Yggtopoteretes/cognee
↗ GitHubKnowledge Engine for AI Agent Memory in 6 lines of code
14,847
Stars
1,494
Forks
62
Watchers
74
Open Issues
Safety Rating A
No hardcoded secrets, malicious code patterns, suspicious dependencies, or prompt injection attempts were detected. The repository is a well-structured, actively maintained open-source project with 14,000+ stars, a clear Apache 2.0 license, published research citations, and standard Python packaging. The README contains only legitimate product documentation and code examples.
ℹAI-assisted review, not a professional security audit.
AI Analysis
Cognee is an open-source knowledge engine for building persistent, dynamic AI agent memory. It ingests data in any format, processes it through a pipeline that combines vector embeddings and graph databases, and continuously learns to surface contextually relevant information for AI agents. It supports local and cloud deployment, multiple LLM providers, MCP integration, and exposes both a Python API and a CLI.
Use Cases
- Building persistent long-term memory for AI agents using knowledge graphs and vector search
- Ingesting and indexing documents, structured data, or unstructured text into a queryable knowledge engine
- Enabling personalized AI assistants that learn from feedback and cross-agent knowledge sharing
- Augmenting LLM responses with graph-based retrieval (GraphRAG) for improved reasoning
- Deploying a self-hosted or managed knowledge infrastructure for enterprise AI pipelines
Tags
Project Connections
MemoryOS
→Both MemoryOS and Cognee provide persistent memory systems for AI agents with vector storage and LLM integration, but Cognee additionally emphasizes knowledge graph (GraphRAG) construction and ontology grounding, while MemoryOS focuses on OS-inspired hierarchical memory tiers.
agentfield
→AgentField is a backend control plane for deploying and operating AI agents with distributed memory and vector search. Cognee could serve as the knowledge engine layer powering AgentField's memory and RAG capabilities.
clawguard
→ClawGuard monitors AI agent activity for security purposes. Cognee's knowledge engine could be integrated as the memory/context backend for the agents ClawGuard monitors, providing auditability and traceability of agent memory state.
ClawWork
→ClawWork benchmarks LLM-based agents on professional tasks. Cognee could enhance ClawWork agents with persistent memory and domain knowledge retrieval, improving task performance across sessions.
Decepticon
→Decepticon orchestrates autonomous multi-agent red team pipelines. Cognee could provide persistent knowledge storage and retrieval across agent runs, enabling the system to learn from prior engagements.