agentic-box/memora
An MCP server that gives AI agents persistent memory with semantic search, graph relations, and structured cross-session storage.
Not currently ranked — collecting fresh signals.
star history
Memora is a Model Context Protocol server implementing a memory layer for AI agents. It provides SQLite-backed persistent storage with optional cloud sync, semantic search using embeddings, knowledge graph relations between memories, and source-backed retrieval. Agents can absorb work into durable memory and retrieve relevant context, TODOs, and related edges via the memory_digest function across sessions.
Frequently asked
- What is agentic-box/memora?
- An MCP server that gives AI agents persistent memory with semantic search, graph relations, and structured cross-session storage.
- Is memora open source?
- Yes — agentic-box/memora is open source, released under the MIT license.
- What language is memora written in?
- agentic-box/memora is primarily written in Python.
- How popular is memora?
- agentic-box/memora has 422 stars on GitHub.
- Where can I find memora?
- agentic-box/memora is on GitHub at https://github.com/agentic-box/memora.