Writing tools for the agent that wrote them

Most of what I have shipped this year has been built using Claude Code. That part is no longer interesting on its own — half the engineering timeline is talking about it. What is more interesting is that one of the projects I have been working on is also for Claude Code to use, written by Claude Code, to extend what Claude Code can do for me. The agent built its own tools, and now the agent uses them. I want to talk about that loop, because the second time you encounter it the implications start to widen out. The project itself is not the point of this post. Pipeline shape: it pulls some external state into a local snapshot, runs deterministic rules over the snapshot to produce a queue of items, and then needs an agent to evaluate each item — decide whether it is real, relevant, and worth surfacing — before producing a final artefact for the user. The evaluation step is the agentic part. When I started, I built it the way you would expect: an in-process loop that called Anthropic’s…