How gajini-prasanth/memori works
Memori is a memory layer for LLM applications, available through Python and TypeScript SDKs and agent integrations. The main workflow attributes an interaction, captures conversation or execution data, stores structured memories, then retrieves relevant memories for later model requests. The Rust core supplies shared augmentation, embedding, retrieval, and worker capabilities; storage can be host-provided or cloud-backed. The README documents OpenClaw, Hermes, and MCP integration capabilities. Some integration internals and unsampled modules are represented at directory level; edges are limited to README statements and indexed source references.
Main entry points: lib.rs Shared Rust engine · __init__.py Python SDK · _network.py Memori Cloud API · pipeline.rs Recall and ranking
Components 20
SDKs and API
- Python SDK
memori/__init__.py - TypeScript SDK
memori-ts/src/memori.ts - Attribution and config
memori/_config.py - LLM adapters
memori/llm/adapters
Memory processing
- Conversation capture
memori/memory/_manager.py - Advanced augmentation
memori/memory/augmentation/_manager.py - Shared Rust engine
core/src/lib.rs - Augmentation pipeline
core/src/augmentation/pipeline.rs - Background workers
core/src/runtime/worker.rs - Embeddings
core/src/embeddings/api.rs
Retrieval and storage
- Recall and ranking
core/src/retrieval/pipeline.rs - Lexical and similarity search
core/src/search - Storage managers
memori/storage/_manager.py - Database adapters and drivers
memori/storage/drivers - Memori Cloud API
memori/_network.py
Agent integrations
- OpenClaw plugin
integrations/openclaw/src/index.ts - Hermes provider
integrations/hermes/src/memori_hermes/__init__.py - MCP integration
integrations/claude-code/index.ts
Other components
- Application developer
- LLM provider
Connections 22
- Application developer → Python SDK: configures
README.md - Application developer → TypeScript SDK: configures
README.md - Python SDK → Attribution and config: sets context
memori/__init__.py - Python SDK → Conversation capture: captures interactions
memori/__init__.py - TypeScript SDK → Shared Rust engine: delegates memory work
- Conversation capture → Memori Cloud API: posts cloud messages
memori/memory/_manager.py - Conversation capture → Advanced augmentation: schedules memory work
- Advanced augmentation → Shared Rust engine: uses shared core
- Shared Rust engine → Augmentation pipeline: runs augmentation
core/src/lib.rs - Shared Rust engine → Background workers: starts jobs
core/src/lib.rs - Augmentation pipeline → Memori Cloud API: requests augmentation
core/src/augmentation/pipeline.rs - Shared Rust engine → Embeddings: creates embeddings
- Shared Rust engine → Recall and ranking: runs recall
core/src/lib.rs - Recall and ranking → Lexical and similarity search: ranks candidates
core/src/retrieval/pipeline.rs - Recall and ranking → Storage managers: fetches candidates
core/src/retrieval/pipeline.rs - Storage managers → Database adapters and drivers: selects drivers
memori/storage/_manager.py - Python SDK → Recall and ranking: recalls memories
memori/__init__.py - LLM adapters → LLM provider: invokes model
- Python SDK → LLM adapters: integrates providers
memori/__init__.py - OpenClaw plugin → Memori Cloud API: captures and recalls
README.md - Hermes provider → Memori Cloud API: captures and recalls
README.md - MCP integration → Memori Cloud API: connects agent clients
README.md