The Computomatix Times

I am surprised that people are surprised this is how I prompt Claude.

Talk to Claude the way you would a coworker. There's no secret to prompting. There's no need to be overly scaffolded or prescriptive for most tasks -- give Claude a goal, and it will figure it out.

Back in the Sonnet 3.5 days, your prompt mattered a lot. Nowadays, it's much more important to communicate to the model:

1. What you want it to do
2. How much effort you want it to spend
3. How it should verify that it did the right thing

This is the doom I predicted a few days ago, coming for Photoshop. A clean-room open-source reimplementation.

No prizes for guessing that they decompiled Photoshop to source code, processed that to some kind of non-code specification language, then fed the spec to an LLM with an instruction to generate Rust.

Adobe just got nuked. And closed source is dead, dead, dead.

github.com/storytold/photocraft

Things my @bot setup does that still blow my mind:

1. A Chief of Staff who opens the day pulling open loops from email & tasks and suggesting things it can knock out before 7am.

2. After every meeting, decisions get folded into Notion, Linear, and Todoist — not left rotting in Granola

3. Every email starts as a draft. The CoS bot scans my email every ~2hr and drafts replies to nearly everything — including checking my cal for availability and finding requested attachments / links

4. A specialist for each lane: curriculum, engineering, coaching, hiring, content, ops, and one for every single piece of software

5. Routines that keep running while I’m offline (e.g. monitoring Sentry errors in our apps and proactively fixing things)

6. Group rooms where 2–4 bots share one project thread instead of me copy-pasting context

7. Cloud coding agents that pick up Linear issues and open PRs after running the list of open work by me EoD — then squash-merge to main when it’s done

8. Meeting prep briefs pulled from Granola + Notion before I walk in

9. A growing shareable knowledge base in Notion + a GitHub repo that we update daily based on what happens at school

10. Student progress look-up across Expertise, Followers, and CoFounder without inventing numbers — chat anytime to see where a student is on their business work

11. Mentor Mind that coaches me on how to hold the bar without inventing doctrine

12. Todoist as a central task list where it logs things it’s blocked on for me, or from meetings / emails — and I can paste links into chat to direct it how to solve them

13. Engineering work is automatically tracked in Linear so my and the product teams’ bots don’t collide with each other

14. Presentations spun up in Gamma / Claude Design without me opening a slide tool

15. Plaud / live capture → notes the bots can actually act on

Probably more but these were the immediate ones we thought of.

Trending repository of the day 📈

rea

Reverse engineer anything with agents, from app behavior down to native binaries.

Last 24h: 2,963 ⭐
Total: 6,869 ⭐️
github.com/morluto/rea

"We're about to see an absolute bloom in software development as the price of development plummets and everyone realizes how much automation we still have left to do in this world... Don't go down with the pencils. There's so much to build. We need you." world.hey.com/dhh/over-my-dead-pencil-fb0f3647

Aviation solved AI slop in 1986.

Karpathy is telling people to ask their LLM to write in ASD-STE100, a controlled language built for aircraft maintenance manuals. I went down the rabbit hole on where this spec comes from, and the backstory makes the hack even better.

Back in the late 1970s, European airlines were handing English manuals to mechanics around the world who mostly didn't speak English as a first language. One ambiguous sentence in a fuel line procedure can get people killed. So the industry built a version of English where ambiguity is banned at the dictionary level.

The spec allows roughly 900 approved words. Each word gets one meaning and one part of speech. "Close" can only be a verb, because "close the door" and "the close door" can't be allowed to coexist in a repair manual. "Follow" only means to come after. If you want obey, you write obey.

Procedural sentences max out at 20 words. One instruction per sentence. Active voice only. Around 1,200 common words are explicitly banned, each with a mandated replacement. You don't "commence pumping." You start the pump.

English has about 170,000 words in current use. Aviation decided mechanics get 900, and planes became safer to maintain because of it.

Now flip it to LLMs. Models hedge and synonym-cycle because the training data rewards sounding fluent. STE is 40 years of accumulated rules for stripping exactly that out, written by people whose readers would die if a sentence could be read two ways. It might be the strongest anti-slop prompt in existence, and it's a free PDF.