PWV Portfolio Feed

Scores aren't pass/fail. They're a baseline and a direction.

Run it honestly, see where agents struggle, and improve from there.

It's open source under MIT, published as an npm package, and runnable anywhere.

Created by Netlify, built in the open with founding contributors who are running it against their own platforms and publishing the results.

AXIS (Agent Experience Index Score) is a CLI tool that runs real scenarios against a live endpoint and scores it across dimensions like discoverability, error handling, and response clarity.

A structured, repeatable read on how agents actually experience your service.

AI agents don't just call models. They call your API. They authenticate, hit errors, parse responses, and decide whether to keep going.

Most services were never built for that.

Today, we're launching AXIS: an open source way to measure how well your service works with AI agents.

I think there’s a clear and useful distinction between things that do the stated job and the moral valence and second degree effects of those things, the difference between good as in moral and good as in functional

Ran hundreds of version control benchmarks comparing Claude vs Codex in usage of the `but` CLI.

Claude performs worse, and I think we finally found the root cause 🥁...

The `but status` output contains characters 
◐ & ● - for some reason Claude overthinks what they mean...

PlanetScale's fault tolerance is built on straightforward principles and architectures. The challenge is in the execution.

Here are the principles we follow to keep our systems reliable:

I’m sure someone suggested this before, but why didn’t “bluep” (pronounced BLOOP) catch on as Bluesky’s “tweet” verb?

what if when you hit Post the app made a little bluep sound

Let's goo!