Memori: Give AI Agents Persistent Memory

Overview

Memori is a memory layer for LLM applications and agents. It turns conversations and agent execution into structured, persistent state, helping agents recall useful context rather than starting each interaction from scratch.

The project offers Python and TypeScript SDKs, integrations for agent frameworks and tools, and options for managed cloud or using your own database. It is aimed at developers and organizations building production agents that need continuity across sessions without replacing their existing model or data infrastructure.

Key Features

• Captures and recalls memory from LLM interactions, with attribution by entity and process.
• Tracks sessions and supports explicit session management.
• Structures memory as attributes, events, facts, people, preferences, relationships, rules, and skills.
• Provides Python and TypeScript SDKs, with integrations for supported LLMs and frameworks.
• Offers agent integrations including OpenClaw, Hermes Agent, and MCP-compatible clients.
• Supports Memori Cloud and a bring-your-own-database option.
• Performs memory augmentation in the background, according to the project documentation.

Use Cases

• Add continuity to customer-support agents that need to remember user preferences between conversations.
• Help coding agents recall project conventions and developer preferences across sessions.
• Preserve execution context, tool calls, and outcomes for agents that perform multi-step workflows.
• Give teams a shared memory layer for production agents operating within existing infrastructure.

Getting Started

Install the Python SDK:

pip install memori

The quickstart uses a Memori API key and an LLM API key. See the README and Memori documentation for configuration and other integration options.

Alternatives

• claude-mem: Claude-Mem focuses on retaining coding-agent project history across sessions, while Memori provides a general memory layer for LLM apps and agents.
• ReMe: ReMe stores searchable, editable Markdown memories for agent runtimes, while Memori captures structured execution and conversation memory across apps.

Considerations

• The basic SDK flow depends on external LLM and Memori credentials. Attribution is required for Memori to create memories.
• The README describes Memori Cloud as managed and documents BYODB separately; choose a deployment and data configuration appropriate for your system.
• Integrations and supported models vary. Check the project documentation for current compatibility before adopting it.
• The supplied license field is NOASSERTION, although the README states Apache 2.0. Verify the repository's license file and terms for your intended use.