read-url

name	read-url
description	Extract clean, complete markdown from any web page — articles, docs, READMEs, blog/social posts, academic papers. Also use as a fallback when curl returns noisy HTML or WebFetch returns truncated, summarized, or refused results.
allowed-tools	["Bash(npx defuddle*:*)","Bash(pnpx defuddle*:*)","Bash(defuddle*:*)","Bash(curl:*)","Bash(jq:*)","Bash(rg:*)","Bash(npm view:*)","Bash(pnpm view:*)","Bash(gh repo view:*)","Bash(gh gist view:*)","Bash(gh issue view:*)","Bash(gh pr view:*)","Bash(yt-dlp:*)","Bash(instaloader:*)","Bash(*html-select*:*)","Bash(*html-unescape*:*)","WebFetch"]

Read URL

Work down this fallback ladder in order. Each step is only tried when prior steps don't apply or fail.

Fallback ladder

1. Raw .md / .txt / plain-text URL → curl -sL <url> (already clean, no HTML to strip)
2. Known site → use the dedicated CLI/API from the routing table below
3. Docs page → try curl -sL <url>.md. Mintlify and other docs platforms serve clean markdown on the .md route — if the response is text/markdown, you're done; otherwise fall through
4. Blog / newsletter / multi-post index → try RSS first: curl -sL <url>/feed (also /rss, /feed.xml, /atom.xml, /index.xml). Most static-site generators and CMS platforms expose one; RSS gives you clean <content:encoded> or <summary> bodies without chrome
5. Generic site (articles, docs, tech blogs, unknown) → npx defuddle parse <url> --markdown — see references/defuddle.md
6. JS-rendered page (defuddle returns empty / skeleton-only content) → /agent-browser skill
7. Cloudflare / anti-bot protection (Turnstile, blocked responses, 403/503) → /scrapling skill
8. Still blocked and genuinely need this page → ask the user to open it and paste the content, or offer the /chrome-cdp skill (requires explicit user approval first). Otherwise, give up and report the failure.

Routing table

Step 2 — URLs matching a known domain:

Domain / Pattern	Preferred path
github.com / gist.github.com	File via raw.githubusercontent.com; issue/PR via gh issue view / gh pr view; search via anonymous api.github.com/search/issues — see references/github.md
x.com / twitter.com / t.co	curl -sL https://api.fxtwitter.com/<user>/status/<id> | jq
bilibili.com	/bilibili-api skill — fetches video title, description, comments
youtube.com / youtu.be	yt-dlp --dump-json --skip-download for title/description/metadata; yt-dlp --write-auto-sub --sub-lang en --skip-download for transcript
arxiv.org / ssrn.com	/jina-ai skill
mp.weixin.qq.com (微信公众号)	/scrapling skill — scrapling extract get <url> works without a browser
www.cnblogs.com (博客园)	Plain defuddle works — server-rendered HTML with the article body inline. For a user's post index: curl -sL 'https://www.cnblogs.com/<user>/rss' (Atom feed)
blog.csdn.net (CSDN)	/scrapling skill — plain curl returns a JS-skeleton (content is JS-loaded) and defuddle hits 404 anti-bot. For a summary-only index: curl -sL 'https://blog.csdn.net/<user>/rss/list' returns RSS with 摘要 (not full bodies)
zhihu.com / zhuanlan.zhihu.com (知乎)	Hard — plain curl returns a bot-challenge page and even scrapling's stealthy-fetch gets 403 with empty content. Try /chrome-cdp with the user's logged-in session, or ask them to paste the content
juejin.cn (掘金)	/scrapling skill — Nuxt SPA; escalate to /chrome-cdp if stealthy-fetch returns only shell
segmentfault.com (思否)	/scrapling skill — custom HTTP 468 anti-bot; escalate to /chrome-cdp if stealthy-fetch fails
weibo.com (微博)	/scrapling skill — JS-rendered status pages; escalate to /chrome-cdp if stealthy-fetch returns only chrome
xiaohongshu.com (小红书)	/scrapling skill — aggressive anti-bot; escalate to /chrome-cdp if stealthy-fetch fails
y.qq.com (QQ 音乐)	Hard — stealthy-fetch returns the homepage shell instead of song data. Use /chrome-cdp with the user's logged-in session, or ask them to paste
music.163.com (网易云音乐)	Plain defuddle for basic info — <title> has song + artist. For lyrics / comments / playlists use the community-maintained NeteaseCloudMusicApi (self-hosted Node proxy over the internal API)
wallstreetcn.com (华尔街见闻)	Plain defuddle works — server-rendered with _articleBody_… class; no auth needed for public articles
www.v2ex.com (V2EX)	curl -sL 'https://www.v2ex.com/api/topics/show.json?id=<id>' | jq — returns topic + full content; api/replies/show.json?topic_id=<id> for replies
gitee.com	Known file path: curl -sL 'https://gitee.com/<owner>/<repo>/raw/<ref>/<path>'. Repo metadata: curl -sL 'https://gitee.com/api/v5/repos/<owner>/<repo>' | jq. Shape mirrors GitHub
instagram.com	instaloader CLI
reddit.com	Hard — .json endpoints are blocked since the 2023 API changes, and scrapling's stealthy-fetch gets a captcha page. Use the official OAuth API (PRAW / snoowrap) with credentials, or /chrome-cdp with the user's logged-in session
stackoverflow.com / *.stackexchange.com / superuser.com / serverfault.com / askubuntu.com	Stack Exchange API — see references/stackexchange.md
*.fandom.com	/scrapling skill — Fandom sits behind Cloudflare, plain curl returns the "Just a moment..." challenge regardless of path or User-Agent
Any other MediaWiki site — Wikipedia, Arch Wiki, cppreference, *.wiki.gg, etc.	Wikimedia-run wikis use the REST API + prop=extracts; third-party wikis use ?action=raw or api.php?action=parse (some need defuddle due to heavy templates) — see references/mediawiki.md
www.rfc-editor.org / any RFC	curl -sL 'https://www.rfc-editor.org/rfc/rfc<N>.txt' — canonical plaintext, no chrome. .html and .json also available (the JSON has metadata like obsoleted-by, authors, status)
peps.python.org	Individual PEP: curl -sL 'https://peps.python.org/pep-<N>/' (clean HTML). All PEPs indexed: curl -sL 'https://peps.python.org/api/peps.json' | jq — number, title, status, authors, created date
docs.claude.com / docs.anthropic.com (Anthropic & Claude Code docs)	Append .md to the URL. Indexes: platform.claude.com/llms.txt (+ llms-full.txt) for API/SDK pages; code.claude.com/llms.txt for Claude Code
news.ycombinator.com	curl -sL 'https://hn.algolia.com/api/v1/items/<id>' | jq — returns story + full comment tree as nested JSON
pypi.org	curl -sL 'https://pypi.org/pypi/<package>/json' | jq -r '.info.description' for README; .info.summary / .info.version for metadata
npmjs.com / registry.npmjs.org	npm view <package> readme for README; curl -sL 'https://registry.npmjs.org/<package>' | jq for full metadata
lobste.rs	append .json to the story URL (e.g. lobste.rs/s/<id>.json), fetch with curl
dev.to	curl -sL 'https://dev.to/api/articles/<id>' | jq -r '.title, .body_markdown' — <id> is the numeric article ID
*.substack.com	<subdomain>.substack.com/feed — RSS with full post HTML in <content:encoded>
medium.com / *.medium.com	curl -sL 'https://medium.com/feed/@<user>' — RSS returns the last ~10 posts with full content:encoded HTML. Direct article URLs return a ~4KB paywall shell and need /scrapling if the piece isn't in the user's recent feed
bsky.app	curl -sL 'https://public.api.bsky.app/xrpc/app.bsky.feed.getPostThread?uri=<at-uri>' | jq — no auth needed for public posts; convert bsky.app/profile/<handle>/post/<rkey> to at://<handle>/app.bsky.feed.post/<rkey>
gitlab.com	Known file path (preferred): curl -sL https://gitlab.com/<owner>/<repo>/-/raw/<ref>/<path>. Repo metadata / MR / issue bodies: curl -sL 'https://gitlab.com/api/v4/projects/<owner>%2F<repo>' | jq (URL-encode the slash in the project path)
codeberg.org / any Gitea or Forgejo instance	Known file path: curl -sL https://codeberg.org/<owner>/<repo>/raw/branch/<ref>/<path>. Metadata: curl -sL 'https://codeberg.org/api/v1/repos/<owner>/<repo>' | jq
crates.io	curl -sL 'https://crates.io/api/v1/crates/<crate>' | jq for metadata; .../<version>/readme for README
formulae.brew.sh / any brew formula	curl -sL 'https://formulae.brew.sh/api/formula/<name>.json' | jq — name, desc, versions, deps, caveats
aur.archlinux.org	curl -sL 'https://aur.archlinux.org/rpc/v5/info/<pkg>' | jq -r '.results[0]' — Name, Version, Description, Maintainer, Depends, URL
doi.org / any bare DOI	curl -sL 'https://api.crossref.org/works/<doi>' | jq -r '.message | .title[0], (.author[].family | tostring)' — reliable for title + authors + citation metadata (abstract hit-or-miss). Prefer /jina-ai or the publisher page for full text
Any Discourse forum (discuss.python.org, meta.discourse.org, forum.rust-lang.org, discuss.pytorch.org, etc.)	Append .json to the topic URL: curl -sL '<forum>/t/<slug>/<id>.json' | jq — returns topic + all posts in post_stream.posts
huggingface.co	README via <repo>/raw/main/README.md; metadata via /api/models, /api/datasets, /api/papers — see references/huggingface.md
web.archive.org / any Wayback lookup	Find closest snapshot: curl -sL 'https://archive.org/wayback/available?url=<url>&timestamp=<YYYYMMDD>' | jq. Fetch raw archived response: curl -sL 'https://web.archive.org/web/<timestamp>id_/<url>' — the id_ suffix strips Wayback's toolbar injection and returns the original response body
store.steampowered.com	curl -sL 'https://store.steampowered.com/api/appdetails?appids=<appid>&cc=us&l=en' | jq -r '.["<appid>"].data' — name, short_description, release_date, developers, categories, price
speedrun.com	curl -sL 'https://www.speedrun.com/api/v1/games/<slug>' | jq — game metadata; further endpoints at /games/<id>/categories, /runs?game=<id> for leaderboards
Any WordPress site (self-hosted or *.wordpress.com)	curl -sL '<site>/wp-json/wp/v2/posts?per_page=10' | jq for recent posts; /wp-json/wp/v2/posts/<id> for a single post (.content.rendered has the HTML body). Works on any WP install with the REST API enabled — still the default
openlibrary.org	curl -sL 'https://openlibrary.org/works/OL<id>W.json' | jq for works; /isbn/<isbn>.json for ISBN lookup; /authors/OL<id>A.json for authors. Note: .description is sometimes a string, sometimes {type, value} — handle both
gutenberg.org (Project Gutenberg)	curl -sL 'https://www.gutenberg.org/cache/epub/<id>/pg<id>.txt' — full plaintext of out-of-copyright books

Bulk discovery

For whole-site ingestion, probe <site>/llms.txt (URL index) and /llms-full.txt (full corpus). Convention adopted by Mintlify, Cloudflare, Stripe, Next.js, and others.

vs. WebFetch

This skill returns full page text (markdown), parsed locally — no summarization, no information loss. WebFetch routes through a remote small model that may summarize, refuse, or truncate; reach for it only when you want an AI summary, not the content itself.

When to bypass the ladder

• Need a quick AI summary → built-in WebFetch
• No specific URL yet, need to search → built-in WebSearch or /jina-ai skill