Cognee

Cognee is an open-source AI memory platform that provides LLM agents with persistent, long-term memory through graph, vector, and relational retrieval methods.

Cognee functions as an open-source memory engine for large language model agents, enabling them to retain project context across sessions. It integrates data from various sources like Slack, GitHub, Linear, Google Drive, and Notion into a single recallable layer, allowing agents to access cited facts and follow domain-specific rules. The platform supports graph, vector, and relational retrieval methods to connect and index information. Users can deploy it self-hosted, via Docker, on-premise, or through the company's cloud service. It includes first-party integrations for Claude Code, Cursor, LangGraph, and OpenClaw, and provides an MCP server for broader agent compatibility. This tool creates a shared memory layer so different agents can search and retrieve information from a unified knowledge base.

Cognee at a glance

Best for: Software engineer
Runs on: open source, self-hosted, API
Pricing: Free plan with 1M tokens; Standard plan adds workspaces at $5/month; Enterprise BYOC engagement with custom pricing.
Released: August 2023
Category: AI agent tool
Website: cognee.ai

Key features

Persistent long-term memory for AI agents
Graph, vector, and relational data retrieval
Integrations with Slack, GitHub, Linear, and Notion
Support for Claude Code, Cursor, and MCP clients
Self-hosted and on-premise deployment options
Code indexing for software repositories
Personalized memory per user and agent

Use it for

Giving AI agents durable memory across sessions
Connecting internal company tools for agent retrieval
Building custom knowledge bases of domain regulations
Reducing token waste by recalling previous task solutions

Alternatives to Cognee

claude-mem — It shares persistent, searchable agent memory across sessions and devices.
ReMe — It enables local-first, file-native long-term memory sharing across multiple AI agents.
Maximem — It delivers fast retrieval and structured memory across frameworks.
MemOS — It delivers ultra-persistent memory, hybrid-retrieval, and token savings.
Memori — Reduces LLM token usage and spend by over 97% with targeted recall.
Supermemory — Integrates long-term memory into AI coding agents with a single prompt.
MemMachine — Provides long-term persistent memory for LLM-based agents.
M-Flow — It achieves a 93.7 percent top-one hit rate using a conical memory architecture.
Tigerless Agent Memory — Maintains agent history in portable, greppable Markdown files without API keys.
OpenViking — Unifies agent memory, knowledge RAG, and skills into a single directory structure.
memU — It gives every connected agent a shared and inspectable memory layer.
Agentmemoryrepo — Enables agents to navigate memories like a wiki and share knowledge across sessions.

All Cognee alternatives