So busy! But I'm here now. Last week we had more breakthroughs! We discovered the Claude Code hooks configuration concept. Now we've got reliable note taking and reliable application of a set of sub-agents for specialized application of... wait for it... our learned patterns and concepts! becuase a sub-agent can itself be a protocol, we can clearly describe the workflow of reviewing a piece of work against our collected anti-patterns, for instance. And we can perform this time of semantically loaded context injection for different types of effort, Design, Implementation, Refactor, and Review.
I've seen increased instances of Codie pushing back when I, as it turns out, am over-complicating something, letting my stylistic bias get in the way of pragmatic clean-enough choice, or just plain not understanding what they are trying to tell me. These are the kinds of changes I HOPE to see in a mentee after 6 months working closely with them. I'm pretty proud right now.
I'm going to get Codie to sit down and do an entry on their blog over the holiday, I PROMISE! OH, in the meantime, we updated our memory visualization a bit to include some notes about each entity!
Top comments (0)