Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
Version History
Release v0.8.2
What This Skill Does
Pinchtab controls a Chrome browser via a local HTTP API, exposing the page's accessibility tree as flat JSON with stable element refs. Designed for AI agents doing web automation, scraping, form filling, and multi-tab workflows. Optimized for low token cost compared to screenshot-only or full DOM approaches.
The accessibility tree output uses 56-93% fewer tokens than full DOM or screenshot approaches, making it significantly cheaper to run browser automation inside AI agent loops.
When to use it
- Scraping product prices from e-commerce sites
- Automating login and form submission on internal tools
- Extracting structured data from paginated web reports
- Taking screenshots of web pages for visual verification
- Running multi-tab browser workflows in CI pipelines
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Go to example.com and find all the navigation links
- 1Start pinchtab server locally on port 9867
- 2Navigate to https://example.com via POST /navigate
- 3Wait 3 seconds for Chrome to render the accessibility tree
- 4Call GET /snapshot?filter=interactive&format=compact to get element refs
- 5Filter results for role=link elements and return their text and href values
A list of navigation links with their labels and URLs extracted from the accessibility tree
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.