10 GitHub repos that defined 2026 so far. Bookmark this list. 1. OpenClaw Peter Steinberger went from 9,000 to over 300,000 stars in months. Personal AI assistant that runs entirely on your devices. Steinberger joined OpenAI shortly after. Repo → github.com/openclaw/openclaw 2. anthropics/skills 135,000 stars. The patterns Anthropic uses internally to extend Claude. The repo that defined the skills ecosystem of 2026. Repo → github.com/anthropics/skills 3. affaan-m/everything-claude-code 141,000 stars. The complete library of Claude Code skills, agents, and commands. The aggregator every serious builder forks. Repo → github.com/affaan-m/everythi… 4. andrej-karpathy-skills Forrest Chang built a single CLAUDE md file in January. 109,000 stars. The most starred single-file repo in GitHub history. Repo → github.com/forrestchang/andr… 5. Hermes Agent Nous Research released this in February. 105,000 stars. The self-evolving AI agent that gets smarter the more you use it. Repo → github.com/NousResearch/herm… 6. obra/superpowers 94,000 stars. Officially accepted into Anthropic's skills marketplace. Multi-agent orchestration without the boilerplate. Repo → github.com/obra/superpowers 7. claude-task-master Multi-agent orchestration on top of Claude Code. Turn one prompt into a coordinated team of specialists shipping a feature while you sleep. Repo → github.com/eyaltoledano/clau… 8. MemPalace Milla Jovovich, the actress from Resident Evil, co-built this with Ben Sigman using Claude Code. Near-perfect score on the LongMemEval benchmark. Repo → github.com/MemPalace/mempala… 9. karpathy/autoresearch Andrej Karpathy released his own research automation framework. 23,000 stars in three days. The closest thing to having Karpathy as your research partner. Repo → github.com/karpathy/autorese… 10. karpathy/nanochat Karpathy released a complete pipeline to train a ChatGPT from scratch for $100. 54,800 stars. The best ChatGPT money can buy is the one you train yourself. Repo → github.com/karpathy/nanochat Save this. Share it with the developer in your life who deserves to know what they missed. 100% free. 100% open source.

Codex Design is becoming real. The real breakthrough isn’t one more design tool. It’s keeping the same context from idea to design output to launch asset. We tested it with Codex Plugins, Open Design, and html-video: from vague idea to polished promo video. All inside Codex.

Vapi charges $0.05/minute per call. Retell charges $0.07/minute per call. A 10-person sales team running 8 hours of AI calls a day pays Vapi $1,200/month. $14,400/year. Someone built the open source alternative to both. It's called Dograh. 1,516 stars on GitHub. You open the drag-and-drop workflow builder. You name your bot. You describe the use case in a sentence. You have a working voice agent in under 2 minutes. No API contracts. No per-minute billing. No vendor lock-in. Here's what it does: → Drag-and-drop workflow builder for inbound and outbound voice agents. No code required. → Bring your own LLM. OpenAI, Anthropic, or any provider you already use. → Bring your own STT. Swap speech-to-text engines without rebuilding your agent. → Bring your own TTS. Full control over voice, latency, and cost. → Outbound calling campaigns. Dial lists, lead qualification, follow-ups. → Inbound call handling. Receptionists, support bots, appointment schedulers. → WebRTC and VoIP support. Real phone calls and browser-based voice. → Built on Pipecat and FastAPI. Production-grade Python backend. → Next.js frontend. Runs entirely in your browser. No desktop app needed. → One Docker command to self-host everything. Running in under 60 seconds. → Full source-level customization. Every line of code is yours to modify. → Data residency on your own infrastructure. No call audio leaves your server. Here's the wildest part: There is no Dograh cloud you are forced to use. There is no per-minute meter running on someone else's servers. There is no closed-source black box deciding what your voice agent can and cannot do. Vapi is SaaS only. Retell is SaaS only. Both are proprietary. Neither lets you see the code. Neither lets you change the code. Neither lets you own the infrastructure. Dograh can never lock you out. Because the code is sitting on your own machine. Vapi: $0.05/minute. A 10-person team at 8 hours/day of calls pays $14,400/year. Retell: $0.07/minute. Same team. Same usage. $20,160/year. Dograh: $0. Unlimited agents. Unlimited calls. Your hardware. Your data. Forever. 344 forks. Built in Python. Maintained by YC alumni and exit founders. BSD-2-Clause licensed. Self-hosted. Free forever. 100% Open Source.

Where will AI be in 1, 2 or 3 years?

This is going to be exciting

Anthropic just leaked their 2026 agent roadmap in 22 minutes. Claude team walked through tools, memory, observability, and the things most builders are 12 months behind. The last 3 minutes alone are worth the watch. Watch it, then save the setup below 👇

Dogfooding Opus 4.7 the last few weeks, I've been feeling incredibly productive. Sharing a few tips to get more out of 4.7 🧵