ReMe vs Memary
Persistent memory for AI agents compared
ReMe and Memary both help AI agents retain useful information across interactions. ReMe organizes memories as editable Markdown files, while Memary centers on graph-based entity relationships and retrieval.
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.
Memary: Add Memory to Autonomous Agents with LlamaIndex
Memary is an open-source memory layer for autonomous agents, built around LlamaIndex, knowledge graphs, and a ReAct chat agent. It stores interaction-derived entities and user context to help agents retrieve relevant information and tailor responses.
| ReMe | Memary | |
|---|---|---|
| Language | Python | Jupyter Notebook |
| License | Apache-2.0 | MIT |
| Stars | 3.7k | 2.7k |
| Forks | 321 | 205 |
| Last analyzed | Oct 23, 2026 | Oct 25, 2026 |
Key differences
- ReMe stores durable memories in Markdown files with frontmatter and wikilinks; Memary stores entity relationships in a knowledge graph, with FalkorDB and Neo4j listed as options.
- ReMe supports BM25 and wikilink search, with optional embeddings; Memary uses recursive and multi-hop graph retrieval and can fall back to external search.
- ReMe offers CLI, HTTP, MCP, and Python interfaces, plus integrations for supported agent runtimes; Memary includes a ReAct-based chat agent and tools for extending it.
- ReMe is described as local-first, with a local service and workspace by default; Memary uses graph databases or local JSON files, with deployment choices affecting persistence and multi-agent setup.
- ReMe is written in Python and uses the Apache-2.0 license; Memary is listed as Jupyter Notebook and uses the MIT license.
- ReMe has 3.5k stars and 306 forks; Memary has 2.7k stars and 206 forks.
Choose ReMe if you…
- want readable, editable Markdown memory under user control.
- need CLI, HTTP, MCP, or Python access to a shared memory workspace.
- want BM25 and wikilink retrieval, with embeddings available as an option.
Choose Memary if you…
- need graph-based retrieval that connects entities across facts or interactions.
- want to track entity frequency and recency and assemble context with multi-hop retrieval.
- plan to prototype a personalized agent using the included ReAct application and tools.
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.