A curated list of autonomous improvement loops, research agents, and autoresearch-style systems inspired by Karpathy's autoresearch.
-
Updated
Aug 10, 2026
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.
This is my implementation of Karpathy's LLM Wiki. A key highlight is the use of agent skills to enforce workflow rigidity and mitigate agent behavioral deviation.
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)
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.
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
Fully local LLM Wiki : Karpathy's knowledge base pattern with Gemma 4, Unsloth Dynamic, and TurboQuant KV cache. Zero dependencies. Runs on a MacBook.
The local system that remembers your projects, maintains itself, and is ready in 5 minutes. But, most of all, a Second Brain that doesn't become a Second Job.
Everyone is building AI second-brains. Some are self-hostable, some are cloud-based. But, which one is actually worth your time & costs?
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
Bio-Autoresearch: Karpathy-Style Autoresearch for Rare Disease Drug Discovery
llm-wiki designed for developers and builders focused on projects building phases. Helps both the agent and builder to stay on track.
A Claude skill that pressure-tests decisions through independent AI advisors with peer review. Based on Karpathy's LLM Council.
基于 Karpathy LLM Wiki 模式的智能体知识库运营 Skill:Ingest 摄入 / Query 回填 / Lint 六维体检,把外部知识蒸馏为智能体可调用的行为规则。
A local-first personal knowledge base built on Obsidian, with AI-assisted document ingestion, classification, summarisation, and synthesis.
Open-source AI evaluation for Africa. A pre-deployment compliance harness for AI in Africa — evidence-linked Context Scores and per-language profiles, open source, built on WAXAL · AfroBench · afri-fertility.
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.
Automate Claude Code optimization using constraints, metrics, and autonomous iterations to drive continuous improvement in research workflows.
Karpathy-inspired coding-agent discipline, packaged natively for Codex.
Add a description, image, and links to the karpathy-inspired topic page so that developers can more easily learn about it.
To associate your repository with the karpathy-inspired topic, visit your repo's landing page and select "manage topics."