Delegate Background Work To Codex Cloud

End-of-lesson check

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

1. What is the recommended first step when assigning work to Codex Cloud?

   1. Write the complete code solution yourself first
   2. Select a pre-built template from the cloud agent library
   3. Define the specific job or task before selecting which tool to use
   4. Choose the most powerful AI model available for the task

2. When delegating work to a cloud agent, what type of task scope produces the best results?

   1. The largest possible scope to minimize round trips
   2. Any scope, as the agent will figure out the boundaries
   3. The smallest useful scope that the agent can realistically complete
   4. A scope that requires multiple follow-up questions

3. After receiving results from a cloud agent, how should the developer test the output?

   1. Show it to other developers for their opinion
   2. Run it again with the same inputs to verify consistency
   3. Run the result as an actual user would, testing real workflows
   4. Deploy it immediately to production

4. Before sharing cloud agent output with others, what three things should a developer inspect?

   1. The cost, execution time, and model version
   2. The diff, data access patterns, and failure path
   3. The prompt used, token count, and API calls
   4. The code style, variable names, and comments

5. What type of tasks are best suited for Codex Cloud agents?

   1. Long-running server processes that need monitoring
   2. Tasks that require access to production databases
   3. Complex architectural decisions requiring stakeholder input
   4. Bounded, parallel tasks that can become branches or pull requests

6. When designing work for a cloud agent, what question should guide the scope?

   1. How many lines of code can the agent generate?
   2. What is the most impressive feature we can add?
   3. Which model has the best reputation for accuracy?
   4. What should the user be able to do when this is finished?

7. What data should an app or cloud agent never expose to users or external systems?

   1. Sensitive data that wasn't intentionally shared, such as credentials or private information
   2. Previous失败的 attempts
   3. Large datasets that would slow down the response
   4. Code that hasn't been tested yet

8. What kind of test proves that a cloud agent's change actually works?

   1. A test that the same model generates the same output
   2. A test that matches the original prompt exactly
   3. A test that verifies the actual user-facing functionality works correctly
   4. A test that checks the code compiles without errors

9. Why is having a rollback path important when using cloud agents?

   1. It enables parallel execution of multiple agent tasks
   2. It makes the agent run faster on subsequent attempts
   3. It allows reverting to a known good state if the agent output is incorrect or harmful
   4. It provides a record of who requested the changes

10. The lesson suggests that while AI can quickly create a working demo, what represents the real developer skill?

    1. Writing documentation for the demo
    2. Making the demo run faster with caching
    3. Turning that demo into something observable, reversible, and safe for others to use
    4. Adding more AI features to the demo

11. How do cloud agent results typically enter the main codebase?

    1. By automatic deployment to production
    2. As branches or pull requests that can be reviewed before merging
    3. Through email approval from the team lead
    4. By directly overwriting the main branch

12. What is the advantage of running tasks in Codex Cloud while working locally?

    1. You can continue local development work while agent tasks run in parallel
    2. The cloud always produces better code
    3. You get free unlimited cloud compute
    4. Your local machine uses less battery

13. When should a developer use the Codex CLI instead of Codex Cloud?

    1. For tight loops where rapid iteration and immediate feedback are needed
    2. When generating documentation
    3. When working without internet connectivity
    4. When the task requires more than 10000 tokens

14. What should a developer verify by inspecting the 'diff' before sharing agent output?

    1. That the diff is formatted consistently
    2. That the diff uses modern JavaScript syntax
    3. That the changes are correct and match the intended functionality
    4. That the diff is shorter than 100 lines

15. What does 'observable' mean in the context of cloud agent output?

    1. The code is easy for new developers to understand
    2. The output appears in the console without errors
    3. The agent's reasoning process is visible
    4. The changes can be monitored, logged, and their effects tracked over time