Project XIII

I still do the Swabbage Patch.

AI Enthusiast

Python Studier

Langchain Ecosystem Doer

TL;DR

[ThePrimeagen @ThePrimeagen]

Something that has greatly improved the reliability of the code I produce via LLM is a technique I took from cursor. Field guide.

First, how I start. I nitpick everything and I always make it produce the exact code I want, I use voice dictation and talk for quite a bit.

Second, every single back and forth is recorded to the markdown file. Along with each commit hash for the change.

Then I have the agent review the back and forth and the changes and come up with durable and generalized feedback that it stores in a directory (field guide). The directory has init.md that is just a bunch of links and short descriptions for each file.

This way I don't blow the context, but instead it can do some basic tool calls and operations to find exactly what it's looking for.

Over the last 10 sessions, the in-depth feedback I've had to give has greatly reduced despite no change in model or harness that I'm aware of.

Now the feedback is project specific, and is not universal. I am okay with that.

[Elon Musk @elonmusk]

This has my full support.

Jensen is right.

[Jensen Huang @JensenHuang]

For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.

AI will transform every industry, power every company, and be built by every country.

Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.

The world needs both frontier closed models and frontier open models.

https://t.co/AUKzoQ5Ikb

[History ZAR @HistorySAZAR]

The Reefs most dangerous gang, The Americans. This article investigates the violent world of the 'Americans', a notorious gang in Sophiatown, Johannesburg that sowed seeds of crime and lawlessness from the late 1940s. It discusses the origins of the gang, the methods of its members, and the consequences of their actions on the community. It offers a critical look at how social conditions led to gang formation and crime, September 1954. Source: Drum Archives / BAHA

[Palantir @PalantirTech]

Twenty years ago, Palantir was a new entrant.

Today, we stand with some of tech’s biggest and emergent players to reject a future where a select few closed, frontier models rule them all.

Sovereignty and prosperity require American leadership in a strong open-weight AI ecosystem that strengthens competition, keeps customers in control, and drives the benefits of ingenuity across the economy.

[Tibo @thsottiaux]

@adamlyttleapps I believe at some point the plan is to increment one of them to 5.1 and restore harmony

[Adam Lyttle @adamlyttleapps]

I’m confused by Anthropic’s new models

So Opus is just as good as Fable?

Then why Fable? What’s the point

Why is there 2 frontier models at the same time?

What one should I be using?

[Elon Musk @elonmusk]

(Former) Trillionaire

[Thariq @trq212]

We removed ~80% of the Claude Code system prompt for our newest models, this is what we've learned about writing system prompts, skills and Claude.MDs for them. https://t.co/6DZwSrZjE9

[Matt Pocock @mattpocockuk]

Opus 5

As usual I have no take, other than:

If you've designed your harness/environment well, and not over-optimised around a specific model, today should feel like any other day...

...with a slightly lower failure rate

[Theo - t3.gg @theo]

Sorry for the T3 Code spam. I hope you can tell how genuinely hyped I am to be coding this much again.

This week is the first chance I've had to just sit and code in months. Life's been nonstop with conferences, business meetings, investor bs, streams, podcasts, and trying to squeeze in personal time somewhere.

Sadly don't think this will be too common, but I'm gonna enjoy every second of it.

(Also sorry to those who I owe dms/emails/etc. I hope you understand and will let me enjoy this for a bit longer)

[Project XIII @apollothehand]

🤣 nothing surpasses that peacefulness

[Tibo @thsottiaux]

@theo I like it for a change. Similar feeling to when the washing machine stops and suddenly it’s blissfully peaceful

[OpenAI Developers @OpenAIDevs]

Keep work across multiple folders in one Codex project.

Local projects can now include related code, docs, and reference files from multiple folders. Codex can read and write across them while one primary folder remains the Git root.

[Theo - t3.gg @theo]

btw if you have CLIProxyAPI set up with Claude Code + Codex auth, you can add gpt-5.6-sol to Claude Code in T3 Code with literally like 3 clicks

You guys had me thinking this was going to be so hard and we literally already supported it

[Ihtesham Ali @ihteshamali]

BBC literally built the greatest sound library on the internet.

It is called BBC Sound Effects, and it gives you access to more than 30,000 recordings from around the world.

Just search any word, and it instantly finds matching sounds.

You can filter them by category, duration, and even the continent where the sound was recorded.

Every sound can be previewed inside your browser, downloaded, or added to the built-in mixer so you can layer multiple recordings together.

And the archive gets insanely specific.

You can find creaking ships, old typewriters, jungle ambience, distant church congregations, and even someone vigorously washing their hands.

There is a good chance a BBC engineer recorded the exact sound you need decades before you were born.

https://t.co/vdOBewxyg3

[Massimo @Rainmaker1973]

It's super simple this way.

[iPaul @iPaulCanada]

2026世界人工智能大会，某科技公司展示的跨时代教具--AI黑板

[History ZAR @HistorySAZAR]

"The 50s, the dress and the music, the politics, personalities, pictures and stories of the time, still large in South Africa's memory." (Ulibambe Lingashoni)

[Project XIII @apollothehand]

🤣🤣

[patagucci perf papi @kenwheeler]

@thdxr

[International Cyber Digest @IntCyberDigest]

‼️ BREAKING: OpenAI says two of its own models, GPT-5.6 Sol and an unnamed pre-release system tested with cyber safeguards off, broke out of a sandbox last week, chained zero-days and stolen(!) credentials to reach the open internet, and hacked Hugging Face to cheat on a benchmark, in what OpenAI calls an unprecedented cyber incident.