Awesome AI Agents › Multi-Agent Frameworks

MemoriLabs/Memori

⭐ 16695 Python added to this list on 2025-12-15 repository created 2025-07-24

Memori is a sophisticated SQL native memory layer designed specifically for large language models (LLMs), AI agents, and multi-agent systems. It acts as a memory fabric for enterprise AI, seamlessly integrating with existing software and infrastructure without being tied to any specific LLM, datastore, or framework. This agnostic design allows it to work with all major foundational models, databases, and document stores, making it highly versatile and adaptable to various AI architectures. The project offers significant performance improvements through advanced augmentation techniques and features a threaded, zero-latency extraction agent. It supports vectorized memories and in-memory semantic search, which enhance the accuracy and relevance of contextual information. Memori employs a third normal form schema that includes the storage of semantic triples, enabling the creation of a knowledge graph for richer data representation. Memori emphasizes ease of use and developer efficiency, reducing development overhead to a single line of code and providing automatic schema migrations. It supports session management to group LLM interactions, allowing for coherent tracking of multi-step agent executions. Attribution is a key concept in Memori, where interactions are linked to entities (such as users) and processes (such as AI agents), enabling the system to create meaningful memories. The project supports a wide range of LLMs including Anthropic, Bedrock, Gemini, Grok (xAI), and OpenAI, and integrates with popular frameworks like Agno and LangChain. It also supports various platforms such as Nebius AI Studio and multiple database integrations including DB API 2.0 compliant drivers, Django ORM, and SQLAlchemy. Supported datastores include CockroachDB, MariaDB, MongoDB, MySQL, Neon, Oracle, PostgreSQL, SQLite, and Supabase. Memori provides advanced augmentation capabilities that enrich memories with attributes, events, facts, people, preferences, relationships, rules, and skills, creating unparalleled context between users and AI agents. This augmentation runs asynchronously in the background, ensuring no latency impact. The project is open source, licensed under Apache 2.0, and offers a developer-friendly environment with extensive documentation, examples, and a supportive community.

https://github.com/MemoriLabs/Memori

advanced-augmentationagentagnoaiai-agentsaiagentanthropicasynchronous-augmentationattributionawesomebedrockchatgptcockroachdbdatastore-agnosticdb-api-2.0developer-friendlydjango-ormenterprise-aiframework-agnosticgeminigrokhacktoberfesthacktoberfest2025knowledge-graphlangchainllmlong-short-term-memorymariadbmemori-aimemorymemory-managementmongodbmulti-agent-systemsmysqlnebius-ai-studioneonopenaioraclepostgresqlpythonragsemantic-searchsession-managementsql-native-memory-layersqlalchemysqlitestate-managementsupabasevectorized-memories