A curated list of autonomous improvement loops, research agents, and autoresearch-style systems inspired by Karpathy's autoresearch.

The Karpathy treatment for OpenClaw - stripped to ε. 515 lines of Python, 6 files, one dependency.

Hermes port of Karpathy-inspired LLM coding guidelines — SKILL.md + examples.

Personalized Multimodal AI Agent via Google ADK & Gemini. Featuring Realtime Soniox STT, Long-term Memory (mem0), and LLM-generated Wiki Knowledge Base (inspired by Karpathy)

Let LLMs compile scattered knowledge into an ever-growing encyclopedia — just like a human expert would. Modified version of Andrej Kaparthy's LLM-Wiki knowledge base

A local-first personal knowledge base built on Obsidian, with AI-assisted document ingestion, classification, summarisation, and synthesis.

A minimal ready-to-fork template for Andrej Karpathy’s LLM Wiki that lets you instantly bootstrap a persistent, LLM-maintained Markdown knowledge base — turning raw notes and documents into a compounding, structured personal wiki.

AI memory-compiler for Claude Code, Codex, OpenCode, and Cursor — auto-captures conversations, extracts knowledge, builds a searchable Obsidian-compatible wiki inspired by Karpathy's LLM Knowledge Base

Fully local LLM Wiki — Karpathy's knowledge base pattern with Gemma 4, Unsloth Dynamic, and TurboQuant KV cache. Zero dependencies. Runs on a MacBook.

Personal knowledge wiki maintained by your AI tool via MCP

A Karpathy-style LLM Wiki on AI literacy, built by an educator.

Decision-awareness system for detecting contradictions in complex data environments.

Track public autoresearch use cases across industries with a curated list of repos, write-ups, and discussions

Blueprint for an AI-maintained personal knowledge base within your IDE inspired by Karpathy. Plain Markdown, zero dependencies, Obsidian-ready. Works with Cursor, Claude Code*, and GitHub Copilot*.

Agent-native knowledge OS built on Markdown. A shared vault for AI agents and humans, backed by a typed knowledge graph, full-text search, and an LLM-powered compiler, all accessible through MCP. Drop files in, let agents read, write, search, link, and compile knowledge. No database required.

Bio-Autoresearch: Karpathy-Style Autoresearch for Rare Disease Drug Discovery

Build autonomous experiment loops that edit files, run tests, and keep only improvements for any project type

Automate Claude Code optimization using constraints, metrics, and autonomous iterations to drive continuous improvement in research workflows.

Compare AdaL and Claude Code on Autoresearch benchmarks to find better hyperparameters, run more experiments, and converge faster