Scott Watermasysk's Development Blog

Engineers love to argue about what “is the way.”

Of all the fun things I have built with AI this year, the ones I use the most are Muv(.ing) and Stooges(.dev).

My dream job at Anthropic: Ensure authentication code emails are sent within 30 seconds MCP auth that lasts more than 24 hours You could fire me after both PRs are merged.

Building with AI doesn’t necessarily mean you have to go faster.

Lewp is my answer to the issues of managing ports for local development, which got infinitely more complicated with agents and worktrees.

I could not figure out which open-source Rails app would be the easiest to showcase using Lewp with PG and Redis, so I gave the Codex App a simple goal to build an RSS reader and walked away.

Claude Design is a good pre-prototype tool, but it could be great if Anthropic invested in it.

A quick recommendation: prototype more.

For anyone who has worked locally on multiple applications, you quickly learn that port conflicts become a nuisance.

I wonder if “shared prompting” could help counter the deep code understanding that AI coding has eroded: two or more people working together with a single prompt/workflow.

We should be prototyping more.

My model breakdown: Fable - the only reason I use Anthropic models for anything.

My typical workflow for larger tasks: research → interview me (grilling session) → generate a spec → tickets → implementation There is some iteration and review along the way.

Lewp was upgraded to 0.2.0 - It now handles releasing of ports when the directory no longer exists (ie, you deleted the worktree or Stooges workspace before you released the ports)

Issue #3 of AI AND Me is out Herdr over Solo, don’t pick a single model, background agents, skills, and my new AI sleeping giant - CloudFlare https://scottw.com/ai-and-me-issue-2026-3/

I have come to really trust my code-quorum skill’s recommendations.

Between life and the constant changes in AI engineering, this post has been nearly a month in the making.

Herdr + Pi is a very flexible combination.

Automatically create @herdrdev workspaces when you add a workspace to http://Stooges.dev - https://gist.github.com/scottwater/9467447340c62875d955b50d16e9e19c

I swear, if I see one more video or post that says, “just ask your agent what three things to build next”… Never do this.

If you’re ever curious how a new tool might fit into your workflow, give @mattpocockuk’s Teach skill a whirl.

Not a bad week.

I took a few minutes to finally connect Stooges → Lewp → Herdr.

Impeccable by @pbakaus has been extremely helpful in spotting UI issues in my various projects (both agent-driven and self-inflicted).

Another reminder that you can do far more with AI than build features and apps nobody asked for.

PlanetScale - It would be great if this remembered my previous settings or had some sort of templates.

There are people on X talking about canceling all of their SaaS subscriptions and building their own personal versions of those services.

If you use Pi and SoloTerm, I made quite a few fixes and updates to the pi-soloterm-tools extension (basically the SoloTerm MCP for Pi).

I have been using @mattpocockuk’s skills quite a bit the last few days, and batch grill me has made a massive improvement.

Debugging with the Sentry MCP/CLI along with the PlanetScale MCP is a massive superpower.

All I want from the next version of MacOS is one single security prompt per application.

Slack needs a “move message to thread” feature or a warning, “are you sure you didn’t meant to reply to. thread” warnig.

Not having Fable available in Pi as part of my subscription is one of the more frustrating parts of my workflow.

KickoffLabs has been running on PlanetScale for about four weeks now, and the process has been super smooth.

One of the biggest mistakes you can make today with AI is going all in on a single harness, tool, provider, etc.

I set up the pi-grok-cli in Pi, which provides adds grok and the rest of the cursor models/etc to Pi.

Let’s go, SoloTerm style.

When coding with AI, take advantage of the opportunity.

The only model decision I know to be 100% accurate is this: don’t rely solely on a single model.

You can create a lot of new things with AI.

Superpowers with Fable has been interesting.

Updated TestFlight build for http://muv.ing/ went out this morning.

Switched my voice-to-text app to FluidVoice this week.

It is surprisingly difficult to get data out of HealthKit and have it match what Apple presents.

I went all in with @sentry this week.

I built a new thing, lewp (loop).

A simple way to make using various models easier with SoloTerm - Just create new agents for each model.

GML 5.2 via @AmpCode has been great.

Scratchpads in @aarondfrancis Solo are a hidden gem.

I hate when console logging (puts in Ruby) is visible while writing specs.