Website Functional Testing Agent

A fully automated, browser-driven smoke-testing agent that takes a URL, runs a deterministic 10-step flow in a real Chromium browser, and produces machine-readable and human-readable reports.

This is the foundation of a digital replica testing agent — a tool that can autonomously validate websites, capture errors, and interact with real pages without human supervision.

---

⚡ Features

• Runs 10 deterministic smoke tests in sequence:

  1. Load page + capture timings (TTFB, DOM, load)
  2. Capture console errors/warnings
  3. Check page title
  4. Check for visible <h1>
  5. Detect the first visible clickable element
  6. Click it + detect navigation/DOM changes
  7. Validate navigation health
  8. Detect presence of form fields
  9. Interact with form (type + submit)
  10. Final correctness check

• Generates:

  • out/run.json — machine-readable structured results
  • out/summary.txt — human summary

• Clean exit codes:

  • 0 = PASS
  • 1 = WARN
  • 2 = FAIL

• CI workflow included (runs on GitHub Actions)

• Zero configuration required

---

📦 Installation

npm install
npx playwright install chromium

⸻

▶️ Run the Agent

npm start https://example.com

Outputs are written to:

./out/run.json
./out/summary.txt

⸻

🧪 Sample Output

summary.txt

URL: https://example.com

step01_load PASS - Load page and collect basic timings step02_console PASS - Check browser console for errors step03_title PASS - Check page title exists step04_h1 PASS - Check for visible

step05_clickable PASS - Detect first visible clickable element step06_click PASS - Click first element and detect navigation/DOM change step07_navcheck PASS - Validate navigation after click step08_formfind SKIP - Detect presence of visible form fields step09_forminteract SKIP - Interact with first form field (type & submit) step10_correctness PASS - Final correctness check

run.json

(Excerpt)

{ "url": "https://example.com", "steps": [ { "id": "step01_load", "status": "PASS" }, { "id": "step02_console", "status": "PASS" }, { "id": "step03_title", "status": "PASS" }, { "id": "step04_h1", "status": "PASS" }, { "id": "step05_clickable", "status": "PASS" }, { "id": "step06_click", "status": "PASS" }, { "id": "step07_navcheck", "status": "PASS" }, { "id": "step08_formfind", "status": "SKIP" }, { "id": "step09_forminteract", "status": "SKIP" }, { "id": "step10_correctness", "status": "PASS" } ] }

⸻

🛠 GitHub Actions CI

This repository includes a CI workflow that runs the agent on push and publishes the test artifacts.

See: .github/workflows/agent-ci.yml

---

🎯 Use Cases • Autonomous UI smoke checks • Overnight monitoring • Pre-deploy validation • Competitive site audits • AI-driven QA agents • “Digital replica” testers for clients or teams

---

📄 License

MIT

Website Functional Testing Agent (Minimal MVP)

Run a tiny Playwright-based agent that executes deterministic smoke steps against a single URL and writes two outputs: out/run.json and out/summary.txt.

Quickstart

1. Install deps: npm install
2. Install browsers: npx playwright install
3. Run: npm start https://example.com

Exit codes

• 0 PASS
• 1 WARN
• 2 FAIL