Mighty Byte

Engineering leader, functional programmer, student of leadership, devotee of skills acquisition who slacklines and rock climbs. It's all about balance!

[Mighty Byte @mightybyte]

@cohere Models can only do these things if the harness that is running them allows it. If we really want to increase our confidence, the harness layer is the place we need to be focusing on.

[Mighty Byte @mightybyte]

@ShaneMac Managing permissions and controlling what agents can do is the wrong way to think about it. You need to give agents only what they need to do a very specific job. Then they can't go wrong.

[Mighty Byte @mightybyte]

He simply pursues his vision of excellence through whatever he is doing, and leaves others to determine whether he is working or playing. To himself, he always appears to be doing both.
- L.P. Jacks

[Mighty Byte @mightybyte]

A master in the art of living draws no sharp distinction between his work and his play; his labor and his leisure; his mind and his body; his education and his recreation. He hardly knows which is which.

[Mighty Byte @mightybyte]

Nice agent.md example designed to help agents generate better production-ready code:
https://t.co/3FxsNMZ9ww

[Mighty Byte @mightybyte]

🤣 Hilariously easy to uncover

[Mighty Byte @mightybyte]

Alternatively, if someone knows about any better ones out there, I'm totally interested in hearing about them. We just need some standard...any standard. I don't care which one other than that it shouldn't have names/artifacts/etc from a particular company / project..

[Mighty Byte @mightybyte]

More people should use this spec for defining agents that ship with a project for the purpose of doing work on the project.

https://t.co/ZPoQKtXHO7

[Mighty Byte @mightybyte]

Current AI has very little ability to discover its own truth. It only mirrors human "truths".

[Mighty Byte @mightybyte]

Your job to make effective use of AI is to figure out how to break down your problem into pieces small enough to fit into a single context window. I like the term "context engineering" to describe what we are fundamentally doing.

[Mighty Byte @mightybyte]

Someone asks, "At what point do you create multiple agents?"

Any time you have a task that can't be done within a single context window of your model, you should consider how you might be able to define agents to do a smaller portion of the work.

[Mighty Byte @mightybyte]

The stuff at the beginning has significant impact on the behavior all through a single session of conversation. The stuff at the end is stuff you didn't realize you needed to know at the beginning.

[Mighty Byte @mightybyte]

“With agents, you can have fresh minds on demand, for the first time in history.”

https://t.co/wT0qztNtzz

[Mighty Byte @mightybyte]

One of the most OP AI techniques I know of is:

Instead of asking AI to do X, ask it to make you a script to do X. Then you use the script to do the task instead of AI.

This idea is less intuitive for non-software people, but it's extremely useful and easier than it seems.

[Mighty Byte @mightybyte]

Excellent analogy

[Charly Wargnier ♨️ @DataChaz]

Vibe coders’ daily struggles, summed up in 60 seconds 😁

[Mighty Byte @mightybyte]

“The hardest things are usually the best. If it’s easy, leave it to somebody else.”
— Tom Platz