Memori

Active
GitHub Python NOASSERTION

Description

Agent-native memory infrastructure that turns agent execution and conversation into structured, persistent state with an LLM-agnostic memory layer, MCP integration, and Python/TypeScript dual SDK support.

Key Features

  • Agent-native memory infrastructure converting conversations and execution into structured persistent state
  • LLM-agnostic design supporting OpenAI, Anthropic, and any model provider
  • Python/TypeScript dual SDK with MCP one-command integration for Claude Code/Cursor/Codex
  • 81.95% accuracy on LoCoMo benchmark using only 4.97% of full-context token footprint
  • Entity attribution system tracking memory provenance by user/agent/process
  • Memori Cloud zero-config deployment and BYODB self-hosted database modes

Use Cases

💡 Cross-session persistent memory and context recovery for AI agents
💡 Long-term tracking of user preferences and interaction history in customer service
💡 Shared coding conventions and project agreements across development teams
💡 Shared state and decision records in multi-agent collaboration
💡 Reducing LLM context window costs while maintaining reasoning quality

Tags

memory agent rag framework python typescript

Categories

🧩 Agent Memory

Quick Start

```bash
pip install memori
# or
npm install @memorilabs/memori
```

Set `MEMORI_API_KEY` and `OPENAI_API_KEY` environment variables, then wrap your LLM client with Memori:
```typescript
import { Memori } from '@memorilabs/memori';
const mem = new Memori().llm.register(client).attribution('user_123', 'support_agent');
```
Conversations are automatically persisted and recalled. MCP integration: `claude mcp add --transport http memori https://api.memorilabs.ai/mcp/`
Visit GitHub Visit Website

Related Projects