🧠 Adaptive Graph of Thoughts

What is 🧠 Adaptive Graph of Thoughts?

Adaptive Graph of Thoughts (AGoT) is a high-performance MCP server that implements the Advanced Scientific Reasoning Graph-of-Thoughts (ASR-GoT) framework. It uses a Neo4j graph database as a dynamic knowledge store and exposes reasoning capabilities through the Model Context Protocol (MCP), enabling seamless integration with AI assistants like Claude Desktop.

How to use 🧠 Adaptive Graph of Thoughts?

Clone the repository, install dependencies with Poetry, run the interactive setup wizard (poetry run python -m agt_setup), then start the server with Uvicorn (poetry run uvicorn adaptive_graph_of_thoughts.main:app --reload). For production, use Docker Compose or a Kubernetes Helm chart. Connect via MCP clients using the /mcp endpoint with Bearer authentication.

Key features of 🧠 Adaptive Graph of Thoughts

• 8-stage graph-based reasoning pipeline with dynamic confidence scoring
• Real-time evidence integration from PubMed, Google Scholar, and Exa Search
• High-performance async FastAPI server backed by Neo4j
• Native MCP support for Claude Desktop and VS Code
• Full Docker and Kubernetes (Helm) deployment support
• Interactive setup wizard for credentials and configuration

Use cases of 🧠 Adaptive Graph of Thoughts

• Multi-step scientific question answering with evidence-backed reasoning
• Hypothesis generation, pruning, and merging from research literature
• Construction and exploration of knowledge graph connectomes from queries
• Synthesis of conclusions with quantified confidence and audit trails
• Cloud-native deployment for scalable scientific AI reasoning

FAQ from 🧠 Adaptive Graph of Thoughts

What are the runtime dependencies for 🧠 Adaptive Graph of Thoughts?

A running Neo4j instance with the APOC library installed is required, along with Python 3.11+, Poetry, and the dependencies defined in pyproject.toml.

How is authentication handled?

All MCP API endpoints require Bearer token authentication. A Bearer token is verified by the Auth Middleware before any request reaches the GoT Processor.

Which external data sources does 🧠 Adaptive Graph of Thoughts integrate with?

It integrates with PubMed, Google Scholar, and Exa Search for real-time evidence retrieval during the reasoning pipeline.

Can 🧠 Adaptive Graph of Thoughts be deployed in containers or cloud environments?

Yes. The project provides a Dockerfile, a production docker-compose.prod.yml, and a Kubernetes Helm chart in the helm/ directory for cloud-ready deployment.

Where does reasoning data live?

All reasoning state (nodes, relationships, sessions) is stored in a Neo4j graph database, either local or remote, as configured in config/settings.yaml or environment variables.