Agents that edit code on the AI Agents track. A coding agent reads files, edits a diff, and runs tests. The pass condition is the test suite, not the model's claim that it fixed the bug.
This lesson assumes you already worked through Computer and browser actions.
The idea in practice
Work on a branch. Apply a patch. Run the tests you named in the task. If they fail, the observation is the failure output, trimmed. Do not commit on red tests.
A concrete check
goal = {
'track': 'AI Agents',
'lesson': 'Agents that edit code',
}
checks = [
'input available at decision time',
'score matches the real decision',
'failure case written down',
]
print(goal['lesson'])
for item in checks:
print('-', item)
Run the sketch locally if you have Python. The printout is a reminder of the checks, not a trained model. Replace the strings with the real inputs from your own example before you treat it as a design.
What usually goes wrong
Trusting 'I fixed it' without running tests ships broken code. When this happens, stop adding parameters or tools. Fix the check, the data, or the permission, then run the same example again.
What to write down
- The input you are allowed to use at decision time.
- The output and the score or pass rule.
- One failure you will test on purpose.
- What you will not claim the system can do.
Practice
Write a task card: repo area, failing test command, and the definition of done.
Self-check
- Say Agents that edit code in one sentence that mentions an input and an output.
- Name the failure mode in this lesson and the check that would catch it.
Done when: you can explain this lesson without the page open, and you have a written failure case.