
Web Access Claude Skill
Extend Claude Code with governed web search, fetch, and CDP browser automation so agents can research and interact with live sites in parallel.
Overview
Web Access Claude Skill is an agent skill for the Build phase that gives Claude Code three-layer web dispatch, CDP browser automation, and parallel sub-agent research patterns.
Install
npx skills add https://github.com/aradotso/trending-skills --skill web-access-claude-skillWhat is this skill?
- Three-layer channel dispatch auto-picks WebSearch, WebFetch, curl, Jina, or CDP per scenario
- CDP proxy attaches to your running Chrome and reuses existing login sessions
- Parallel sub-agent task splitting for multi-page research workloads
- Cross-session site experience accumulation for repeatable domain patterns
- Three browser click modes including JS /click and coordinate /clickAt variants
- Three-layer channel dispatch across WebSearch, WebFetch, curl, Jina, and CDP
- Three documented browser click modes for CDP automation
Adoption & trust: 845 installs on skills.sh; 31 GitHub stars; 0/3 security scanners passed (skills.sh audits).
What problem does it solve?
Claude Code can search and fetch URLs but you still lack rules for dynamic sites, logged-in browsers, and parallel web tasks.
Who is it for?
Indie developers wiring Claude Code for authenticated browsing, dynamic pages, and large parallel web research jobs.
Skip if: Builders who only need occasional static page fetches with no automation, or environments that cannot run local Chrome/CDP.
When should I use this skill?
User asks to give Claude internet access, browse with Claude Code, use CDP automation, or run parallel web research agents—per skill triggers.
What do I get? / Deliverables
Your agent follows a documented dispatch and CDP workflow to browse, scrape, and split web work across sub-agents with reusable site memory.
- Configured dispatch strategy for web tasks in agent sessions
- CDP-connected browsing flows including click and scroll automation patterns
Recommended Skills
Journey fit
Internet access is installed while you are assembling the agent’s tooling layer, not as a one-off launch tactic. The skill wires dispatch rules, CDP proxy, and sub-agent splitting directly into how Claude Code reaches the public web.
How it compares
Skill playbook for tool orchestration—not a replacement for Shopify or vendor-specific API integration skills.
Common Questions / FAQ
Who is web-access-claude-skill for?
Solo builders using Claude Code who need structured internet access beyond bare WebSearch and WebFetch, including CDP-driven Chrome and parallel agent splits.
When should I use web-access-claude-skill?
During Build agent-tooling when you install browsing automation, choose between fetch and CDP per site, or set up parallel web research sub-agents.
Is web-access-claude-skill safe to install?
Browser automation can access logged-in sessions and the open web; review the Security Audits panel on this Prism page and constrain permissions in your own environment.
SKILL.md
READMESKILL.md - Web Access Claude Skill
# Web Access Skill for Claude Code > Skill by [ara.so](https://ara.so) — Daily 2026 Skills collection. A skill that gives Claude Code complete internet access using three-layer channel dispatch (WebSearch / WebFetch / CDP), Chrome DevTools Protocol browser automation via a local proxy, parallel sub-agent task splitting, and cross-session site experience accumulation. ## What This Project Does Claude Code ships with `WebSearch` and `WebFetch` but lacks: - **Dispatch strategy** — knowing *when* to use which tool - **Browser automation** — clicking, scrolling, file upload, dynamic pages - **Accumulated site knowledge** — domain-specific patterns reused across sessions Web Access fills all three gaps with: | Capability | Detail | |---|---| | Auto tool selection | WebSearch / WebFetch / curl / Jina / CDP chosen per scenario | | CDP Proxy | Connects directly to your running Chrome, inherits login state | | Three click modes | `/click` (JS), `/clickAt` (real mouse events), `/setFiles` (upload) | | Parallel dispatch | Multiple targets → sub-agents share one Proxy, tab-isolated | | Site experience store | Per-domain URL patterns, quirks, traps — persisted across sessions | | Media extraction | Pull image/video URLs from DOM, screenshot any video frame | ## Installation **Option A — Let Claude install it:** ``` 帮我安装这个 skill:https://github.com/eze-is/web-access ``` or in English: ``` Install this skill for me: https://github.com/eze-is/web-access ``` **Option B — Manual:** ```bash git clone https://github.com/eze-is/web-access ~/.claude/skills/web-access ``` The skill file is `~/.claude/skills/web-access/SKILL.md`. Claude Code loads all `SKILL.md` files under `~/.claude/skills/` automatically. ## Prerequisites ### Node.js 22+ ```bash node --version # must be >= 22 ``` ### Enable Chrome Remote Debugging 1. Open `chrome://inspect/#remote-debugging` in your Chrome 2. Check **Allow remote debugging for this browser instance** 3. Restart Chrome if prompted ### Verify Dependencies ```bash bash ~/.claude/skills/web-access/scripts/check-deps.sh ``` Expected output: ``` ✅ Node.js 22.x found ✅ Chrome DevTools reachable at localhost:9222 ✅ curl available ``` ## CDP Proxy — Core Component The proxy is a lightweight Node.js WebSocket bridge between Claude and your Chrome instance. ### Start the Proxy ```bash # Start in background (auto-exits after 20 min idle) node ~/.claude/skills/web-access/scripts/cdp-proxy.mjs & # Confirm it's running curl -s http://localhost:3456/ping # → {"status":"ok"} ``` ### Full HTTP API Reference ```bash # ── Tab management ────────────────────────────────────────── # Open new tab, returns tab ID curl -s "http://localhost:3456/new?url=https://example.com" # → {"targetId":"ABC123","url":"https://example.com"} # Close a tab curl -s "http://localhost:3456/close?target=ABC123" # → {"closed":true} # ── Page content ──────────────────────────────────────────── # Execute JavaScript, returns result curl -s -X POST "http://localhost:3456/eval?target=ABC123" \ -d 'document.title' # → {"result":"Example Domain"} # Get full page HTML curl -s -X POST "http://localhost:3456/eval?target=ABC123" \ -d 'document.documentElement.outerHTML' # ── Interaction ───────────────────────────────────────────── # JS click (fast, works for most buttons) curl -s -X POST "http://localhost:3456/click?target=ABC123" \ -d 'button.submit' # Real mouse click via CDP (use for upload triggers, canvas elem