Sureffi

Augmenting agents with cyberware
C# is not like the other languages
Founder, dev, dropout (neovim + 3x Arch btw)

[Sureffi @Sureffi]

A ```dot fence in Claude Code becomes a drawing, in place, as the reply streams.

To install:
/plugin marketplace add sureffi/drawer
/plugin install drawer@drawer

https://t.co/tX0oTkOynZ

[Sureffi @Sureffi]

@dexhorthy do you have anything public about those three grey boxes you could point me at? genuinely curious where the frontier is at here

[Thariq @trq212]

working at a higher level of abstraction has always required understanding the lower level ones, I don't think coding agents changes this

[Sureffi @Sureffi]

@dexhorthy which part is the "kinda"?

[Sureffi @Sureffi]

@dexhorthy the bars were not filled

[Sureffi @Sureffi]

Swap the reader, and the same doc can go from compact to redundant, or from compact to cryptic.

[Sureffi @Sureffi]

A doc's meaning is relative to its interpreter, so there's no true "good" documentation. There's only documentation that's efficient relative to *this* reader.

[Sureffi @Sureffi]

@0yield Imagine multiplayer combat surf in the browser. I’d play that

[Sureffi @Sureffi]

@wieslawsoltes But why tho?

[Sureffi @Sureffi]

@unixterminal I don’t want to to spend my tokens on job applications. Kinda dont want to spend my time either :D what do?

[Sherwood @shcallaway]

We deleted 800k lines of unit tests from the @sazabi monorepo earlier this week.

Three hypotheses:

1- Unit tests "lock-in" slop code by making it hard to change or remove

2- Unit tests slow down dev cycles by making your agent do more work locally and by increasing CI times

3- Unit tests waste money by making your agent generate more tokens and by adding useless CI runs

Will report back soon with the results!

[Sureffi @Sureffi]

Yk those docs that frontier models like to write for your codebases? It's making them worse btw, measurably so

[Sureffi @Sureffi]

@Sauers_ Those messages go through unix sockets. You can send them directly to sessions yourself. Makes no sense to me why hide them if you can read them anyway lmao

[Sureffi @Sureffi]

They keep changing little things that subtly mess with your work. Sometimes i find new skills i never wanted. Then we spend time with claude to figure out how to disable because the harness is already bloated enough, only for something else to come back a few updates later.

Another example is the idle notification that claude now subscribes to when they send messages to other sessions. It used to be enough that the session would report back, and i never had any issues with that. Now the harness slops my context on every message. Claude even got annoyed by it and saved a note in memory to not do that.

I dont know how to disable that, and i dont think claude does either — as is with a lot of things. The mods and hooks and whatever are cool concepts, but ”just prompt it” doesnt really work when the first step is to launch a dedicated research agent, fill 100k tokens of context, and then you might get something that works half the time. You go through all of that, just so they can break it again in the next update.

[Alex Christou @alexchristou_]

Claude Code just shipped mods...

I made one that connects you to a multiplayer MW2 Quick scoping lobby

Every other player in the match is a person waiting for their Claude to finish.

[Sureffi @Sureffi]

@seflless @dexhorthy I’m actually running this exact bench right now. It is quite fun indeed!

[Grady Booch @Grady_Booch]

In the face of generative AI coding agents, the role of human-led architecture becomes even more important in the crafting of software-intensive systems.

All architecture is design, but not all design is architecture. Architecture represents the set of significant design decisions that shape the form and function of a system, where significant as measured by cost of change.

As such, remember that coding agents are reasonably good at transforming prompts into code, but they are not particularly good at making decisions that balance the static and dynamic forces on a system in order to yield a reasonably optimal solution: this is where human judgment, human understanding, human knowledge, and human experience shine.