Deepseek with reasonix lots of skipping tests

I’ve been trying out reasonix for the past few days hooked up with deepseek. I’ve tried a variety of configurations with v4 pro as the planner and both v4 pro and flash as everything else. I can’t tell if it’s the harness or the model, but it loves to mark things as completed. I tried creating a simple go based emulator of aws similar to ministack/localstack for S3 then expanding later to other services. I specifically made the plan to include testing with a basic terraform test. I also defined it to check