TestSprite - automated testing for modern coders. Try it here: https://www.testsprite.com/
Try the hallucinating vibe coder, puppy GIF edition: https://chaos-button-cascade.lovable....
In this HauseBuild+ episode, I walk through an Agentic Coding workflow: use Cursor’s Planning feature to keep AI work focused, then validate the result with TestSprite so we only ship what actually works. We build a tiny Next.js app called “Wins of Today” (public wins that expire after 24 hours, a “cheer” button with throttling, mobile-first UI, no external APIs). After approving Cursor’s plan, we run TestSprite to generate and execute UX/E2E checks, read the dashboard results, and have the agent apply fixes before we re-run the suite.
If you’re curious how to keep scope tight, catch real bugs, and still move fast, this is a practical, no-hype walkthrough.
──────────────
Chapters
00:00 Intro — the “move one button, ship ten features” problem
00:38 Goal & approach: Agentic Coding (plan → test → ship)
01:37 Cursor Planning explained (approve before code is written)
02:01 Meet TestSprite and how it fits the workflow
02:45 App concept: “Win of Today” (24-hour public wins)
03:30 Prompt & requirements (Next.js, SQLite, Tailwind, throttled cheers)
04:20 Tech picks & constraints (no external APIs, rate-limits)
05:12 Plan details → approve → let it build
07:56 First run: posting a win; cheer-button throttle bug
09:03 TestSprite overview (why & when to use it)
09:27 Setup: sign-up, quick install, API key
10:33 Ask Cursor to run TestSprite against the project
11:26 Dashboard: what’s being tested and how
16:18 Watch tests execute on the local app
17:52 Failures found: long-text validation & a11y messaging
19:11 Report & generated test code overview
20:45 Auto-fix prompts to Cursor + rerun tests
23:08 Takeaways & wrap
What you’ll learn
How Agentic Coding keeps AI work aligned (approve a plan before any code).
Using TestSprite to generate and run UX/E2E tests locally and in the dashboard.
Building a small Next.js + SQLite + Tailwind app with ephemeral posts and rate-limited reactions.
Reading results, reproducing failures, prompting fixes, and re-running for confidence.
If you try this flow, drop your country in the comments and tell me what you’re building.