skills.rest
obispobruno
obispobrunoCOMMUNITY
Explore all skills by this author →
GitHub

agent-browser

Automate browser interactions via a CLI using accessibility-tree snapshots and element refs.

1|Updated Aug 1, 2024
Search Tags:#screenshots#cli#browser automation#web scraping#form filling#cdp#qa testing
One-click install
npx skills add https://github.com/obispobruno/dotfiles --skill agent-browser-obispobruno
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agent-browser
Source: https://github.com/obispobruno/dotfiles/tree/main/dot_agents/skills/agent-browser
Command: npx skills add https://github.com/obispobruno/dotfiles --skill agent-browser-obispobruno

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires agent-browser, and includes references (resource) components.

What problem does it solve? AI agents and developers need a programmatic way to interact with websites—navigating pages, filling forms, clicking buttons, and extracting data—without parsing raw HTML or maintaining fragile CSS selectors. This Skill provides a fast Rust-based CLI that drives Chrome/Chromium over CDP and exposes compact element references that keep token usage low. ## Core Features & Use Cases - Snapshot-and-Interact Workflow: Capture an accessibility-tree snapshot with @eN element refs, then click, fill, select, or drag elements directly by ref. - Sessions, Auth & State: Named isolated sessions, persistent profiles, saved auth state, OAuth/2FA handling, and encrypted state at rest. - Capture & Observability: Screenshots, PDF export, video recording, Chrome DevTools profiling, network interception, and a dashboard on port 4848. - Use Case: Automate a login flow once, save the authenticated state, then reuse it to scrape a dashboard, capture a full-page screenshot, and record a video of the run for CI evidence. ## Quick Start Install agent-browser with npm, run "agent-browser skills get core" to load the workflow guide, then ask the agent to open a website and take a snapshot of its interactive elements.

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 actions with agent-browser?▼

Run agent-browser open with a URL, then agent-browser snapshot -i to get interactive element refs like @e1. Use those refs with commands such as click, fill, and select to interact with the page, re-snapshotting after any navigation.

How do I fill out and submit a web form programmatically?▼

Navigate to the form, run agent-browser snapshot -i to discover field refs, then use agent-browser fill for text inputs, select for dropdowns, and check for checkboxes. Click the submit button ref and wait for the resulting URL or network idle.

Does agent-browser support saving login sessions and cookies?▼

Yes, agent-browser state save writes cookies and storage to a JSON file, and state load restores it. You can also use --session-name for automatic persistence or --profile for a full Chrome user data directory, with optional encryption via AGENT_BROWSER_ENCRYPTION_KEY.

Can agent-browser automate Electron apps like VS Code or Slack?▼

Yes, load the specialized skill with agent-browser skills get electron for desktop apps such as VS Code, Slack, Discord, Figma, and Notion. A separate slack skill covers workspace automation like checking unreads and sending messages.

Why does agent-browser report a ref not found error?▼

Element refs are invalidated whenever the page changes, so old refs fail after navigation or dynamic updates. Run agent-browser snapshot -i again to obtain fresh refs before interacting.

What are the limitations of agent-browser profiling and recording?▼

Profiling only works with Chromium-based browsers, caps at 5 million trace events, and has a 30-second stop timeout. Video recording outputs WebM, adds slight overhead, and some headless environments have codec limitations.

Related Skills

Explore More Software Engineering Skills→
MantisWareMantisWare

import

Import workspace templates from local files, URLs, or GitHub repositories with collision handling and secret prompts.

Community
Advanced
vercel-labsvercel-labs
41.7k

derive-client

Generates standalone API clients from browser network traffic recorded in HAR files.

Official
Intermediate
vm0-aivm0-ai
76

scrapeninja

Extract data from anti-bot-protected websites via a scraping API.

Official
Intermediate
VYRE-StudiosVYRE-Studios

requesting-code-review

Perform systematic code reviews and categorize feedback into critical, important, and minor issues.

Official
Intermediate