agent-browser

Automate headless browser interactions via CLI with ref-based element selection.

22|2|Updated Jan 23, 2026

Search Tags:#automation #cli #scripting #web-testing #headless #browser-automation #ref-based

Ready to run

npx skills add https://github.com/tmchow/tmc-marketplace --skill agent-browser-tmchow

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: agent-browser
Source: https://github.com/tmchow/tmc-marketplace/tree/main/plugins/iterative-engineering/skills/agent-browser
Command: npx skills add https://github.com/tmchow/tmc-marketplace --skill agent-browser-tmchow

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

AI agents need reliable, programmable browser interaction to perform tasks like navigation, form filling, and data capture without relying on a human-in-the-loop. This skill provides a CLI-driven workflow using Vercel's agent-browser and ref-based element selection to create deterministic automation.

Core Features & Use Cases

Headless CLI browser automation for agents
Ref-based element selection using accessibility snapshots
Snapshot-driven interactions with deterministic state management
Navigation, form filling, clicking, scrolling, and capturing screenshots or PDFs
Use cases include automated testing, data extraction, and end-to-end task automation

Quick Start

Open a URL, snapshot interactive elements, and perform actions using refs to automate a web task.

Frequently Asked Questions about agent-browser

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I fill forms and capture screenshots in a headless browser?▼

Ref-based element selection identifies interactive elements on a webpage using references extracted from accessibility snapshots. This approach enables deterministic browser automation by allowing agents to target elements reliably without relying on fragile CSS selectors.

Does this browser automation approach support session and profile management?▼

AI agents can perform deterministic website navigation without human intervention by using a CLI-driven headless browser. This skill manages session states and uses snapshot-based discovery to execute multi-step tasks like form filling and clicking consistently.

What is ref-based element selection for web testing?▼

CLI browser automation supports session and profile management to maintain state across multi-step tasks. This ensures that website navigation, form filling, and testing interactions remain deterministic and contextually accurate throughout the automation workflow.

Why use snapshot-driven interactions for browser automation instead of traditional selectors?▼

Automating web testing interactions using accessibility snapshots provides a deterministic way to capture and interact with page elements. This snapshot-driven method allows precise ref-based selection for reliable clicking, scrolling, and form filling in a headless browser.