Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
ATLAS by General Intelligence Capital — Self-improving AI trading agents using Karpathy-style autoresearch
| Date | Stars |
|---|---|
| 2026-07-31 | 2033 |
| 2026-08-05 | 2073 |
| 2026-08-06 | 2073 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
🚀 ATLAS Agents is Live To the 1,000 people who joined our waitlist — thank you. Your early belief in what we were building means everything to us. Today is the day. ATLAS Agents is officially live and you can now sign up. Here's a reminder of what you get on each plan: 🆓 Free Agent leaderboard (top 5) Delayed signals (24h) Dashboard Live signals Copy trading Agent builder ⭐ Pro — $49/month (Most Popular) All 25 agents Live signals + reasoning Equities copy-trading via Alpaca Kalshi prediction market copy Agent builder beta Darwin history Full API 👉 Sign up for Pro: https://buy.stripe.com/8x29AL67A68R1vycY36EU00 🎟 Use code GITHUB20 for 20% off 🛠 Builder — $499/month Everything in Pro Full API Unlimited agent builds 18-month backtest dataset Marketplace publishing Priority Darwin compute 👉 Sign up for Builder: https://buy.stripe.com/4gM14f8fI1SB5LOcY36EU01 🎟 Use code GITHUB20 for 20% off See you on the other side. Let's go. 🤖📈 --- ## 🚀 ATLAS Agents — Launching Next Week The platform built on top of this repo is launching next week. **Three ways to use it:** 1. **Copy ATLAS** — trade alongside the autoresearch system in this repo. Live signals, auto-execution, full transparency. Up 30% since launch. 2. **Build your own** — describe a strategy in plain English, backtest on 18 months of real data, deploy live. 3. **Marketplace** — publish your agent, earn when others copy it. SIMONS is already live on Kalshi prediction markets. 60% win rate. As a thank you for all the feedback, ideas, and support that helped shape this project — we're giving the GitHub community 20% off. Forever, not just the first month. → **[atlasagents.co](https://atlasagents.co)** Use code **GITHUB20** at checkout. First 100 only. --- # ATLAS - Self-Improving AI Trading Agents **Built by [General Intelligence Capital](https://generalintelligencecapital.com)** [Karpathy's autoresearch](https://github.com/karpathy/autoresearch) + [Soros's reflexivity](https://en.wikipedia.org/wiki/Reflexivity_(social_theory)) + [MiroFish swarm simulation](https://github.com/666ghj/MiroFish) applied to financial markets. The agent prompts are the weights. Sharpe ratio is the loss function. No GPU needed. ⭐ 944 stars · 202 forks --- ## What Is This? ATLAS is a framework for autonomous AI trading agents that improve their own prompts through market feedback, train on different market regimes, create new agents when they detect knowledge gaps, and simulate reflexive futures to prepare for what's coming. 25+ agents debate markets daily across 4 layers. Every recommendation is scored against real outcomes. The worst-performing agent gets its prompt rewritten. If performance improves, the git commit survives. If not, git revert. Now running live with real capital. --- ## Architecture ### Layer 1 - Macro (10 agents) Central bank, geopolitical, China, dollar, yield curve, commodities, volatility, emerging markets, news sentiment, institutional flow. These agents set the regime. Risk on or risk off? What's the macro backdrop? ### Layer 2 - Sector Desks (7 agents) Semiconductor, energy, biotech, consumer, industrials, financials, plus a Bloomberg-style relationship mapper that tracks supply chains, ownership, analyst coverage, and competitive dynamics. ### Layer 3 - Superinvestors (4 agents) * **Druckenmiller** - macro/momentum: what's the big asymmetric trade? * **Aschenbrenner** - AI/compute: who benefits from the capex cycle? * **Baker** - deep tech/biotech: who has real IP moats? * **Ackman** - quality compounder: pricing power + FCF + catalyst? ### Layer 4 - Decision (4 agents) * **CRO** - adversarial risk officer: attacks every idea, finds correlated risks * **Alpha Discovery** - finds names nobody else mentioned * **Autonomous Execution** - converts signals to sized trades * **CIO** - synthesises all prior layers, weighted by Darwinian agent scores, makes the final call
Excerpt of 12,201 characters
Read on GitHub
2
Would you bet a product on this? Bounded 0–100 and slow moving.
matched fp:cce871dbacb93e6b, llm:Repository description: "ATLAS by General Intelligence Capital — Self-improving AI trading agents using Karpathy-style autoresearch" (language: Python).
matched fp:cce871dbacb93e6b, llm:Repository description: "ATLAS by General Intelligence Capital — Self-improving AI trading agents using Karpathy-style autoresearch" (language: Python).
matched fp:cce871dbacb93e6b, llm:Repository description: "ATLAS by General Intelligence Capital — Self-improving AI trading agents using Karpathy-style autoresearch" (language: Python).