ck:agent-browser

Automate browser workflows using snapshot-derived element references for clicks and form fills.

Updated Apr 9, 2026
One-click install
npx skills add https://github.com/jangtrinh/demo-Lam --skill ck-agent-browser-jangtrinh
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ck:agent-browser
Source: https://github.com/jangtrinh/demo-Lam/tree/main/.claude/skills/agent-browser
Command: npx skills add https://github.com/jangtrinh/demo-Lam --skill ck-agent-browser-jangtrinh

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

It solves the problem of running reliable browser automation with minimal context by letting an AI capture a lightweight, interaction-ready “snapshot” of the page.

Core Features & Use Cases

  • Snapshot-driven interaction (refs): Take accessibility-aware snapshots to get stable element references (like @e1, @e2) and then click/fill/type using those refs after each page change.
  • Long, self-verifying sessions: Support multi-step autonomous flows with re-snapshot loops to reduce breakage from dynamic UIs.
  • CI/cloud and debugging support: Use local or cloud browsers (e.g., Browserbase), plus recording, screenshots, PDF export, and network controls for debugging and testing.
  • Use case: Automate logging into an app, navigating through multiple pages, and validating results by repeatedly snapshotting and interacting with only the currently relevant elements.

Quick Start

Open the target site, take an interactive snapshot, fill the login email field, click the sign-in button, then take a new snapshot to continue the workflow.

Frequently Asked Questions about ck:agent-browser

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

FAQPage Schema
How do I automate web form filling with AI agents without breaking on dynamic UIs?

Browser automation for AI agents uses snapshot-derived element references to interact with web forms reliably. By taking accessibility-aware snapshots after each page change, agents capture stable refs like @e1 to click, fill, and type without losing track of elements.

What is ref-based snapshot interaction for cloud browser testing?

Ref-based snapshot interaction captures a lightweight accessibility tree of the page to generate stable element references. AI agents use these refs to execute click, fill, and select commands in cloud browsers, enabling long, self-verifying sessions that reduce breakage from dynamic UI updates.

Can I run long, self-verifying web testing workflows with session and state persistence?

Long, self-verifying web workflows are supported through re-snapshot loops and session persistence. Agents navigate forward or back, take interactive snapshots, and validate results by repeatedly interacting with currently relevant elements across multi-step autonomous flows.

Does browser automation for AI agents support debugging with screenshots, PDFs, and recordings?

Browser automation for AI agents supports debugging with screenshots, PDF export, and recordings. It works with local or cloud browsers like Browserbase, providing network controls and visual outputs to debug interactive web workflows and verification loops.

How do I start automating a login flow using interactive snapshots and element refs?

Open the target site, take an interactive snapshot to get element refs, fill the login email field, and click the sign-in button. Take a new snapshot after the page changes to continue the workflow, using ref-based commands for reliable execution.

Why do my AI agent web automation scripts fail when the page content changes dynamically?

Scripts fail because dynamic UI updates break traditional selectors. Snapshot-driven interaction solves this by capturing the accessibility tree after each page change, providing fresh element refs for every click, fill, or type command to maintain reliable execution.