I made nawc because I loved spec-driven development with agents, but hated how inconsistent markdown is at describing intent. I needed a more streamlined approach, where intent is captured with more deterministic probes that are easy to refine and iterate on.
The solution is NAWC - an extensible HTML-based notebook that lives in your codebase. It has built-in plugins that add custom components that can reference live source and run interactive examples and tests right inside the notebook. It also ships with a set of agent skills, inspired by Matt Pocock's skill collection, designed to help you refine a rough idea into a polished TDD-ready design through prototypes you can see and interact with.
I'd especially love to hear some feedback from people who find it difficult to keep up with the code when working with AI agents.
I'd love for this tool to be a way to help people have a deeper understanding of the code they (have their agents) write, and would love to know what else I can add to help support that goal.
NAWC is open sourced under an AGPL license over at: https://github.com/Narwhster/nawc
You can read more about my process coming up with the idea for NAWC, as well as future projects that stem from it over at: https://narwhster.com/notes/nawc-and-nit/building-nawc-and-n...