Charley Cunningham
16 posts

Charley Cunningham
@mureytasroc
Rust Evangelist
San Francisco, CA Katılım Kasım 2012
81 Takip Edilen15 Takipçiler
Charley Cunningham retweetledi

I've left OpenAI and the Codex team to build Blackstar: A new hardware company building the future of human-computer interaction.
We believe that software is solved. Building apps is now easy, but the next meaningful improvement in human-AI communication requires changing the OS & hardware. That's why we're building a new device entirely.
I'm also excited to announce our $12m seed round led by @AbstractVC, with participation from @naval, @SVAngel, @chapterone, and Timeless, among other amazing angels who've supported us from the old Alex days.

English

Codex has a very annoying bug now that your queue question sometimes would just cut in earlier and mess up everything @thsottiaux
Worth a reset?
English

@thsottiaux bug with queueing up msgs / steers while /compact is running. it hangs infinitely. to repro, do /compact, then queue up a bunch of steers. it hangs indefinitely and only way out is to send a new msg and esc to interrupt
English

@omarsar0 All that just to be mogged by Claude Code using grep
English

AGENTS dot md files don't scale beyond modest codebases.
Lots of discussions on this lately.
If you're building serious software with Claude Code or any agentic tool, a single AGENTS dot md will eventually fail you. This paper shows what comes next.
A 1,000-line prototype can be fully described in a single prompt. A 100,000-line system cannot. The AI must be told, repeatedly and reliably, how the project works, what patterns to follow, and what mistakes to avoid.
Single-file manifests hit a ceiling fast.
This new paper, Codified Context, documents a three-tier infrastructure built during real development of a 108,000-line C# distributed system across 283 sessions over 70 days.
The system uses a three-tier memory architecture: a hot-memory constitution (660 lines, always loaded), 19 specialized domain-expert agents (9,300 lines total) invoked per task, and a cold-memory knowledge base of 34 specification documents (~16,250 lines) queried on demand via an MCP retrieval server.
Across 283 sessions, this produced 2,801 human prompts, 1,197 agent invocations, and 16,522 autonomous agent turns, roughly 6 autonomous turns per human prompt, with a knowledge-to-code ratio of 24.2%.
Crucially, none of it was designed upfront: each new agent and specification emerged from a real failure, a recurring bug, an architectural mistake, a convention forgotten, and was codified so it could never require re-explanation again, turning documentation into load-bearing infrastructure that agents depend on as memory, not reference.
Paper: arxiv.org/abs/2602.20478
Learn to build effective AI agents in our academy: academy.dair.ai

English

@theo > Not the outcome I expected, better if I’m being totally honest
> Not the outcome, I expected better if I’m being totally honest
GIF
English

Charley Cunningham retweetledi

Every company and every person will one day have a magical library that personalizes their computing experience. We're the team building it.
Meet Pacific:
Pacific@pacificint_
English

@IterIntellectus ???? Its a pay plan not a lump sum wtf
English

Elon musk has now 1 trillion dollars in his bank account
That’s a thousand times $1 billion dollars!
He could give every single human on earth $1B and still be left with $992B!
Let that sink in
vittorio@IterIntellectus
ONE TRILLION DOLLARS
English

@0xMasonH Check out this PR (super short): github.com/fmhall/Petrosi…
English

Thanks for reading this epic chess saga. The code for the bot is open source, and can be found here: github.com/fmhall/Petrosi….
(15/15)
English

How my Reddit bot became ranked #2 out of 55k.
A story about chess, cheating, and the most passionate community on Reddit.
🧵 👇 (1/15)
English

@elonmusk Miners using renewables means that they are crowding out other potential users / driving up prices. They just use the cheapest power available. Mining also wastes GPUs which are very valuable to society (especially for ML). Move to proof of stake!
English












