Repolode
Submit
agentscope-ai

agentscope-ai/ReMe

A local-first Markdown memory workspace that lets AI agents store, search, and consolidate long-term knowledge as editable files.

Open-source alternative to:Mem.aiNotion AI

View on GitHubWebsite
Stars
3,655
Forks
319
Open issues
46
Last commit
yesterday

What is it?

ReMe is a memory management kit for AI agents. It turns conversations and external resources into ordinary Markdown files with frontmatter and wikilinks, organized into session, daily, and digest layers. Users and agents can read, edit, sync, and back up these files with standard tools, while indexes and graphs under the metadata folder can be rebuilt at any time.

The system follows a capture, index, consolidate, and recall loop. Automatic memory distills useful facts from conversations, auto resource turns files into source-linked daily cards, and auto dream consolidates recent notes into long-term digest nodes. Search combines BM25 keyword matching, optional embeddings, and wikilink expansion to return line-level passages with their neighbors.

ReMe can be used as a CLI, HTTP service, MCP server, or embedded Python API, and it includes a browser-based Studio for browsing and editing the workspace. Integrations are provided for agent runtimes such as Claude Code, Codex, OpenClaw, Hermes, and DeepSeek Harness.

Key features

  • 01Stores memory as readable Markdown files with frontmatter and wikilinks.
  • 02Hybrid retrieval with BM25 keyword search, optional embeddings, and wikilink expansion.
  • 03Automatically distills conversations and imported text or image resources into daily notes.
  • 04Consolidates recent notes into long-term digest nodes on a schedule.
  • 05Exposes memory through CLI, HTTP, MCP, and Python APIs.
  • 06Includes a browser-based Studio for editing and browsing memory graphs.

Use cases

  • Giving a personal coding or assistant agent persistent memory of preferences and project context.
  • Sharing one local memory workspace across several agent runtimes.
  • Maintaining a personal knowledge base that you can read and edit directly.
  • Keeping research notes linked to their original sources.

Quick start

  1. 1

    Install ReMe from pip

    pip install "reme-ai[core]"
  2. 2

    Start the ReMe service and leave it running

    reme start
  3. 3

    Check the running service

    reme health_check

Things to know

  • Requires Python 3.11 or newer.
  • LLM features such as auto_memory and auto_dream need an LLM API key.
  • Embedding-based semantic search is disabled by default and needs manual configuration.

Topics

#agent#memory#markdown#rag#knowledge-base#bm25#mcp#local-first#ai-agents#dsh-plugin

FAQ

What is ReMe?
A local-first Markdown memory workspace that lets AI agents store, search, and consolidate long-term knowledge as editable files. ReMe is a memory management kit for AI agents. It turns conversations and external resources into ordinary Markdown files with frontmatter and wikilinks, organized into session, daily, and digest layers. Users and agents can read, edit, sync, and back up these files with standard tools, while indexes and graphs under the metadata folder can be rebuilt at any time.
Is ReMe free to use? What license does it have?
ReMe is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
Is ReMe still maintained?
Actively maintained. The last commit to ReMe was on Oct 19, 2026.
What can I use ReMe for?
Giving a personal coding or assistant agent persistent memory of preferences and project context. Sharing one local memory workspace across several agent runtimes. Maintaining a personal knowledge base that you can read and edit directly. Keeping research notes linked to their original sources.
How do I install ReMe?
Install ReMe from pip: pip install "reme-ai[core]" Start the ReMe service and leave it running: reme start Check the running service: reme health_check
Where can I get ReMe?
The source code and installation instructions are on GitHub: https://github.com/agentscope-ai/ReMe

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects