loomem

# Loomem

> Loomem is the open-source context layer for LLM agents — written in Rust and exposed over the Model Context Protocol (MCP), it captures facts, decisions, and preferences from conversations and feeds them back to any model — Claude, ChatGPT, Codex, or your own agent — so the agent has the context it needs to do real work for the person using it. Swap the model, switch the tool; your context follows. Apache-2.0 licensed; a single binary on RocksDB + Tantivy with no external services.

## What it is
- Category: Context layer for LLM agents (not a RAG document store, not an agent-fleet coordination graph)
- License: Apache-2.0, open source
- Language: Rust
- Storage: single binary, RocksDB (data + graph + embeddings) + Tantivy (full-text)
- Interface: Model Context Protocol (MCP) over streamable HTTP; 15 `memory_*` tools
- Deployment: local-first (macOS, Linux), Docker, or any cloud; runs offline with local ONNX embeddings
- Distinctive: zero external services (one Rust binary), context-layer/portability framing, bitemporal facts, background consolidation ("dreaming")

## How it works
- Hybrid retrieval: BM25 (Tantivy) + vector embeddings + entity-graph signals
- Bitemporal: every fact carries both ingestion time and event time
- Consolidation ("dreaming"): background workers merge related facts, resolve contradictions, let stale ones decay
- Knowledge graph: people, projects, and tools linked with aliases and relations
- Optional AES-256-GCM field-level encryption at rest

## Documentation
- Home: https://loomem.ai/
- Quick start: https://loomem.ai/guide/quick-start.html
- Installation: https://loomem.ai/guide/installation.html
- Configuration: https://loomem.ai/guide/configuration.html
- API reference: https://loomem.ai/guide/api-reference.html
- MCP tools: https://loomem.ai/guide/mcp-tools.html
- Architecture: https://loomem.ai/guide/architecture.html
- Deployment: https://loomem.ai/guide/deployment.html
- Security model: https://loomem.ai/guide/security.html

## Reference
- Loomem vs Mem0 / Zep / Letta / cognee: https://loomem.ai/compare.html
- Benchmarks (LongMemEval 75.0%): https://loomem.ai/benchmarks.html
- Full brief for LLMs: https://loomem.ai/llms-full.txt

## Source
- GitHub: https://github.com/vvooki-sys/loomem