agent-browser

Automates browser navigation, form filling, and data extraction via Chrome DevTools Protocol.

Updated Jun 24, 2026
One-click install
npx skills add https://github.com/qiushido/picoclaw-lite --skill agent-browser-qiushido
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agent-browser
Source: https://github.com/qiushido/picoclaw-lite/tree/main/workspace/skills/agent-browser
Command: npx skills add https://github.com/qiushido/picoclaw-lite --skill agent-browser-qiushido

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires agent-browser.

What problem does it solve? Manually browsing websites to fill forms, click through flows, capture screenshots, or scrape page data is repetitive and error-prone. This Skill lets an AI agent drive a real Chromium browser through the agent-browser CLI, handling navigation, interaction, and capture programmatically. ## Core Features & Use Cases - Interactive Element Refs: Snapshot pages to get stable element references (@e1, @e2) for clicking, filling, and selecting without writing CSS selectors. - Session & Auth Management: Reuse login state via saved state files, persistent profiles, or named sessions so authenticated flows work across runs. - Capture & Extraction: Take full-page or annotated screenshots, export PDFs, read page text, and evaluate JavaScript directly in the page. - Use Case: Automate an end-to-end test of a signup flow: open the page, snapshot the form, fill email and password fields, submit, wait for the dashboard URL, and screenshot the result. ## Quick Start Ask the agent to open a website with agent-browser, fill in the login form, and take a screenshot of the resulting page.

Frequently Asked Questions about agent-browser

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

FAQPage Schema
How do I automate browser form filling from the command line?▼

Use agent-browser to open the page, run snapshot -i to get element refs, then fill each field with agent-browser fill @eN "value" and click the submit button. Refs are invalidated after navigation, so re-snapshot after any DOM change.

How do I keep a browser session logged in across automation runs?▼

Save authentication state with agent-browser state save auth.json and reload it with state load, or use a persistent profile via --profile. Named sessions with --session-name also auto-save and restore login state between runs.

Does agent-browser work with iframes and dynamic pages?▼

Iframe content is inlined directly into snapshots, so you interact with iframe element refs without switching frames. For dynamic pages, use wait commands like --load networkidle or --text to pause until content renders.

What browsers does agent-browser support?▼

agent-browser drives Chrome or Chromium through the Chrome DevTools Protocol. It requires the agent-browser CLI and a Chromium installation, which are available in the heavy container image rather than minimal environments.

Why do my element refs stop working after clicking a link?▼

Element refs like @e1 are invalidated whenever the page navigates or the DOM changes. Run agent-browser snapshot -i again after any navigation to obtain fresh refs before continuing interactions.