Open-source AI memory platform for agents with self-hosted knowledge graph memory across sessions
- Stars17.8k
- Forks1.9k
- Open Issues77
Apache-2.0
- Python
- TypeScript
- JavaScript

About Cognee
Cognee is an open-source AI memory platform for AI agents. It gives agents persistent long-term memory across sessions by ingesting data and building a self-hosted knowledge graph, so agents can recall and use context beyond a single interaction.
Cognee combines vector embeddings, graph reasoning, and ontology generation grounded in cognitive science. Documents become searchable by meaning and connected by relationships that evolve with the knowledge base. Agents can use session memory, graph-backed recall, auto-routing across search strategies, and dataset deletion.
Cognee runs locally, self-hosted, in Docker, on-prem, or on Cognee Cloud. It installs as a Python package with pip, poetry, uv, or another Python package manager, and has plugins for OpenClaw and Claude Code. Research from Markovic et al. covers optimizing knowledge graph interfaces for LLM reasoning.
Key features
- Persistent long-term memory for AI agents across sessions
- Ingests data and builds a self-hosted knowledge graph
- Combines vector embeddings, graph reasoning, and ontology generation
- Session memory with background sync to the graph
- Auto-routing chooses a search strategy for recall queries
Details
- First released
- 2023
- Self-hosting
- Local · Docker · on-prem
- Cloud
- Cognee Cloud
- Runtime
- Python 3.10 to 3.14
- Retrieval
- Graph · vector · relational
- Integrations
- OpenClaw · Claude Code
