mnemosyne-oss/mnemosyne

How mnemosyne-oss/mnemosyne works

Mnemosyne is a local-first, SQLite-backed memory layer for agents. The main workflow is an agent storing a memory, then recalling relevant context; the runtime also supports fact extraction, multi-tier and canonical memory, consolidation, and agent-facing integrations. Groups distinguish interfaces, memory processing, retrieval, and persistence/operations. README statements and the supplied source excerpts/index support the shown relationships; unsampled subsystems are included where documented, but their internal calls are not inferred.

Main entry points: memory.py Memory API · beam.py BEAM Recall · embeddings.py Dense Embeddings · batch_tool.py Batch Operations

Components 30

Agent Interfaces

Python SDK mnemosyne/__init__.py
MCP Tools mnemosyne/mcp_tools.py
Command Line mnemosyne/cli.py
Hermes Provider integrations/hermes/src/mnemosyne_hermes/__init__.py
OpenWebUI Bridge mnemosyne/integrations/openwebui_tool.py
OpenClaw Provider mnemosyne/integrations/openclaw.py
VS Code Extension integrations/vscode-mnemosyne/src/client.ts
Obsidian Plugin integrations/obsidian-mnemosyne/main.ts

Memory Processing

Fact Extraction mnemosyne/extraction/client.py
Write Filtering mnemosyne/core/filters.py
Memory API mnemosyne/core/memory.py
Batch Operations mnemosyne/batch_tool.py
Veracity Consolidation mnemosyne/core/veracity_consolidation.py
Canonical Facts mnemosyne/core/canonical.py
Episodic Graph mnemosyne/core/episodic_graph.py
Content Canonicalization mnemosyne/core/canonical.py

Recall and Retrieval

BEAM Recall mnemosyne/core/beam.py
Dense Embeddings mnemosyne/core/embeddings.py
Memory Triples mnemosyne/core/triples.py
Annotations mnemosyne/core/annotations.py

State and Operations

Memory Banks mnemosyne/core/banks.py
Journal mnemosyne/core/journal.py
Runtime Configuration mnemosyne/core/config.py
Memory Importers mnemosyne/core/importers/__init__.py
Instance Sync mnemosyne/core/sync.py
Recall Diagnostics mnemosyne/core/recall_diagnostics.py
Model Refresh mnemosyne/core/model_refresh.py
User and Agent Context mnemosyne/core/user_agent.py
Health Diagnostics mnemosyne/diagnose.py

Other components

Agent or User

Connections 34

Agent or User → Python SDK: calls README.md
Python SDK → Memory API: stores or recalls mnemosyne/__init__.py
Agent or User → MCP Tools: uses README.md
MCP Tools → Memory API: accesses memory
Agent or User → Command Line: runs commands README.md
Command Line → Memory API: stores or recalls
Agent or User → Hermes Provider: uses provider README.md
Hermes Provider → Memory API: accesses memory integrations/hermes/src/mnemosyne_hermes/__init__.py
Agent or User → OpenWebUI Bridge: uses tool README.md
OpenWebUI Bridge → Memory API: calls
Agent or User → OpenClaw Provider: uses provider README.md
Agent or User → VS Code Extension: uses extension
Agent or User → Obsidian Plugin: uses plugin
Memory API → Write Filtering: filters writes mnemosyne/core/memory.py
Memory API → BEAM Recall: delegates recall mnemosyne/core/memory.py
Memory API → Dense Embeddings: uses vectors mnemosyne/core/memory.py
Memory API → Memory Banks: uses banks mnemosyne/core/memory.py
Memory API → Journal: records changes mnemosyne/core/memory.py
Memory API → Memory Triples: uses triples mnemosyne/core/memory.py
Memory API → Annotations: uses annotations mnemosyne/core/memory.py
Memory API → Content Canonicalization: uses canonicalization mnemosyne/core/memory.py
Memory API → Episodic Graph: uses graph hermes_memory_provider/__init__.py
Batch Operations → BEAM Recall: uses recall mnemosyne/batch_tool.py
Batch Operations → Veracity Consolidation: uses consolidation mnemosyne/batch_tool.py
BEAM Recall → Dense Embeddings: uses vectors
BEAM Recall → Recall Diagnostics: reports diagnostics
Canonical Facts → Memory Banks: shares memory store
Runtime Configuration → Memory API: configures runtime mnemosyne/core/memory.py
Memory Importers → Memory API: imports memories
Instance Sync → Memory API: syncs instances
Model Refresh → Runtime Configuration: reads settings mnemosyne/core/model_refresh.py
Model Refresh → Content Canonicalization: uses canonicalization mnemosyne/core/model_refresh.py
Dense Embeddings → User and Agent Context: identifies client mnemosyne/core/embeddings.py
Health Diagnostics → Dense Embeddings: checks embeddings

Drawn by GitDiagram on November 15, 2026 from the repository's file tree, README and source files. Read as Markdown ·