← ClaudeAtlas

browse-oplisted

Headless browser automation. Use when the user asks to open a URL, check a website, scrape data, fill a form, test a page, read web content, take a screenshot, interact with a web app, or any task requiring browser access. TRIGGER when: user mentions a URL, asks to "open", "browse", "check", "visit", "navigate to", "scrape", "read the page", "screenshot", "test in browser", "fill the form", "click", "what does [site] look like", or any web interaction. DO NOT TRIGGER when: user explicitly asks to use Claude in Chrome (visible browser), or the task is purely about reading local files.
vb-tyagi/OP-browse-the-web-skill · ★ 0 · AI & Automation · score 70
Install: claude install-skill vb-tyagi/OP-browse-the-web-skill
# Browse Op Headless browser for Claude Code. Two tiers: agent-browser (default, ultra-lean) and Playwright CLI (fallback). All commands require `export PATH="/opt/homebrew/bin:/usr/local/bin:$PATH"` prepended. **For the full command reference → Read `~/.claude/skills/browse-op/BROWSER_REFERENCE.md`** --- ## Tier 1: agent-browser (Default — Always First) ~200-400 tokens per snapshot. Rust-native daemon. Persists between commands. ### Cheatsheet — Essential Commands ```bash # Open & read agent-browser open <url> agent-browser wait --load networkidle agent-browser wait <ref> # wait for specific element agent-browser snapshot -i -c # interactive-only, compact — leanest output # Get info without full snapshot agent-browser get text [ref] agent-browser get url agent-browser get title # Interact agent-browser click <ref> agent-browser fill <ref> "text" agent-browser select <ref> "value" agent-browser press Enter agent-browser hover <ref> # Capture agent-browser screenshot [path] agent-browser screenshot --annotate # labeled screenshot for vision agent-browser pdf <path> # JavaScript agent-browser eval "document.title" # Cleanup agent-browser close ``` ### Token-Saving Rules - Always use `snapshot -i -c` (interactive + compact) unless you need full page structure - Chain commands with `&&` when you don't need intermediate output - Use `get text <ref>` for specific content instead of re-snapshotting - Only re-snapshot after DOM changes