agentleFS
miguelclavelmiguelclavel/ux-research-skills/skills/ai-browser-walkthrough/SKILL.md

ai-browser-walkthrough

Skill1 starsChanged 38 days agoMITView on GitHub

Use an AI agent driving a real browser to attempt tasks on a live site or prototype and report where it got stuck, before testing with people. Use when someone wants a quick automated usability pass, a keyboard only walkthrough, a pre test of a flow, or to check a prototype works end to end with Claude in Chrome, Playwright or a similar browser tool.

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Don't run it on production with real accounts, payments or personal data.

    line 31

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

AI browser walkthrough

An AI agent can't tell you what people feel, but it can try your flow in a real browser in minutes and show you where it breaks. Use it before a usability test, never instead of one.

Steps

  1. Write the tasks as scenarios, exactly as you would for people ("Find out if you could pay less for car insurance"), plus success criteria.
  2. Run each task in a real browser (Claude in Chrome, Playwright, or similar). For each step record: what the agent looked for, what it clicked or typed, what happened, and a screenshot at every decision point.
  3. Run it three ways:
    • As a first time visitor, using only what's visible on screen
    • Keyboard only: Tab, Shift Tab, Enter, Space, arrows and Escape, noting where focus is lost or invisible
    • On a phone sized window (about 390 pixels wide)
  4. Flag friction: steps that needed guessing, labels that didn't match the goal, dead ends, errors, slow loads, missing states (empty, loading, error), and anything that only worked by luck.
  5. Check the basics on each page: headings, labels on form fields, alt text, contrast on key text, and console errors.

Output

Per task: success or failure, the step by step path with screenshots, and friction points ranked by severity. Then a list of hypotheses to test with real people.

Be honest about the limits

The agent reads pages differently from people: it may find things a person wouldn't and miss confusion a person would feel. Label every finding "AI walkthrough", and never report it as user research.

Don't

  • Don't run it on production with real accounts, payments or personal data. Use test accounts and sample data.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.