Building an agentic workflow, part 2: testing what done means
Working through the dot-finance prototypes, including a misleading interruption test, interface corrections, CI storage failures, and a supervised controller trial.

Series
Tracking my learning in AI Development
Working through the dot-finance prototypes, including a misleading interruption test, interface corrections, CI storage failures, and a supervised controller trial.

Setting up my own agentic development loop around dot-finance, with repository context, validation, independent review, and limits on correction attempts.
