| name | agent-browser |
| description | Browser automation CLI for AI agents. Use when you need to interact with the mrbro.dev site — navigating pages, checking for errors, capturing screenshots, verifying WCAG compliance, filling forms, clicking buttons, or testing any browser-based behavior. Triggers include "open the site", "take a screenshot", "check for errors", "verify accessibility", "test the deployed page", "scrape content", or any task requiring browser interaction with https://mrbro.dev or a local preview server. |
| license | MIT |
| metadata | {"author":"marcusrbrown","version":"1.0"} |
| allowed-tools | Bash(npx agent-browser:*), Bash(agent-browser:*) |
Browser Automation for mrbro.dev with agent-browser
Use agent-browser (via npx agent-browser) to interact with the site in CI or from workflow_dispatch events.
Core Workflow
Every browser session follows this pattern:
- Navigate:
npx agent-browser open <url>
- Snapshot:
npx agent-browser snapshot -i (get element refs like @e1, @e2)
- Interact: Use refs to click, fill, select
- Re-snapshot: After navigation or DOM changes, get fresh refs
npx agent-browser open https://mrbro.dev
npx agent-browser wait --load networkidle
npx agent-browser snapshot -i
Essential Commands
npx agent-browser open <url>
npx agent-browser close
npx agent-browser snapshot -i
npx agent-browser snapshot -i -C
npx agent-browser snapshot -s "#main"
npx agent-browser click @e1
npx agent-browser fill @e2 "text"
npx agent-browser press Enter
npx agent-browser scroll down 500
npx agent-browser get text @e1
npx agent-browser get url
npx agent-browser get title
npx agent-browser wait @e1
npx agent-browser wait --load networkidle
npx agent-browser wait --url "**/blog"
npx agent-browser wait 2000
npx agent-browser screenshot
npx agent-browser screenshot --full
npx agent-browser screenshot --annotate
npx agent-browser diff url <url1> <url2>
npx agent-browser diff screenshot --baseline before.png
Example Prompts for mrbro.dev
Check site for JavaScript errors
npx agent-browser open https://mrbro.dev && npx agent-browser wait --load networkidle
npx agent-browser snapshot -i
Verify all pages load without errors
for PATH in "/" "/about" "/projects" "/blog"; do
npx agent-browser open "https://mrbro.dev${PATH}" && npx agent-browser wait --load networkidle
npx agent-browser screenshot --full
done
Check WCAG accessibility (visual scan)
npx agent-browser open https://mrbro.dev && npx agent-browser wait --load networkidle
npx agent-browser snapshot -i
npx agent-browser screenshot --full accessibility-check.png
Verify dark/light theme toggle
npx agent-browser open https://mrbro.dev && npx agent-browser wait --load networkidle
npx agent-browser screenshot light-theme.png
npx agent-browser snapshot -i -C
npx agent-browser click @e5
npx agent-browser wait 500
npx agent-browser screenshot dark-theme.png
npx agent-browser diff screenshot --baseline light-theme.png
Test navigation links
npx agent-browser open https://mrbro.dev && npx agent-browser wait --load networkidle
npx agent-browser snapshot -i
npx agent-browser click @e2
npx agent-browser wait --url "**/about"
npx agent-browser get url
Test with local preview server
npx agent-browser open http://localhost:4173 && npx agent-browser wait --load networkidle
npx agent-browser snapshot -i
npx agent-browser screenshot local-preview.png
Compare deployed vs local build
npx agent-browser diff url https://mrbro.dev http://localhost:4173 --wait-until networkidle
Command Chaining
Chain commands with && for efficiency — the browser persists between commands:
npx agent-browser open https://mrbro.dev && npx agent-browser wait --load networkidle && npx agent-browser screenshot --full site.png
Integration with CI (workflow_dispatch)
In a GitHub Actions workflow step:
- name: Check site with agent-browser
run: |
npx agent-browser open https://mrbro.dev
npx agent-browser wait --load networkidle
npx agent-browser screenshot --full site-check.png
npx agent-browser snapshot -i
npx agent-browser close