Claude Skill MartBrowse skillsWhat is a Skill?
Back to list

Ultimate Browsing (Blocked-Web Escalation)

A three-tier escalation skill that reaches pages blocked by 403, Cloudflare, JS-only rendering, or login walls.

Web & APIAdvanced67,5315,507AI score 7/10Last updated: Aug 9, 2026

What it does

An escalation skill for when a normal browse/fetch fails. It routes the request first, then climbs from the cheapest tool to the heaviest.

  • Tier 1 insane-search: curl_cffi TLS impersonation for WAF/403 bypass, yt-dlp (1858 sites) for subtitles/metadata, official public endpoints (Reddit .json, HN Firebase, Twitter syndication), Wayback/archive.today surrogate snapshots with provenance tagging, Jina Reader, and a Playwright real-Chrome fallback.
  • Tier 1.5 agent-reach: platform-native readers for Xiaohongshu, Douyin, Weibo, Bilibili, V2EX, WeChat articles plus Twitter/Reddit/LinkedIn/GitHub.
  • Tier 2 Chrome stealth: CloakBrowser (fingerprint-patched stealth Chromium) driven by agent-browser over CDP for clicks, forms, screenshots, video playback, and cookie login.

A bundled script extracts local browser cookies (macOS Keychain / libsecret / DPAPI) and can inject them into the CDP session, with 0600 permissions and handling warnings documented.

Who it's for

  • Researchers and data collectors who keep hitting Cloudflare or 403 walls
  • Anyone who regularly needs content from Chinese social/video platforms
  • Developers QA-testing login-gated or JS-rendered web apps

The skill itself says to use web-search for plain queries and webfetch for ordinary pages.

Examples

  1. "This article URL returns 403 — pull the body text" → Tier 1 engine detects the WAF and auto-tries TLS impersonation and archive snapshots.
  2. "Grab this YouTube talk's EN/KO subtitles and summarize" → yt-dlp extracts captions with no browser at all.
  3. "Log into our web app and screenshot the dashboard" → extract/inject cookies, then click and capture with CloakBrowser.

· · · Install guide · · ·

Install in the Claude app (no terminal)
  1. Download the ZIP with the button below.
  2. In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
  3. Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.
Download ZIP
Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the packages/shared-skills/skills/ultimate-browsing folder from the GitHub repo code-yeongyu/oh-my-openagent into my ~/.claude/skills/ultimate-browsing/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/code-yeongyu/oh-my-openagent.git && mkdir -p ~/.claude/skills && cp -r oh-my-openagent/packages/shared-skills/skills/ultimate-browsing ~/.claude/skills/

This is a third-party skill. Check the source repository before installing.

  1. Open a terminal and cd into a working directory.
  2. Clone the repo: git clone https://github.com/code-yeongyu/oh-my-openagent.git
  3. Create the skills folder: mkdir -p ~/.claude/skills
  4. Copy just this skill: cp -r oh-my-openagent/packages/shared-skills/skills/ultimate-browsing ~/.claude/skills/
  5. Make sure Python 3 and yt-dlp are available (pip install yt-dlp if needed). The Tier 1 engine auto-installs its remaining deps on first run.
  6. Optional (Tier 1.5): export the channel env vars you have access to, e.g. JINA_API_KEY, TWITTER_*, and run gh auth login for GitHub.
  7. Optional (Tier 2): follow references/chrome-stealth.md to install CloakBrowser and agent-browser, and confirm the CDP port (default 9242).
  8. Restart Claude Code, then ask something like "this URL is blocked — read it for me" to trigger the skill.
  9. Caution: never keep exported cookie files in shared temp dirs or commit them, and check the target site's terms of service before bypassing protections.
View source on GitHubLicense: NOASSERTION