VividKit Crystal clear AI coding
VI
Choose Your Setup Get Started with AgentKit Continue with ClaudeKit Migrate from ClaudeKit to AgentKit Clean cutover from ClaudeKit What Is AgentKit? How AgentKit Works AgentKit Kits, Skills, Agents & Hooks AgentKit Runtime Support AgentKit Desktop App AgentKit Configuration AgentKit CLI Quick Reference AgentKit Skills Cheatsheet AgentKit Workflows AgentKit Statusline Update AgentKit CLI & Kits AgentKit Helper Clean reinstall of AgentKit AgentKit Troubleshooting What is ClaudeKit? ClaudeKit CLI Guide Command Finder Commands Cheat Sheet Workflow Recipes CLI Quick Reference Claude Mechanics Permission Modes ClaudeKit Hooks Coexistence ClaudeKit x Codex CK → Other coding tools How ClaudeKit Works Inside ClaudeKit Finding Your Unknowns Session Recovery Fix from Logs UI/UX Guide Accounts Switcher IDE Config Remote Control Codex App System Diagramming UI Review Gate Promotions Support VividKit

Quick Ref / Command

Marketing Kit

/ak:agent-browser

Profile-independent browser automation

Use the agent-browser CLI for profile-independent browser and desktop automation: compact snapshots, screenshots, form fills, scraping, exploratory QA, cloud browsers, Electron apps, and Slack workflows.

01

Classify need

02

Check profile dependency

03

Install or refresh

04

Load live guidance

Rule 01

Prefer agent-browser for autonomous browser work that does not need the user's real profile state.

Rule 02

Always load live CLI skill content because command details ship with the installed binary.

Rule 03

Accessibility snapshots and stable refs are the core interaction primitive, not raw page dumps.

Rule 04

Choose specialized workflows for Electron, Slack, dogfood QA, Browserbase, AgentCore, and Vercel Sandbox.

Execution Map

Core lanes from input to output, aligned with the command's real execution path.

01

Start

Input, scope, route

  1. 1 Classify need Use this skill when a fresh or tool-managed browser is acceptable for testing, screenshots, form fills, scraping, QA, Electron, Slack, or cloud sessions.
  2. 2 Check profile dependency If the task needs the user's actual Chrome profile, cookies, account, or workspace state, stop and route to ak:chrome-profile.
02

Work

Agent / skill execution

  1. 3 Install or refresh Use npm install -g agent-browser, agent-browser install, upgrade, and --version as needed so the installed CLI is current.
  2. 4 Load live guidance Before commands, run agent-browser skills get core, or --full when detailed command references and templates are needed.
03

Verify

Gate, review, validation

  1. 5 Select specialty Load electron, slack, dogfood, vercel-sandbox, agentcore, or browserbase/cloud setup guidance when the task leaves ordinary web pages.
  2. 6 Open session Start the local, Browserbase, AgentCore, or sandbox browser session and use the dashboard on port 4848 for observability when useful.
04

Close

Report, handoff, artifact

  1. 7 Snapshot and act Use accessibility-tree snapshots and stable element refs for compact state, then click, fill, navigate, scrape, or capture screenshots.
  2. 8 Recover stale state When elements disappear or commands go stale, re-run snapshots, close stale sessions, or refresh the CLI guidance instead of guessing.

Sample Prompt

Concrete invocations for each flag, subcommand, mode, or route available in the live workflow.

Inspect a site Recommended 
/ak:agent-browser https://example.com
Use when:
A page can be opened in a fresh managed browser for observation or screenshots.
Expected:
Live CLI guidance, a browser session, snapshots, interactions, and evidence from the page.
Automate a task 
/ak:agent-browser fill signup form on staging
Use when:
You need clicks or form fills without using the user's real Chrome login state.
Expected:
A profile-independent automation flow with compact snapshots, browser evidence, and recovery steps for blocked navigation.
Scrape a page 
/ak:agent-browser scrape the pricing table from https://example.com
Use when:
You need compact page data or screenshots without the user's real Chrome profile.
Expected:
A managed browser session with snapshot refs, extracted table evidence, and saved artifacts.

Handled Scope

  • Browser automation
  • Accessibility snapshots
  • Screenshots and forms
  • Scraping
  • Exploratory QA
  • Cloud browsers
  • Electron and Slack automation

Next