skills.rest
qqlcx5
qqlcx5COMMUNITY
Explore all skills by this author →
WebsiteGitHub

ego-browser

Automate browser tasks in isolated task spaces using a Chromium-based CLI runtime.

Updated Aug 15, 2026
Search Tags:#testing#screenshots#browser automation#web scraping#form filling#cdp#chromium
One-click install
npx skills add https://github.com/qqlcx5/skills-hub --skill ego-browser-qqlcx5
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: ego-browser
Source: https://github.com/qqlcx5/skills-hub/tree/main/skills/ego-browser
Command: npx skills add https://github.com/qqlcx5/skills-hub --skill ego-browser-qqlcx5

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve? AI agents need to interact with real websites—opening pages, filling forms, clicking buttons, and extracting data—without hijacking the user's own browser session or losing login state. This Skill gives agents an isolated browsing context that reuses the user's login state while keeping automation separate from normal browsing. ## Core Features & Use Cases - Isolated Task Spaces: Each agent task runs in its own browsing context with its own tabs, inheriting the user's login state, with explicit handoff and takeover controls when user intervention (login, captcha) is needed. - Three Automation Workflows: A semantic workflow using snapshotText() refs and locators for standard pages, a visual workflow using screenshots and coordinate clicks for canvas apps like Google Docs or Figma, and a direct DOM/CDP workflow for custom data extraction. - Rich Helper API: Preloaded Node.js helpers for navigation, clicking, typing, scrolling, file upload, screenshots, network fetches, and raw CDP calls, all driven through Bash heredocs. - Use Case: Ask the agent to log into a site, search for items, scrape the results table, and take a verification screenshot—it will create a task space, drive the browser across multiple rounds, and close the space when done. ## Quick Start Ask the agent to open a website with ego-browser, fill in a form, and report back the extracted page content.

Frequently Asked Questions about ego-browser

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

FAQPage Schema
How do I automate browser tasks with an AI agent?▼

Run browser operations through the ego-browser nodejs heredoc, starting with useOrCreateTaskSpace to create an isolated context, then openOrReuseTab to load pages. Use snapshotText() to observe the page and helpers like click, fillInput, and typeText to act on elements.

How do I fill forms and click buttons on a web page programmatically?▼

Call snapshotText() to get a semantic tree with ref numbers, then use click('@N') or fillInput('@N', value) targeting those refs. Stable loc= values from the snapshot or raw CSS selectors also work for long-lived element references.

Can browser automation reuse my existing login sessions?▼

Yes, each task space inherits the current user's login state by default, so agents can operate on authenticated sites without separate logins. The isolated context prevents the automation from disturbing your normal browser windows.

What platforms does ego-browser support for installation?▼

The bundled install script supports macOS only, downloading the correct arm64 or x64 DMG and installing the ego lite app. On other platforms, users must download ego lite manually from the official website.

Why does browser automation fail with a user is controlling error?▼

This error means the user has taken control of the task space through the browser GUI, and it is a hard stop. Do not retry the operation; ask the user to confirm, then resume with takeOverTaskSpace only after explicit confirmation.

When should I use screenshots instead of DOM selectors for automation?▼

Use the visual workflow with captureScreenshot() and coordinate clicks for canvas-like apps such as Google Docs, Figma, maps, and virtualized editors. DOM selectors and fillInput often target hidden fields rather than the real editing surface in these apps.

Related Skills

Explore More Software Engineering Skills→
celeroncoderceleroncoder

ego-browser

Automate Chromium browser tasks through a Node.js CLI with isolated agent task spaces.

Community
Advanced
sungaoxiang-backendsungaoxiang-backend

evidence-alipay-transfer-handler

Extract structured card_features from Alipay transfer screenshots.

Community
Advanced
15491715061549171506

novelty

Verifies research idea novelty via multi-source search and cross-model review.

Community
Advanced
ubay1ubay1
1

react-server-components

Render React components on the server to reduce client JavaScript bundle size.

Community
Intermediate