AGENTS.md Scope And Precedence In Codex

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-codex-agents-md-scope-creators

1. What is the primary function of an AGENTS.md file in a project using Codex?

   1. It stores the project's API keys and authentication tokens for the agent to use
   2. It provides project-specific instructions that Codex reads to follow local conventions
   3. It compiles the code into executable binaries for distribution
   4. It serves as the main documentation for end users learning the codebase

2. In agentic coding, what does the term 'scope' refer to?

   1. The total number of lines of code in the project
   2. The amount of computational resources available to the agent
   3. The specific, bounded portion of work an agent is tasked to complete
   4. The geographic location where the code will be deployed

3. When multiple instructions conflict in a Codex project, what determines which instruction takes precedence?

   1. The number of times each instruction has been executed previously
   2. The length of each instruction in characters
   3. Precedence rules that establish which instruction wins based on hierarchy and specificity
   4. The chronological order in which instructions were written

4. Which principle from the lesson emphasizes completing the smallest useful unit of work?

   1. Run the result as a user, not as a fan of the tool
   2. Inspect the diff, data access, and failure path before sharing
   3. Name the job before naming the tool
   4. Write the smallest useful scope the agent can finish

5. What does it mean to 'run the result as a user, not as a fan of the tool'?

   1. Prefer using official tool documentation over community tutorials
   2. Use the most advanced features of the agent regardless of user needs
   3. Share the tool with as many people as possible to gain fans
   4. Test the output from an end-user perspective rather than admiring the tool's capabilities

6. Which question helps determine what data an application should never expose?

   1. What is the agent's favorite programming language?
   2. What data should the app or agent never expose?
   3. What is the current server time?
   4. How many users have logged in today?

7. What defines a 'reversible' change in the context of agentic coding?

   1. The code can be compiled without errors
   2. The agent can run the same task twice
   3. There exists a rollback path if the output is wrong
   4. The output can be converted to a different file format

8. The lesson states that real skill in agentic coding involves transforming an AI demo into something that is:

   1. Observable, reversible, and safe enough for another person to use
   2. Fully automated without any human oversight
   3. Faster, larger, and more complex than the original demo
   4. Written in the newest programming language available

9. In Codex, what determines which guidance file's rules apply when multiple guidance files exist?

   1. The longest file by line count
   2. The file with the most recent modification date
   3. Random selection by the agent
   4. Precedence rules that establish which instruction wins

10. What is the recommended approach when defining tasks for Codex?

    1. Name the job before naming the tool
    2. Avoid naming tasks and rely on implicit understanding
    3. Use only technical jargon in task descriptions
    4. Name the tool first, then describe the job it should perform

11. What does 'observable' mean in the context of professional-grade agentic coding?

    1. The agent runs on a visible public server
    2. The code can be seen by anyone on the internet
    3. The agent's actions and outputs can be monitored and verified
    4. The project has a graphical user interface

12. Why is it important to inspect the failure path before sharing agent output?

    1. To find syntax errors that prevent compilation
    2. To make the code longer and more impressive
    3. To understand what happens when things go wrong and ensure safety
    4. To count the number of functions in the code

13. What question should be answered to determine if a change works correctly?

    1. What is the agent's model version number?
    2. What test proves the change works?
    3. How many lines were added to the documentation?
    4. What is the favorite color of the lead developer?

14. What question helps identify the rollback path if agent output is wrong?

    1. What is the agent's response time?
    2. What is the current temperature setting?
    3. What rollback path exists if the output is wrong?
    4. How many users have complained about bugs?

15. What question helps define what users should be able to do when a task is finished?

    1. What should the user be able to do when this is finished?
    2. What programming language was used?
    3. What is the server's IP address?
    4. How much memory does the agent use?