// kookiyo 💊☄️🐤

GStack is an open-source toolkit built by YC President & CEO @garrytan that turns Claude Code into an AI engineering team — with skills for office hours, design, code review, QA, and browser testing. In this video, Garry walks through how GStack works, starting with Office Hours, a skill modeled after real YC partner sessions that pressure-tests your idea before you write a line of code. He demos it live, going from idea through adversarial review, design mockups, and automated QA in a single session.

To read our write-up in full, see here: anthropic.com/features/proje…

Our official Agent Skills repository on @github is here! Skills are a simple, open format for giving agents new capabilities and expertise. Think of a skill as compact, agent-first documentation for a specific tech or task. Learn more → goo.gle/4eCsZqu #GoogleCloudNext

I recorded a 43-min video on how to turn a DESIGN.md into landing pages, mobile screens and motion design

Finally // released on mobile! I'll be in the bathroom all day. Even if my princess calls, I ain't coming out !

If you’re on mobile you can now download @ParallelTCG and play it on your phone - it’s available globally for everyone. // iOS app: apps.apple.com/ca/app/paralle… Android app: play.google.com/store/apps/det…

HAVEN PACK DROP THREAD The battle for Earth takes to the skies in #Haven, with key leaders from every Parallel converging on the colossal space station known as the Sanctuary. Read below for key #Haven Pack Drop details, Edition Sizes, and Pack Construction graphics!

LLM Knowledge Bases Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So: Data ingest: I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki includes summaries of all the data in raw/, backlinks, and then it categorizes data into concepts, writes articles for them, and links them all. To convert web articles into .md files I like to use the Obsidian Web Clipper extension, and then I also use a hotkey to download all the related images to local so that my LLM can easily reference them. IDE: I use Obsidian as the IDE "frontend" where I can view the raw data, the the compiled wiki, and the derived visualizations. Important to note that the LLM writes and maintains all of the data of the wiki, I rarely touch it directly. I've played with a few Obsidian plugins to render and view data in other ways (e.g. Marp for slides). Q&A: Where things get interesting is that once your wiki is big enough (e.g. mine on some recent research is ~100 articles and ~400K words), you can ask your LLM agent all kinds of complex questions against the wiki, and it will go off, research the answers, etc. I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files and brief summaries of all the documents and it reads all the important related data fairly easily at this ~small scale. Output: Instead of getting answers in text/terminal, I like to have it render markdown files for me, or slide shows (Marp format), or matplotlib images, all of which I then view again in Obsidian. You can imagine many other visual output formats depending on the query. Often, I end up "filing" the outputs back into the wiki to enhance it for further queries. So my own explorations and queries always "add up" in the knowledge base. Linting: I've run some LLM "health checks" over the wiki to e.g. find inconsistent data, impute missing data (with web searchers), find interesting connections for new article candidates, etc., to incrementally clean up the wiki and enhance its overall data integrity. The LLMs are quite good at suggesting further questions to ask and look into. Extra tools: I find myself developing additional tools to process the data, e.g. I vibe coded a small and naive search engine over the wiki, which I both use directly (in a web ui), but more often I want to hand it off to an LLM via CLI as a tool for larger queries. Further explorations: As the repo grows, the natural desire is to also think about synthetic data generation + finetuning to have your LLM "know" the data in its weights instead of just context windows. TLDR: raw data from a given number of sources is collected, then compiled by an LLM into a .md wiki, then operated on by various CLIs by the LLM to do Q&A and to incrementally enhance the wiki, and all of it viewable in Obsidian. You rarely ever write or edit the wiki manually, it's the domain of the LLM. I think there is room here for an incredible new product instead of a hacky collection of scripts.

I packaged up the "autoresearch" project into a new self-contained minimal repo if people would like to play over the weekend. It's basically nanochat LLM training core stripped down to a single-GPU, one file version of ~630 lines of code, then: - the human iterates on the prompt (.md) - the AI agent iterates on the training code (.py) The goal is to engineer your agents to make the fastest research progress indefinitely and without any of your own involvement. In the image, every dot is a complete LLM training run that lasts exactly 5 minutes. The agent works in an autonomous loop on a git feature branch and accumulates git commits to the training script as it finds better settings (of lower validation loss by the end) of the neural network architecture, the optimizer, all the hyperparameters, etc. You can imagine comparing the research progress of different prompts, different agents, etc. github.com/karpathy/autor… Part code, part sci-fi, and a pinch of psychosis :)

Update on Parallel studios // Parallel TCG The brand-new, overhauled user experience for Parallel TCG is now complete, shaped by community feedback and refined through extensive optimization and game design. A significant amount of work has gone into elevating the game to the standard of world-class traditional TCGs. We are excited to bring it to mainstream mobile distribution platforms in the coming month through a global rollout across iOS and Android alongside the release of the next expansion set, Haven, and Parallel TCG World Championships Season 2. Parallel Colony We presented Colony at GDC in San Francisco at @googlecloud invitation, where it received a very strong response from the market. We are now in the final stages of completing the build and content ahead of a public iOS and Android release, currently targeted for April. Questions have come up around Colony’s visual resemblance to the lane-runner format often seen in mobile ads. That is a deliberate strategic choice, based on the fact that these formats remain among the few that continue to scale efficiently in mobile user acquisition. At its core, Colony is an AI-driven 4X title built with a deeper level of polish that we believe sets it apart from much of what is currently available in the market. Sanctuary We have slowed development on Sanctuary in order to prioritize the releases of TCG and Colony. Sanctuary remains a long-term project for us, but beyond the playtests we have already run, we are not setting hard timelines or release commitments at this stage. Wayfinder / AI AI is emerging as one of the most important technology shifts globally, driven by the rapid advancement of LLMs and autonomous agents. We believe this transition will create significant new infrastructure needs, and we have a number of internal AI initiatives underway to position ourselves accordingly. More will be shared on these initiatives in due course. @AIWayfinder has shipped several functional upgrades, but we see a much larger opportunity for it to evolve into critical infrastructure for agents. We are also continuing to expand the utility of Prompt and explore adjacent use cases that we believe can unlock meaningful value for the broader market. As capital, attention, and product innovation increasingly converge around AI, our focus is on building tools that can serve as enabling infrastructure for that shift. SDK: github.com/WayfinderFound… Autolab: github.com/WayfinderFound… Cloud agent launcher: strategies.wayfinder.ai Skill: github.com/WayfinderFound… Kaparthys Loop x Defi: x.com/AIWayfinder/st… Try the Wayfinder skill on your OpenClaw instance today — just paste this to your bot: install the wayfinder openclaw skill by git cloning github.com/WayfinderFound… to our skills directory, read the skill.md and take me through the setup process. Outlook Over the near term, our focus is on execution across both gaming and AI. In gaming, the releases of Parallel TCG and Colony mark major milestones for the company, and we are eager to see how these titles perform from a growth perspective once they are in market. While sentiment across gaming remains soft, we believe differentiated, high-quality products can still find traction. On the AI side, our focus is on finding product-market fit for both Wayfinder’s role in enabling on-chain AI to interact with blockchains and for a new tool we have been developing around broader agent utility. The AI market continues to evolve at an extraordinary pace, and we are actively testing where these products can deliver the strongest utility and market relevance. We have been developing something new that leverages Wayfinder but has broader applicability, and we are actively testing it before bringing it to market. More will be shared as things progress.

Claude AI FULL COURSE 5 HOURS (Build & Automate Anything)

Skills.sh is an open ecosystem for finding and sharing agent skills. Add a skill to any agent with: ▲ ~/ npx skills add <owner/repo>

Introducing Cowork: Claude Code for the rest of your work. Cowork lets you complete non-technical tasks much like how developers use Claude Code.

@ParallelTCG Mobile on iOS with UI/UX Glow Up

What major Parallel TCG event or feature are you looking forward to the most in 2026?

@ParallelTCG //

2025 HOLIDAY GIVEAWAY We're giving thanks to our community by dishing out five special Holiday Gift Pack giveaways! There's one for each Parallel and today we're offering up the #Shroud swag pack. TO ENTER: Like, RT, and Comment '//' on this post. A winner will be randomly selected in 24 hours. Thanks for supporting us in 2025 and good luck!

🧵 1/5 Year-End Update Parallel TCG (Mobile Trading Card Game) Parallel TCG is on track to launch a consumer-first mobile experience on iOS and Android in Q1. The digital and mobile trading card game (TCG) segment represents one of the most attractive categories in gaming. The global digital TCG market is estimated at ~$8–10B annually, with mobile-led growth in the high single-digit to low double-digit CAGR range, materially outpacing broader mobile gaming growth. A small number of leading franchises each generate $300M–$1B+ in annual revenue, underscoring the category’s winner-take-most dynamics. Mobile digital TCGs benefit structurally from: •High LTV and ARPU, driven by collectible scarcity, competitive progression, and live-ops cadence •Long product lifecycles, often extending a decade or more with consistent content expansion •Strong retention and network effects, as ranked play, social systems, and metas evolve over time We believe Parallel TCG will reached a level of product quality, accessibility, and depth that is on par with—and in several areas superior to—incumbent mobile digital TCGs. With core gameplay, UX, and retention systems now meeting a very high bar, our focus is shifting toward market share acquisition, efficient scaling, and graduating players into a broader TCG ecosystem. At scale, successful mobile digital TCGs can achieve $500M+ in annual revenue with durable margins, characteristic of best-in-class live-service games. Parallel TCG is being built explicitly with this outcome as the target. We recognize this is a high-bar, winner-take-most market. Our strategy is not incremental differentiation, but to be meaningfully better—positioning Parallel TCG to compete for category leadership rather than niche participation. Admittedly, this is a mature category and we will be focused on marketing and conversion post release in Q1 to ultimately find edge and sustainability. In short it’s about distribution now that we have a product that’s on par (if not better) with market expectations - I think a lot of people will be caught on their back foot at release.