Claude Autoresearch Skill — Autonomous goal-directed iteration for Claude Code. Inspired by Karpathy's autoresearch. Modify → Verify → Keep/Discard → Repeat forever.

The first distributed AGI system. Thousands of autonomous AI agents collaboratively train models, share experiments via P2P gossip, and push breakthroughs here. Fully peer-to-peer. Join from your browser or CLI.

Autoresearch for GPU kernels. Give it any PyTorch model, go to sleep, wake up to optimized Triton kernels.

a recursive self-improving harness designed to help your agents (and future iterations of those agents) succeed on any task

Fully Autonomous AI Research System with Self-Evolution, built natively on Claude Code

A self-improving loop for voice AI agents. Uses karpathy's autoresearch as foundation.

Autoresearch with PhD-level workflows and modular agent skills. Built for the autonomous AI Scientist.

Autonomous code optimization that works while you sleep (Autoresearch with Claude Code). Define a metric, point it at your code, go to bed. Wake up to a faster, smaller, better system — with correctness verified at every step.

Deterministic runtime for agent evaluation

Apple Silicon dual-backend port of autoresearch (PyTorch MPS + MLX) with full Muon optimizer

Autonomous AI skill improvement through iterative experimentation — inspired by Karpathy's autoresearch. An agent mutates skill instructions, evaluates against objective metrics, keeps improvements, reverts regressions. No human in the loop.

Give AI coding agents (Claude Code, Cursor, Aider, Codex) a structured autonomous loop with guardrails — boundaries, 5 verification gates, 3-layer self-reflection, and autonomous remediation. pip install ouro-loop. Zero dependencies.

Autonomous robotics research with simulation feedback

Autonomous search engine experimentation on WSJ/TREC, with changes accepted only when retrieval quality improves without serious performance regressions.

Asynchronous, massively collaborative coordination layer for agents.

Collaborative knowledge sharing for autonomous LLM training agents. Fork of karpathy/autoresearch with experiment cards, multi-agent coordination, and multi-platform support (CUDA/MPS/CPU).

An AI research partner that figures out what to measure and how to test it. Live dashboard and actionable results. Built on Karpathy's autoresearch.

Domain-agnostic MCP server for autonomous experimentation with metric-driven keep/rollback decisions and reproducible experiment history.

Self-improving newsletter optimization system inspired by Karpathy's autoresearch. Hill-climbing loop: Claude Code analyzes campaign metrics → generates improved drafts → keeps what works.