TrustGraph is an open-source Context Operating System that enables organizations to build, manage, and deploy intelligent context graphs. Transform fragmented enterprise data into AI-optimized context structures that power accurate, trustworthy AI agents.
- Build Context Graphs — Automatically extract entities, relationships, and knowledge structures from your enterprise data
- Manage Context — Organize, version, and govern your context graphs with enterprise-grade tools and custom ontologies
- Deploy Intelligent Agents — Run AI agents grounded in your own precise context with full visibility and control
- Maintain Full Sovereignty — Keep your data and AI stack entirely under your control, deployed on-prem, in the cloud, or on bare metal
- Automated Entity & Relationship Extraction — AI-powered agents automatically identify key concepts and connections in your data
- Ontology-Driven Graphs — Define what should be extracted, not just what can be extracted, for consistent, controlled knowledge representation
- Multi-Format Data Support — Process PDFs, documents, databases, APIs, and structured data sources simultaneously
- Vector Embedding Integration — Automatic semantic embeddings mapped to graph relationships for hybrid retrieval
- Context Cores — Package and version your processed context for reuse across projects and deployments
- Collections — Organize knowledge by domain, project, or dataset with enterprise governance controls
- Flow Configuration — Design flexible data processing pipelines with runtime control and prompt management
- Observability & Telemetry — Monitor processing status, costs, performance, and agent behavior in real-time
- GraphRAG Queries — Intelligent retrieval combining graph structure and semantic search for deep contextual understanding
- Agentic Workflows — Build sophisticated agents that understand relationships, perform reasoning, and make decisions based on your knowledge
- Model Context Protocol (MCP) — Connect agents to external tools, APIs, and services while maintaining grounded context
- Multi-Model Support — Deploy local open-source models or connect to Anthropic, OpenAI, Google, Mistral, and other LLM providers
- Production Deployment — Kubernetes-native, fully containerized, ready for enterprise scale
- Cost Observability — Real-time tracking of token usage, inference costs, and resource consumption
- Access Controls & Secrets Management — Enterprise security with fine-grained permissions and credential handling
- Flexible Storage — Graph databases (Neo4j, Cassandra, Memgraph), vector stores (Qdrant, Pinecone, Milvus), and support for structured data
Anthropic Claude • OpenAI • Google AI Studio • Google VertexAI • Mistral • Cohere • AWS Bedrock • Azure OpenAI
Ollama • LM Studio • vLLM • Hugging Face TGI • Llamafiles
Qdrant • Pinecone • Milvus
Neo4j • Apache Cassandra • Memgraph • FalkorDB
AWS • Azure • Google Cloud • OVHcloud • Scaleway
Prometheus • Grafana
Model Context Protocol (MCP) for seamless agent integration with external APIs and tools
- Have Questions? Join our Discord
- Found a Bug? Open an issue
- Need Help? Check the documentation
- Ready to Contribute? See the contributing guide