Case studies for each project — context, problem, approach, outcome, and what I’d still change.
Testing
A four-stage prompt framework for testing an existing codebase with an LLM, built to trade "write more tests" for knowing what to test and who should test it, then proven against a real open-source app.