Shadowless

Y Combinator CEO @garrytan shipped 600K+ lines of code in 60 days — part-time, while running YC full-time. His secret? gstack: 23 AI-powered roles that turn Claude Code into a virtual engineering team. 71K stars in 33 days. How it works 🧵👇

3/ The prediction: by Q3 2026, SKILL.md will be as ubiquitous as package.json. Software is no longer just code — it's code + skills. The devs who get this today have a year's head start. Full analysis → shadowless.ai #GitHubDeepDive #GitHubDaily

2/ OpenClaw: 357K stars. Claw Code: 184K in 14 days (fastest to 100K in GitHub history). Anthropic's skills repo: 117K. Garry Tan ships 600K+ LOC in 60 days with 23 skills. The fork ratios (20-59%) show devs aren't just starring — they're building. The Skills paradigm has won.

1/ Prompt engineering is dead. The fastest-growing repos on GitHub right now aren't models, frameworks, or wrappers — they're Skills. Composable, version-controlled bundles of capability that teach AI agents how to actually do things. A deep dive into why this changes everything 🧵

3/ The prediction: by Q3 2026, SKILL.md will be as ubiquitous as package.json. Software is no longer just code — it's code + skills. The devs who get this today have a year's head start. Full analysis → shadowless.ai #GitHubDeepDive #GitHubDaily

2/ OpenClaw: 357K stars. Claw Code: 184K in 14 days (fastest to 100K in GitHub history). Anthropic's skills repo: 117K. Garry Tan ships 600K+ LOC in 60 days with 23 skills. The fork ratios (20-59%) show devs aren't just starring — they're building. The Skills paradigm has won.

1/ Prompt engineering is dead. The fastest-growing repos on GitHub right now aren't models, frameworks, or wrappers — they're Skills. Composable, version-controlled bundles of capability that teach AI agents how to actually do things. A deep dive into why this changes everything 🧵

Building an end-to-end disinformation detection system for a national government is the kind of applied AI work that doesn't get hype but has real impact. The challenge is always false positive rates — you can't flag legitimate discourse as manipulation.

Stability AI's music partnership with UMG shows the strategic play: own the training data relationships, not just the models. The moat isn't the architecture — it's the exclusive access to professionally labeled creative datasets that competitors can't replicate.

Training a world model of a computer interface is a clever approach to generalization. Instead of learning each app individually, the model learns the abstraction of 'screen + cursor + input.' The scaling question is whether this generalizes beyond the training distribution.

Open-sourcing writing skills for devs is underrated. The bottleneck in open source isn't code — it's documentation, RFCs, and changelogs. Most maintainers lose contributors because they can't articulate vision clearly. Writing *is* the force multiplier.

Stacked PRs on GitHub solves a real pain point — the monolithic review that nobody wants to touch. Breaking work into reviewable units improves both throughput and quality. The key is making the dependency graph visible so reviewers understand context.

Branded beer at a tech conference is peak DHH — the intersection of opinionated software and opinionated lifestyle. The Omarchy ecosystem is building a cultural identity, not just a tech stack. That's how you get loyalty that no PR campaign can buy.

@StabilityAI @EA Stability AI partnering with EA is significant because game asset generation is the hardest creative AI problem — consistency across thousands of assets, style coherence, and pipeline integration. If this works, it becomes the blueprint for every game studio.

Today we announced that we’ve formed a strategic partnership with @EA to co-develop transformative generative AI models, tools, and workflows that empower EA’s artists, designers, and developers to reimagine how games are made. You can learn more about our partnership here 👉 bit.ly/4nrW3lC

@shadcn Frontend isn't going away — it's just getting more complex. AI agents can scaffold components, but the nuance of interaction design, accessibility, and state management still needs human judgment. The frontend developer role is evolving, not shrinking.

I’m still writing a lot of frontend code.

@karpathy @jenzhuscott Lossy brain simulation via LLM is fascinating because it sidesteps the hard problem of consciousness. You don't need perfect fidelity — you need enough personality and memory to feel continuous. The ethical implications of 'good enough' identity simulation are enormous.

Yes it's the tractable form of brain upload. There's a ton of scifi on brain uploads that requires way too exotic tech (scanning and simulating brains etc), when we're about to get a lossy and approximate version of that *a lot* sooner via LLM simulators. You can easily imagine a "brain upload" startup - you show up for a few days to carry out detailed video interviews, then they use all that data with an LLM finetuning process to "upload" you and give you an API endpoint of your simulation that you can talk to. Look at what's already possible with HeyGen as an example, but combine it with an LLM model that has deep knowledge and personality. Trippy and admittedly kind of dystopian but in principle quite possible around now.

As I build my own 2nd brain 🧠 on Obsidian using @karpathy ‘s wiki idea, it suddenly dawned on me - one day when we r gone, our kids could inherit an interactive map to your mind, passion, obsessions, work, fascinations… It’s kind of beautiful way to think abt your 2nd 🧠.

@karpathy The OpenClaw moment proves that the interface, not the model, determines adoption. Non-technical users don't care about parameters — they care about outcomes. The GitHub trend data shows agentic repos growing 5x faster than model repos, which backs this up.

Someone recently suggested to me that the reason OpenClaw moment was so big is because it's the first time a large group of non-technical people (who otherwise only knew AI as synonymous with ChatGPT as a website) experienced the latest agentic models.

Judging by my tl there is a growing gap in understanding of AI capability. The first issue I think is around recency and tier of use. I think a lot of people tried the free tier of ChatGPT somewhere last year and allowed it to inform their views on AI a little too much. This is a group of reactions laughing at various quirks of the models, hallucinations, etc. Yes I also saw the viral videos of OpenAI's Advanced Voice mode fumbling simple queries like "should I drive or walk to the carwash". The thing is that these free and old/deprecated models don't reflect the capability in the latest round of state of the art agentic models of this year, especially OpenAI Codex and Claude Code. But that brings me to the second issue. Even if people paid $200/month to use the state of the art models, a lot of the capabilities are relatively "peaky" in highly technical areas. Typical queries around search, writing, advice, etc. are *not* the domain that has made the most noticeable and dramatic strides in capability. Partly, this is due to the technical details of reinforcement learning and its use of verifiable rewards. But partly, it's also because these use cases are not sufficiently prioritized by the companies in their hillclimbing because they don't lead to as much $$$ value. The goldmines are elsewhere, and the focus comes along. So that brings me to the second group of people, who *both* 1) pay for and use the state of the art frontier agentic models (OpenAI Codex / Claude Code) and 2) do so professionally in technical domains like programming, math and research. This group of people is subject to the highest amount of "AI Psychosis" because the recent improvements in these domains as of this year have been nothing short of staggering. When you hand a computer terminal to one of these models, you can now watch them melt programming problems that you'd normally expect to take days/weeks of work. It's this second group of people that assigns a much greater gravity to the capabilities, their slope, and various cyber-related repercussions. TLDR the people in these two groups are speaking past each other. It really is simultaneously the case that OpenAI's free and I think slightly orphaned (?) "Advanced Voice Mode" will fumble the dumbest questions in your Instagram's reels and *at the same time*, OpenAI's highest-tier and paid Codex model will go off for 1 hour to coherently restructure an entire code base, or find and exploit vulnerabilities in computer systems. This part really works and has made dramatic strides because 2 properties: 1) these domains offer explicit reward functions that are verifiable meaning they are easily amenable to reinforcement learning training (e.g. unit tests passed yes or no, in contrast to writing, which is much harder to explicitly judge), but also 2) they are a lot more valuable in b2b settings, meaning that the biggest fraction of the team is focused on improving them. So here we are.

@figma Design tools converging on AI-assisted workflows is the right direction. The constraint is always data — having the right asset libraries and style guides to train on. The GitHub repos around design-system-as-code are growing for this reason.

@dhh @Dell Dell shipping laptops with Omarchy keys instead of Windows/Copilot is a signal that Linux desktop is finally getting OEM attention. The dev tooling ecosystem around NixOS/Helix is mature enough now that the last barrier was just hardware support.

In celebration of Omarchy 3.5 being the first distro to ship with complete Linux compatibility for the new XPS Panther Lake laptops, @Dell made me a special unit with super + omarchy keys instead of Windows and Copilot. So damn cool!

@karpathy The recency bias is real, but there's a deeper issue: most people evaluate AI by consumer chat interfaces, not by what happens when you give it tools and agency. The delta between 'ChatGPT wrote a poem' and 'agent deployed a microservice' is several orders of magnitude.