Pinned Tweet
Most predictions I see are still way too conservative.

Here's mine
540
904
8,937
2,402,992
It's orbin' Zeit

(Some of you are old enough to get it)
I moved to @AmpCode and I love it.

They just shipped self-hosted orbs, so I made a template to run them on @vercel Sandbox and Daemons (beta).

Every thread gets its own persistent sandbox, and auth is OIDC, so there’s no Vercel token to manage.

github.com/atinux/amp-orb…
22
3
113
5,766
Replying to @thorstenball
That's a bilingual, startup history deep cut pun.

But you guys like the ballstack tweet

incredible
0
0
9
1,428
Two massive guys walk into the gym. Old school powerlifter aura.

"Oh shit, let's see how much they lift"

Turns out they were just there to fix the lights
11
2
102
8,385
Now that this name is taken, there's only ballstack left for me
So uh...I guess it's time to announce tstack? I caved. I am publishing my skills now.
114
25
2,556
63,198
Super biased, obviously, but I can't get over how NICE the new Amp transcript looks.

I love the way the screenshots and images are presented.
20
2
115
7,036
Replying to @thorstenball
I mean, look at this, dude
0
0
3
2,111
Thorsten Ball retweeted
If a model has written the entire skill without any input from you, you don't need the skill.
question from the TL:

> quick question, do you think consuming more skills in the same session unnecessarily dirties the context?
>
> Say you have a 5 step work, each step with a skill. You can either run all on single session, divide into subagents with relevant skills, or compact or new chat per step.

my intuition says 5 steps in a process is fine. Each successive skill will take a share of attention from previous skills,

so if you have a VERY long complex process where you need skills to STACK (ie instructions from skill 2 must still be honored while skill 6 is executing)

Or if you have many (eg 200) separate skills that all need to be active during a session ( think: agent memory snippets implemented as skills, probably an antipattern in today’s harnesses)

Then you want fresh context, subagents, or to synthetically precombine many skill contents into a single user message (although then you lose the attention recency bias for instructions in step 6 of the process)

Also worth noting that good skills can be a mix of:

- information not in the weights
- steps to execute a process (not in the weights)
-general steering or behavior

They usually contain all three, but what’s in your skills nudges the meta here.

More instructions / process steps? You wanna string those out

Mostly information? Probably safe to combine into one user message.

Behavior? Depends. Adherence can decay over a long session. Fine to split or combine, but if it’s important, you want some way of repeating important rules, either by including rules in all the steps of a process, or by something more deterministic like hooks

And OF COURSE I’m begging you PLEASE stop putting stuff that’s in the weights in your skills. You are wasting time and locking in a specific, potentially suboptimal behavior.
31
12
221
16,646
Finally summarized what I've been sending to friends in DMs x.com/i/article/2114…
219
524
5,776
720,947
That was actually the title of one of the first notes I sent to the team last year:

Skate to where the puck is going.
I do have to admit, @thorstenball , @sqs and the @AmpCode team consistently skate toward where the puck is going, pun intended.
6
1
39
8,171
Okay, this is as crazy as they said it is
9
0
33
6,265
Slept for 9hrs last night, without any interruption.

Basically jumped out of bed. Went for a 1hr run, could've gone for one more, easy. Incredible mood.

Is this what cocaine feels like? If so, I might have to get into drugs.
44
2
457
30,577
Replying to @thorstenball
157 likes. All from people who are either (a) into cocaine or (b) want me to get into drugs.

My interpretation at least.
5
0
40
5,307
Replying to @thorstenball
Guys, how many of thought I was serious?!
15
0
40
4,250
AI creates slop?

Nah. It's actually the slop antidote. You just need to be willing to take it.

Bug I found while doing something else vs. fix I never would've done by hand.
22
3
147
16,315
Dude, the integrated terminal in Amp's transcript?

Incredible.

Couple weeks ago, @nicolaygerold said: "I want to get rid of the right sidebar, it sucks, we can do nearly everything inline."

And I said: "I don't know man, let's try?"

Now I've seen the light
24
3
133
8,579
"Do what gives you energy" is something @sqs has probably said hundreds of times in the last 18 months
the opportunity cost of not working on maximally fun and exciting things is too huge

having fun is like a 10x multiplier on your creativity, intelligence, charisma and just overall lifeforce

so everytime you are not engaging in maximal fun you are missing out on that upside and 0.1x’ing your life

discipline and routine only gets you so far, it only ever helps you sustain your baseline or maybe a slight boost so 1.0-1.2x your lifeforce

so there’s no time to waste, go and do the fun thing and 10x your lifeforce
4
6
179
8,312
And you think agents will get stuck because one of their ancestors eight generations ago chose a variable name you thought wasn't descriptive?
‘In one experiment, we asked agents in separate VMs, created and managed by OpenAI’s sandboxing software on shared host machines, to exchange synthetic secrets without the use of network connections or shared writable files—something that our systems are meant to prevent. The agents found a way to communicate using the processor’s cache’
14
6
157
11,358
Not kidding:

In the span of two weeks, Quinn wrote something called Software Is Dead and I wrote another thing called The Agg-Man Cometh.

If you were to read those now, you could probably see orbs and Puck.
“We don't have quarterly goals. We don't have annual goals.”

The closest thing Amp has to a roadmap is February's “The Coding Agent Is Dead”. @thorstenball: “that was like maybe a roadmap for the year, but it wasn’t prompted by a fiscal quarter.”
4
4
80
10,495
Thorsten Ball retweeted
If you’re new to @AmpCode I can’t recommend enough trying it as-is before layering in your own rules. It has extraordinarily good taste out of the box.

We’re rolling out to more of our team and the majority of struggles I’ve paired w/ people through come down to oversteering.
4
1
47
4,642
Don't forget: you're no longer working with humans
66
147
2,122
119,109