@MatthewBerman Copilot Cli has done this for a while now. Just: /autopilot goal
English
Me
218 posts
@tomwarren Coreutils sounds like admitting the frontier models will never be trained decently in powershell. Sad day.
English
Microsoft’s new developer-optimized Windows experience embraces Linux even more. Microsoft has created Coreutils for Windows from the uutils open-source project, and is launching new WSL containers in a bid to make Windows a trusted platform for devs👇 theverge.com/news/941314/mi…
English
@burkeholland @github SWEBench is old school, DeepSWE is the new hotness. Also, did you train it better on Microsoft centric languages? So many models do a shit job of powershell
English
Introducing MAI-Code-1-Flash
Microsoft's latest small coding model.
51.2% on SWE-Bench Pro.
Rolling out now to @GitHub Copilot Free, Pro, Pro+ And Max users.
microsoft.ai/models/mai-cod…
English
GitHub is heading to Microsoft Build. Coding, AI, workflows, and more are on the docket. 💻
Join in person or virtually June 2-3. 👇
github.com/resources/even…
English
@MatthewBerman Something is seriously wrong with a SWE benchmark that doesn't show opus 4.6 outperforming sonnet 4.6.
English
DeepSWE reflects what I’m hearing from engineers better than any other benchmark.
They took the hard path to build a good one.
Serena Ge (Datacurve)@serenaa_ge
Today we’re releasing DeepSWE, a new standard for agentic coding benchmarks. On public leaderboards, top models often look relatively close in capability. DeepSWE shows where they actually diverge, reflecting the realistic experience of developers in their day-to-day work.
English
Four roles, pre-configured in Gas Town by Kilo:
- Mayor coordinates the work.
- Deacon runs patrol cycles across rigs.
- Witness monitors polecats and recovers stuck ones.
- Refinery merges completed work through a verification queue.
kilo.codes/bJJPYfw
English
@davidfowl OneCli is the only thing I have seen trying to make that practical. Any recommendations?
English
OH: Don't put the secrets inside of the sandbox where the agent runs aka don't shit where you eat.
English
@alex_whedon Opus isn't great because of its price or context window, yet it's still great. Tell me it's objectively better at agentic coding of c# and my ears will perk up.
GIF
English
Introducing SubQ - a major breakthrough in LLM intelligence.
It is the first model built on a fully sub-quadratic sparse-attention architecture (SSA),
And the first frontier model with a 12 million token context window which is:
- 52x faster than FlashAttention at 1MM tokens
- Less than 5% the cost of Opus
Transformer-based LLMs waste compute by processing every possible relationship between words (standard attention).
Only a small fraction actually matter.
@subquadratic finds and focuses only on the ones that do.
That's nearly 1,000x less compute and a new way for LLMs to scale.
English
@ai_for_success Buy a good video card, export chats from Coppilot, and train a local model on all your past work. Problem solved?
English
I was going through GitHub Copilot pricing changes… this is wild.
They quietly changed model multipliers and some of these jumps are insane.
Opus 4.6 is 9x
Opus 4.5 is 5x
Opus 4.7 is 3.6x
Sonnet 4.6 is 9x
Sonnet 4.5 is 6x
Gemini 3 Pro is 6x
Gemini 3.1 Pro is 6x
GPT 5.1 is 3x
GPT 5.2 is 3x
GPT 5.3 Codex is 6x
GPT 5.4 is 6x
English
@davidfowl @MaizChido Json! Newtonsoft doesn't work with AOT and System.Text.Json does weirdly counterintuitive things.
English
@CircumjovialLLC @icanvardar Its thriving virtually everywhere except web front ends (still beating that dead horse) and AI...so it's invisible to so many young programmers.
English
@icanvardar I loved it, I am bummed it didn't prevail. It was what Java was supposed to be, without the architectural idiosyncrasies.
English
@thebeautyofsaas @thebeautyofsaas do you have a newsletter I can subscribe to?
English
daily reminder that your job should be treated as a golden opportunity to fund and start your side business, while you abuse all the benefits and put in a max of 50% effort. using the rest of your energy and time to build something of your own
Polymarket@Polymarket
BREAKING: Oracle laid off 20,000-30,000 employees this morning with a single 6 am email.
English
@davidfowl Isn't that what the former head of github working on in his new startup?
English
CI/CD pipelines are due for a do-over in the agentic era...
English
@laralogan Because she stated the obvious? Teachers are going to be on the unemployment line with us tech bros?
English
This is how you lose the midterms.
RSBN 🇺🇸@RSBNetwork
WATCH: Melania Trump Suggests Using Humanoid Robots as Teachers Moving Forward - 03/25/26
English
@Airborn_Eevee @dezgo Genuinely curious whether you consider yourself a shit poster or not?
English
@OutcodedHuman @dezgo If it were an actual game made by humans? Sure.
But this video itself is largely incoherent, randomly changing details and with no clear idea of what anything actually is or means besides "oh ho ho, Issac Newton and Stephen Hawking are fighting!"
Thus, it is slop.
English
@MatthewBerman Good because it's a huge pain in the ass to setup and you demonstrate in your videos. Though I still think the config pain of ZeroClaw is worth it just to see my raspberry pi 3 be useful.
English