Prompt It Yourself
Menu

Make your agent prove its work

  1. 1Hey Claude, create me an automated test suite for our project.
  2. 2Check the test coverage and make sure we are testing every important feature and possible failures.
  3. 3Run the test, fix any problem in the code, not in the test, and run them again.
  4. 4Make testing a permanent part of this project, keep the tests up to date and run them whenever we make changes.

Tip. A test suite won't catch every mistake, but it helps your agent check its own work, catch a problem early and stay accountable.

The story

Computers are predictable

Computers are predictable

We've got used to computers being predictable. You press a button, and you expect the same result every time.

Confident, and wrong

An AI agent can make a mistake and sound completely confident about it. It says done, with a big thumbs up, while what it built leans and cracks behind it.

Ask it to prove it

So when your AI agent builds something for you, don't just ask if it works. Ask it to prove it.

Fix the code, not the test

Then the agent runs the tests. When one fails, the fix goes into the code, not into the test, and the tests run again.

Make testing part of the project

Make testing part of the project

One test run is not enough. Make testing a permanent part of the project: the agent keeps the tests up to date and runs them whenever you make changes.

Not perfect, accountable

A test suite won't eliminate every mistake. It helps your agent check its own work, catch a problem early, and stay accountable.

Build your own Udemy clone

Build your own Udemy clone

Want to build your own Udemy clone, where your AI agent creates the courses for you? That's exactly what we build in my free Fundamentals course, made for non-technical people.

Find more like this

What it solves

More prompts

All 49 prompts