Ask For The Test Before The Fix

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-coder-tests-first-agent-creators

1. What is the primary benefit of having an AI agent reproduce a bug with a failing test before writing a fix?

   1. It automatically writes the fix for you
   2. It reduces the number of tests needed
   3. It makes the code run faster
   4. It proves the bug exists and gives a clear target for the fix

2. In test-driven development, what does the 'red-green' cycle refer to?

   1. Writing a failing test, then making it pass, then refactoring
   2. Testing on separate staging and production environments
   3. Writing code first, then tests to verify it
   4. Using red and green color coding in the IDE to mark files

3. A developer wants an AI agent to fix a login bug. What should they ask the agent to do FIRST?

   1. Check the production database for corruption
   2. Write the code to fix the login issue
   3. Deploy a temporary patch to production
   4. Write a test that reproduces the login failure

4. Why is it important to 'run the result as a user, not as a fan of the tool'?

   1. To save development time and costs
   2. To verify the actual functionality works for end users
   3. To impress stakeholders with the demo
   4. To make sure the tool looks visually appealing

5. What does it mean to 'name the job before naming the tool'?

   1. Define the problem and goal before selecting how to solve it
   2. Give your project a catchy title
   3. Pick a framework before writing any code
   4. Choose a programming language before starting

6. An AI agent can create a working demo quickly. What makes that demo safe enough for another person to use?

   1. It has a polished user interface
   2. It is observable, reversible, and has tests
   3. It uses the latest AI model available
   4. It uses the most popular libraries

7. What does 'scope' mean in the context of an AI coding task?

   1. The amount of memory the program uses
   2. The specific, contained piece of work the agent should complete
   3. The visual area of a web page
   4. The number of files in the project

8. Why might an AI agent 'patch around the symptom' instead of fixing the root cause of a bug?

   1. Because it didn't first reproduce the actual failure with a test
   2. Because the code is too complex
   3. Because the user didn't ask nicely
   4. Because the model's training data is outdated

9. What is a 'rollback path'?

   1. A function that reverses the order of a list
   2. A navigation route to return to a previous page
   3. A way to undo a change if something goes wrong
   4. A backup of the database

10. What is the 'smallest useful scope' for an AI coding task?

    1. Just the documentation and comments
    2. A single, complete piece of functionality the agent can finish
    3. The entire application feature set
    4. A rough sketch of the idea without code

11. In continuous integration, what role do automated tests play?

    1. They automatically verify code changes work correctly
    2. They are optional and rarely used
    3. They slow down the build process
    4. They replace the need for code reviews

12. When deploying AI-generated code, what question should be asked about data?

    1. How fast the data loads in milliseconds
    2. What color theme the data uses
    3. What data the app or agent should never expose
    4. Which programming language the data prefers

13. What risk exists when letting an AI agent fix a bug without first seeing a failing test?

    1. The tests will take too long to run
    2. The agent might patch symptoms rather than fix the root cause
    3. The agent will refuse to write code
    4. The test will cause the program to crash

14. What makes code 'observable'?

    1. It uses descriptive variable names
    2. It has many comments explaining each line
    3. It is open source and public
    4. It can be monitored and its behavior tracked

15. When should an AI agent consider the 'rollback path'?

    1. Never, that's not the agent's responsibility
    2. Only at the very start of a new project
    3. After the code has already been deployed
    4. When fixing bugs or making changes that might need to be undone