agent-browser

Guide Codex to reproduce local browser scenarios and capture evidence.

Updated Apr 30, 2026
One-click install
npx skills add https://github.com/softmg/hop-and-fill --skill agent-browser-softmg
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: agent-browser
Source: https://github.com/softmg/hop-and-fill/tree/main/.agents/skills/agent-browser
Command: npx skills add https://github.com/softmg/hop-and-fill --skill agent-browser-softmg

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Enables Codex to reproduce browser-based issues on local surfaces by guiding it to use the in-app browser and browser automation tools to inspect local apps and URLs.

Core Features & Use Cases

  • Compatibility alias for legacy workflows referencing agent-browser.
  • Reproduce UI bugs on local surfaces such as localhost, 127.0.0.1, ::1, or file:// URLs.
  • Capture concrete evidence: opened URLs, actions taken, screenshots, console errors, and observed behavior, with clear fail states when tooling is unavailable.

Quick Start

Tell Codex to open a local web app in the in-app browser and walk through reproducing the issue.

Frequently Asked Questions about agent-browser

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

FAQPage Schema
How do I reproduce local browser bugs using automation?

To reproduce local browser bugs, guide the agent to open local URLs like localhost or file:// in the in-app browser, take actions, and capture screenshots and console errors as evidence. It walks through the scenario automatically to observe failures.

Can I capture console errors and screenshots from a localhost web app?

Yes, you can capture console errors and screenshots from a localhost web app. The agent uses the in-app browser and automation tools to record opened URLs, actions taken, and observed behavior to document the issue concretely.

Does browser automation work with file:// URLs and 127.0.0.1?

Browser automation works with file:// URLs, 127.0.0.1, and ::1 local surfaces. The agent inspects these local applications by guiding the in-app browser to interact with the page and collect reproduction evidence.

What happens when browser automation tooling is unavailable?

When browser automation tooling is unavailable, the agent provides clear failure messaging. It relies on Codex browser capabilities and the in-app browser, returning explicit fail states if the required tooling is missing to prevent silent errors.

What's the best way to document UI issues on local surfaces?

The best way to document UI issues on local surfaces is to instruct the agent to open the local app in the in-app browser and reproduce the issue. It captures concrete evidence including actions taken, screenshots, and console errors.