ReMe vs karpathy-llm-wiki
Markdown knowledge and memory tools for AI agents compared
ReMe and karpathy-llm-wiki both use Markdown to give AI agents durable knowledge beyond a single conversation. ReMe centers on an evolving memory workspace with multiple service interfaces and retrieval methods, while karpathy-llm-wiki packages a curated, citation-based wiki workflow for compatible agent tools.
ReMe: Build a File-Based Memory Workspace for AI Agents
ReMe is a Python memory-management kit for agents, built around a local-first, self-evolving knowledge base stored as Markdown files. It helps agent users capture conversations and resources, refine durable knowledge, and retrieve relevant context across supported agent integrations.
karpathy-llm-wiki: Build Agent Skills LLM Wikis
Packages Karpathy’s LLM Wiki workflow as an Agent Skills skill for Claude Code, Cursor, Codex, and other compatible tools. Ingest sources into a curated Markdown knowledge base, query it with citations, and lint its links and index.
| ReMe | karpathy-llm-wiki | |
|---|---|---|
| Language | Python | Python |
| License | Apache-2.0 | MIT |
| Stars | 3.7k | 2.7k |
| Forks | 321 | 314 |
| Last analyzed | Oct 23, 2026 | Nov 3, 2026 |
Key differences
- ReMe captures conversations and resources, then refines them into daily and long-term memory; karpathy-llm-wiki ingests source material into curated wiki pages.
- ReMe retrieves passages with BM25, wikilink expansion, and optional embeddings; karpathy-llm-wiki answers from wiki pages with citations and is designed around grep and file reading.
- ReMe offers a CLI, HTTP service, MCP server, Studio interface, and Python API; karpathy-llm-wiki is an Agent Skills workflow installed for compatible tools.
- ReMe is Apache-2.0 licensed; karpathy-llm-wiki is MIT licensed.
- ReMe requires Python 3.11 or newer and supports Docker and Compose; karpathy-llm-wiki documents skill installation and copy-based setup.
- ReMe has regular commits and releases, while karpathy-llm-wiki has had no commits in the last 90 days and no release in over a year.
Choose ReMe if you…
- need memory capture and consolidation across conversations, resources, and agent sessions.
- want workspace operations through a CLI, service interfaces, MCP, or Python API.
- need passage-level search with BM25 and the option to enable embeddings.
Choose karpathy-llm-wiki if you…
- want a curated Markdown wiki built from source material and queried with citations.
- prefer an Agent Skills workflow for Claude Code, Cursor, Codex, or other compatible tools.
- want link and index linting for a project-directory knowledge base.
Compare the numbers side by side, or add more projects
This comparison is generated with AI from the OSRepos analyses of both projects. Always check each project's repository and documentation before choosing.