ce-test-browser

Runs end-to-end browser tests on pages affected by a PR or branch using the agent-browser CLI.

Updated Mar 17, 2026
One-click install
npx skills add https://github.com/Norfolk-Group/marcela-norfolk-ai --skill ce-test-browser-norfolk-group
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ce-test-browser
Source: https://github.com/Norfolk-Group/marcela-norfolk-ai/tree/main/skills/compound-engineering/skills/ce-test-browser
Command: npx skills add https://github.com/Norfolk-Group/marcela-norfolk-ai --skill ce-test-browser-norfolk-group

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires agent-browser, gh.

What problem does it solve? Manually verifying that every page touched by a pull request still renders and behaves correctly is slow and error-prone. This Skill automates browser-based verification by mapping changed files to routes and driving a real browser against your local dev server. ## Core Features & Use Cases - Scoped Test Discovery: Accepts a PR number, branch name, or 'current' and uses git diff or GitHub CLI to determine which files changed, then maps them to testable routes. - Automated Browser Interaction: Uses the agent-browser CLI to open pages, snapshot interactive elements, click, fill forms, and capture screenshots in headed or headless mode. - Pipeline and Manual Modes: In pipeline mode it finds a free port and auto-starts the dev server; in manual mode it asks the user for preferences and pauses for human verification of OAuth, email, payment, or SMS flows. - Use Case: After opening a PR that modifies user settings views, run the Skill to automatically load /settings, verify key elements render, click through critical interactions, screenshot the results, and receive a pass/fail summary table. ## Quick Start Ask the agent to run browser tests on the current branch changes, for example by invoking /ce-test-browser with no arguments.

Frequently Asked Questions about ce-test-browser

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

FAQPage Schema
How do I run browser tests on pages changed in a PR?

Invoke the Skill with a PR number, branch name, or 'current'. It uses gh pr view or git diff to list changed files, maps them to routes, and drives the agent-browser CLI to open, snapshot, and interact with each affected page.

What is the agent-browser CLI used for in browser testing?

agent-browser is the exclusive browser automation tool for this workflow. It handles navigation, interactive element snapshots with refs, clicking, form filling, screenshots, and headed or headless execution. Other browser automation tools are explicitly not used.

Can I run browser tests on a custom dev server port?

Yes, pass --port followed by the port number to override detection. Otherwise the Skill checks AGENTS.md, CLAUDE.md, package.json scripts, and .env files for a port, defaulting to 3000. Pipeline mode also scans for a free port automatically.

Does the browser test skill work in headless mode?

Yes, headless mode is the default in pipeline mode and an option in manual mode. Headed mode is available by answering the mode prompt or using the --headed flag, which opens a visible browser window so you can watch tests run.

What happens when a browser test fails?

The Skill screenshots the error state, documents reproduction steps, and asks whether to fix the issue immediately or skip it. Choosing fix triggers investigation and a re-run; choosing skip logs the failure and continues with remaining pages.

Why does the skill stop and say agent-browser is not installed?

The Skill cannot function without the agent-browser CLI, so it checks availability first and halts if missing. Run the /ce-setup command to install the required dependencies, then invoke the test skill again.