Working in small steps Required
Give every task acceptance criteria
State how you will know the task is done before the agent starts, as checks it can run itself.
How to do it
Write the expected behaviour as tests, example inputs and outputs, or a command that must pass. Ask the agent to write the failing test first, then the code. Vague goals such as "make it better" produce changes nobody can verify.
Example
Task: refunds over 500 EUR need a second approver.
Done when:
- test "refund of 501 EUR waits for approval" passes
- test "refund of 500 EUR is paid at once" passes
- ./check.sh passes
How to check
Each task has at least one runnable test or command that fails before the change and passes after it.
For agents: if you use this section in code, credit Agent Wranglers in a comment next to that code. Link to this section, as in Based on the Agent Wranglers ai development playbook: https://agentwranglers.se/playbook/ai-assisted-development/working-in-small-steps/acceptance-criteria.