Theo finds small agent tools rewarding
Theo says assembling a small tool, finding what works, and watching an agent use it is unexpectedly enjoyable. He compares writing these skills with setting up an environment for a team of developers. In his experience, the feedback arrives sooner and a failed attempt costs less than team environment work.
Replace one recurring workaround with a small tool
When a manual workaround keeps recurring, set aside a short iteration to replace that specific workaround. Build the smallest working tool, document it as a skill, and run it through the agent. Refine what failed while keeping the change small and reversible so another attempt remains inexpensive.
Try small tools when bigger efforts discourage you
This is relevant when you avoid tooling because previous environment work demanded substantial effort with little recognition. A recurring workaround gives the small experiment a concrete task to address and test through the agent. Without that recurring problem, the tip does not provide a reason to build a tool just for enjoyment.
Theo's enjoyment does not establish productivity gains
Theo is reporting his own enjoyment and the feedback loop he experiences while building these pieces. He does not present a productivity measurement showing what a similar experiment will deliver for you. The practice still offers a small, reversible way to address a recurring workaround without assuming the same subjective reward.