skills.rest
leonardoacosta
leonardoacostaCOMMUNITY
Explore all skills by this author →
GitHub

agent-browser

Automates browser inspection, testing, and session control through the agent-browser CLI.

Updated Aug 2, 2026
Search Tags:#screenshots#cli#browser automation#web testing#cdp#sessions#accessibility snapshot
One-click install
npx skills add https://github.com/leonardoacosta/agents --skill agent-browser-leonardoacosta
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agent-browser
Source: https://github.com/leonardoacosta/agents/tree/main/skills/agent-browser
Command: npx skills add https://github.com/leonardoacosta/agents --skill agent-browser-leonardoacosta

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? It gives AI agents a governed way to drive a real browser for web automation, inspection, and testing without inventing stale CLI flags or mishandling sessions, profiles, and CDP endpoints. ## Core Features & Use Cases - Version-matched CLI guidance: Verifies the installed agent-browser CLI and loads its bundled core skill before issuing any commands, failing closed when unsupported. - Session and profile governance: Enforces task-unique session prefixes, stable named sessions, approval gates for private profiles and external CDP endpoints, and owned-resource-only cleanup. - Use Case: An agent needs to test a web app login flow. It checks the CLI version, loads the core skill, creates a named session, takes fresh accessibility snapshots after each navigation, and cleans up only its own sessions. ## Quick Start Ask the agent to use agent-browser to open a web page, take an accessibility snapshot, and click a specified element in a named session.

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 tasks with the agent-browser CLI?▼

First verify the CLI with command -v agent-browser and agent-browser --version, then load its bundled core skill via agent-browser skills get core. Use a task-unique session prefix and one stable named session per task, taking fresh accessibility snapshots after page changes.

How do I test a web app using browser automation sessions?▼

Create a stable named session for the task, navigate to the app, and interact using fresh accessibility snapshots rather than stale element references. Clean up only the sessions your task owns when finished.

Can agent-browser use an existing private browser profile?▼

Yes, but it requires explicit approval for the exact profile path, same-user ownership verification, and a profile lock check. It never creates missing profile directories or copies credentials, and reports temporary snapshot or signed-out caveats.

What happens if the agent-browser CLI is not installed?▼

The skill stops and reports the missing CLI rather than substituting stale instructions or inventing flags. The same fail-closed behavior applies when a requested specialized skill such as electron or dogfood is unavailable.

How does agent-browser handle shared CDP endpoints?▼

When two approved sessions share one CDP endpoint, each session pins itself to its own tab using the upstream --pin-tab behavior. It never probes arbitrary ports or switches into another session's tab.

Related Skills

Explore More Software Engineering Skills→
majesticlabs-devmajesticlabs-dev
47

skill-first

Scan installed skills and load a matching skill before starting a task.

Official
Basic
jikig-aijikig-ai
12

reproduce-bug

Reproduce GitHub-reported bugs via logs, code inspection, and Playwright.

Official
Intermediate
microsoftmicrosoft
190.1k

integrated-browser

Document the architecture of the VS Code integrated Chromium browser view.

Official
Advanced
HouseGarofaloHouseGarofalo
3

browserless

Automate headless Chrome browser tasks via REST API and WebSocket.

Official
Intermediate