Context Visualizer for Agents
An early attempt at context engineering — building a tool to visualize and navigate agent context, born out of a failed hypothesis-tester automation.
This was a project that occurred along side a few attempts at context engineering. I’m early in figuring out exactly how to engineer context to support agents, and this demonstrated that I have a lot to learn.
A separate project that I tried building out this week was a hypothesis tester automation. I’d put a fresh hypothesis into Notion and then an agent would spend the week testing it through cold emails, building things, etc. This failed miserably. The agent had no clue what to do, there was no alignment between what I thought the approach should be and what the agent did, and it was a mess.
So, I’m learning from the ground up. I realized that you can’t shoot for the moon with AI automation and context engineering. You need to build small building blocks that work effectively and then go from there.
I will also flag that this was my first time messing around with Claude Design which was fun. I was getting tired of Claude Code’s ugly front-end designs so I’m grateful they built that.