Showing growth as a developer requires evidence, and evidence requires a system. My earlier attempts at journaling all died the same way: too much friction, too little return. LLMs changed that arithmetic.

This talk describes the system I've now run for a year. What it captures, how the model helps without writing my thoughts for me, and how the open-source template works. It isn't an app and it isn't a SaaS: it lives in your IDE, in your own workflow, and you're meant to modify it.

I'll cover what changed over the year, including sharper writing prompts and MCP servers pulling context from the tools I already use, plus the pitfalls worth knowing about before you start your own.