Using Jev to classify and triage beads quickly and efficiently compared to an LLM

[harrymunro]

As I've moved into Gas Town like territory myself with multiple agents running overnight (also heavily inspired by Steve's Wheelhouse) and generating more beads than I can deliver with my customer interactions, one thing I struggle with is doing large sweeping triages on my beads.

For example I sometimes need to go through my backlog and categorise all beads which are blocked on a human, or all related to a particular feature, or the size of beads, or whether they are agent-ready. That kind of thing.

Previously I have leant on my favourite LLM for this.

Then Jev came along in the past day and I've realised that it's exactly what my beads needed.

So I ended up building a system today called beadsort.

It's a simple concept: a sidecar CLI which labels beads according to your requirements.

I'm hoping that it proves an unlock for extra efficiency in our software factories :)

Here's the github link - MIT licenced: https://github.com/harrymunro/beadsort