ReMe

RAG & Memory 💻 Python ⚖️ Apache-2.0 🟢 Actively maintained
3.4k stars

ReMe is a local-first, self-evolving personal knowledge base for AI agents. It stores durable memory as readable, editable Markdown files with frontmatter and wikilinks, and provides hybrid search (BM25 + optional embeddings) for precise recall. It is designed for developers building agent systems that need persistent, shared memory across sessions and tools.

ReMe demo
🎞️ Demo from the project README

✨ Key features

  • Memory as Markdown files with frontmatter and wikilinks
  • Self-evolving knowledge base from conversations and resources
  • Hybrid search: BM25, optional embeddings, wikilink expansion
  • One workspace shared across multiple agents
  • Multiple interfaces: CLI, HTTP, MCP, Python API
  • Rebuildable indexes and metadata

🎯 Use cases

  • Give AI agents long-term memory across sessions
  • Build a personal knowledge base from chat logs and documents
  • Enable multiple agents to share and evolve a common memory
  • Create a searchable archive of procedures and facts
  • Integrate memory into agent frameworks like DeepSeek Harness or OpenClaw

📦 Installation

🧰 Requirements: Python 3.11+; optional LLM API keys for auto_memory/auto_dream; optional embedding API key for semantic search; Node.js 22.13+ only for source build of Studio.

pip install "reme-ai[core]"

Install from source:

git clone https://github.com/agentscope-ai/ReMe.git
cd ReMe
pip install -e reme_studio -e ".[core]"
cd reme_studio
npm ci
npm run build:static
cd ..

🚀 Usage

reme start
in another terminal
reme write \
  path=digest/wiki/quick-start-demo \
  name="Quick Start Demo" \
  description="A first ReMe memory node" \
  content="# Quick Start Demo\n\nReMe stores agent memory as readable Markdown.\n\nRelated: [[digest/wiki/memory-as-file.md]]"
reme search query="agent memory markdown" limit=5
reme read path=digest/wiki/quick-start-demo start_line=1 end_line=20

⚠️ Good to know

Embeddings are disabled by default; proactive only reads topics and does not act on its own.

❓ FAQ

What Python version is required?

ReMe requires Python 3.11 or higher.

Do I need an LLM API key to use ReMe?

No, basic file operations, BM25 search, wikilink traversal, and reading proactive topics work without LLM credentials. LLM keys are required only for auto_memory, auto_resource, and auto_dream.

How do I enable semantic search?

You need to uncomment components.as_embedding and components.embedding_store in reme/config/default.yaml, and change components.file_store.default.embedding_store from "" to default. Then set the embedding API key and base URL in your environment.

Can I use ReMe with multiple agents?

Yes, ReMe is designed to be shared across agents. It provides native integrations for DeepSeek Harness, OpenClaw, QwenPaw, Claude Code, Hermes, and CLI agents via SKILL.md.

📊 Repository

Stars★ 3,415
Forks🍴 293
Open issues🐛 26
Last commit🕒 Sep 4, 2026
Created📅 Aug 2024
Language💻 Python
License⚖️ Apache-2.0

🤖 Overview, features, install steps and FAQ were generated from the project's README on Sep 4, 2026. Always check the original source before running commands.