mnemosyne
Zero-cloud AI memory that works everywhere. SQLite-backed. One pure-Python dependency.
What is mnemosyne?
Zero-cloud AI memory that works everywhere. SQLite-backed. One pure-Python dependency.
Overview of mnemosyne
What it is
Mnemosyne is a universal, Hermes-first memory layer for AI agents. It is SQLite-backed, requires zero cloud services, and has a single pure-Python dependency. It works with any agent framework, including Cursor, Claude Code, Codex, Windsurf, OpenWebUI, Pi, OpenClaw, and custom agents.
Key features
- SQLite-backed, zero-cloud, single pure-Python dependency.
- Works with any agent framework via MCP (stdio/SSE/Streamable HTTP) or Python SDK.
- Memory banks for per-domain isolation (e.g., work, personal).
- Temporal recall with recency boost and temporal halflife.
- Entity extraction and LLM-driven fact extraction from natural language.
- Temporal triples (knowledge graph) support scoped to a memory bank.
- CLI for direct memory operations, export/import, and bidirectional sync between instances.
Use cases
- Adding persistent memory to AI agents (e.g., storing user preferences, session context).
- Cross-platform memory layer for tools like Cursor, Claude Code, Codex, and Windsurf.
- Domain-specific memory banks for different projects or contexts.
- Storing and recalling facts with temporal awareness and entity extraction.
Getting started
Install Mnemosyne: bash pip install mnemosyne-memory # With all features (vector search + MCP server) pip install 'mnemosyne-memory[all]' Add to an MCP-based agent (e.g., Cursor, Claude Code): json { 'mcpServers': { 'mnemosyne': { 'command': 'mnemosyne', 'args': ['mcp'], 'env': {} } } } Use the Python SDK: python from mnemosyne import remember, recall remember('User prefers dark mode interfaces') results = recall('user preferences')
License and pricing
Mnemosyne is open source under the MIT license. No pricing information is provided beyond being open source.
Based on the project's public README and repository metadata.