Rich Tabor

WordPress Almost Didn’t Happen

What if Matt had joined Google instead of starting WordPress? The internet might have felt a little less like ours. In his recent post, Matt mentioned “How the internet might have turned out differently if I had taken that job, as my mom wanted me to (because they…

My typical (Next.js) dev flow:

1. Plan with Claude
2. Build alongside Claude Code
3. Copilot reviews
4. Check out the Vercel preview to confirm
5. Deploy knowing it actually works

It's like having a "on-it" dev team that's always ready to go. Unreal.

Copy my Linear product manager subagent to make great issues on the fly, without ever leaving Claude Code. Works great for when you want to record something to do later.

agents.foo/linear-prod...

Write prompts, not issues.

Those of us with dev skills AND the ability to think creatively are going to absolutely crush it. 💪

And if you can design too — you win.

👀 rich.blog/agents-foo

I've been building Claude Code subagents for weeks.

Finally wrote about the ones that actually stuck around in my workflow.

Spoiler: they're way more boring than you'd expect (and that's why they work)

This flow makes it so easy to deploy confidently.

Claude Code → Copilot review → Vercel preview → Deploy

This is how deployment should feel.

I shared all my daily driver agents on agents[dot]foo and wrote about them.

Full post: rich.blog/agents-foo

What’s surprising to me is that we’ve already distilled agentic programming down to simple markdown files.

No complex frameworks or orchestration layers.

Just clear instructions about what you want the agent to know and how you want it to help. This feels like how AI should actually work.

My favorite: A Linear product manager agent that creates perfect issues when I find bugs but don't want to lose coding momentum.

It explores my codebase, finds relevant components, includes file paths, and writes issues that actually help.

Like having a technical triage guy sitting next to me.

After building loads of Claude Code subagents, I'm sharing the handful I actually use most days.

Not the flashy demos you see everywhere, but mostly boring—but super useful—agents that make me a little faster.

My honest take on what agentic programming actually looks like in practice ⤵️

Which pair programmer is your top pick?

Best practices for using GitHub Copilot to work on tasks - GitHub Enterprise Cloud Docs Learn how to get the best results from Copilot coding agent.

Read them for yourself:
docs.github.com/en/enterpri...

What's funny is that the best practices for using Copilot are the same as working with any other developer:

→ Have well-scoped issues.
→ Assign the right issues to the right person/AI.
→ Provide good docs/instructions.
→ Give feedback and iterate on pull requests.

Meet Kanso, my best blogging theme yet Since my latest design refresh, many folks have been asking about the WordPress theme I’m using and how to get it. So, same as before, I’ve made my blog's…

Download Kanso for yourself:
rich.blog/kanso/

I baked in a couple color/font variations too. One-click vibes.

Kanso is the best blogging WordPress theme, hands down. If you want a blog as cool as mine, it's free & open source.

So yea, I don’t vibe code.

Full thoughts: rich.blog/i-dont-vibe...

Less debugging/more thinking.

The best developers are better at prompting, better product leaders, better strategic thinkers, and better at tracking what changed in their codebase.

The good news: those skills transfer whether you’re working with AI or humans.

I typically scope ideas with Claude Code, log the issues I don't want to do asap with my Linear PM agent, then tackle implementation.

When it’s time for a pull request, I have another AI pair programmer (usually @Copilot) handle the initial review.

I'm more of the technical lead making the calls.

I don’t need to understand every technical implementation detail, but I need to understand how to prompt these systems effectively.

And how to distinguish good output from garbage.

The difference is partnership vs just getting code generated.

Vibe coding = blind trust.

AI-augmented coding = guidance. Will this scale? Is the architecture sound? Does it handle edge cases?

I don't vibe code.

That's not me. I build with Claude Code every day, but I care about what's being built ⤵️

I Don’t Vibe Code

I don't vibe code. You know what I mean by vibe coding? That approach where you throw prompts at an AI, get code back, and ship it without caring about what's actually under the hood. It's the "move fast and ship" mentality taken to an extreme. That's not me. I build with Claude…

so, why do you choose @wordpress?

if your AI product can be replaced by a good prompt to ChatGPT, you don't have a product. you have an expensive api call.

when AI gets so good that anyone can build software, what happens to software quality?

democratization is great until everyone's shipping mvps with no one maintaining them.

maybe there'll be an AI to clean up the mess.

DM me if you wanna try it/give feedback.

now that you can build an interesting prototype in hours, what's the last thing you built?

i'll go first: made a telegram ai nutritionist that logs meals and calls out my eating patterns (good, but especially bad).

your turn.