line 23
  3. Very specific:

    Run from this skill directory (skills/web-to-markdown):

    line 27
  4. Very specific:

    If input is already proxy-formatted (https://defuddle.md/https://...

    line 52
  5. Very specific:

    Use a two-stage strategy for WeChat/Zhihu/Feishu:

    line 56
  6. Very specific:

    If stage 1 fails or returns blocked/shell content, fallback to puppeteer-extra browser impersonation.

    line 59
  7. Distinctive:

    Remove known login modals and backdrop overlays (best effort).

    line 63
  8. Distinctive:

    If special-site extraction fails due to anti-bot checks, account-only pages, or network limits, report failure clearly and ask for fallback …

    line 69
  9. Very specific:

    When --json is used, return:

    line 75

“Common” is measured: kept by at least 6 in 10 of the 8 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Web To Markdown

Convert URLs into usable Markdown by applying domain-aware fetching routes, then return the cleaned content directly.

Quick Workflow

  1. Normalize and validate the input URL.
  2. Select route:
  • r.jina.ai: general web + X/Twitter.
  • defuddle.md: YouTube transcript/content extraction.
  • special-browser-fetch: WeChat/Zhihu/Feishu.
  1. Return markdown text (or JSON metadata if needed).

For generic URLs (non-YouTube, non-WeChat/Zhihu/Feishu), use this fallback chain:

  • try r.jina.ai first,
  • if it fails, fallback to direct HTTP fetch + Readability,
  • if direct fetch still fails or returns shell-like content, fallback to browser extraction.

Commands

Run from this skill directory (skills/web-to-markdown):

npm install
node scripts/url_to_markdown.mjs <url>

Return metadata with markdown:

node scripts/url_to_markdown.mjs <url> --json

Force special-site browser extraction:

node scripts/fetch_special_sites.mjs <url> --json

Routing Policy

  • Default route: https://r.jina.ai/<url>.
  • YouTube (youtube.com, youtu.be): https://defuddle.md/<url>.
  • X/Twitter (x.com, twitter.com): https://r.jina.ai/<url>.
  • WeChat/Zhihu/Feishu: run scripts/fetch_special_sites.mjs.
  • If input is already proxy-formatted (https://defuddle.md/https://... or https://r.jina.ai/https://...), normalize back to the original URL and re-apply routing.

Special-Site Extraction Behavior

Use a two-stage strategy for WeChat/Zhihu/Feishu:

  1. Try cuimp HTTP/TLS impersonation first, then clean HTML with Mozilla Readability.
  2. If stage 1 fails or returns blocked/shell content, fallback to puppeteer-extra browser impersonation.
  • HTTP stage impersonates modern Chrome TLS/HTTP profile via cuimp.
  • Browser stage impersonates a modern Chrome user agent and standard sec-ch-ua headers.
  • Remove known login modals and backdrop overlays (best effort).
  • Scroll the page to trigger lazy-loaded article blocks.
  • Parse cleaned document with Mozilla Readability.
  • Convert extracted HTML body to Markdown via Turndown.
  • Resolve browser executable from CHROME_PATH first, then system Chrome/Chromium/Edge paths.

If special-site extraction fails due to anti-bot checks, account-only pages, or network limits, report failure clearly and ask for fallback input (for example raw page text).

Output Contract

For normal usage, output markdown only.

When --json is used, return:

  • source: backend source (r.jina.ai, defuddle, cuimp, browser-readability).
  • strategy: selected route (r-jina, defuddle, special-http-fetch, special-browser-fetch-fallback).
  • requestedUrl: original input.
  • resolvedUrl: normalized/final URL.
  • markdown: extracted markdown body.

Resources

  • references/routing-and-notes.md: domain routing rules and operational caveats.
  • scripts/url_to_markdown.mjs: primary entrypoint.
  • scripts/fetch_special_sites_http.mjs: WeChat/Zhihu/Feishu HTTP impersonation fetcher (cuimp JS).
  • scripts/fetch_special_sites.mjs: two-stage extractor (HTTP-first, browser-fallback).

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.