NexusMind-2.0 is an AI reasoning framework that uses a Neo4j graph database to perform advanced scientific reasoning through a graph-of-thoughts approach (ASR-GoT). The MCP integration allows AI clients (such as Claude Desktop or other MCP-enabled applications) to send structured prompts and receive reasoning traces, results, and confidence scores generated by the system. Core capabilities include graph-based query processing, dynamic multi-dimensional confidence evaluation, modular reasoning stages, and a containerized deployment path via Docker for easy setup. To get started with usage, run the FastAPI/ASR-GoT server locally or in a container, ensure Neo4j with APOC is accessible, and connect via an MCP client. The server exposes endpoints to submit scientific queries, retrieve intermediate reasoning steps, and obtain final conclusions along with context graphs.

Prerequisites:

• Python 3.11+ and Poetry for dependency management
• Neo4j with APOC installed and reachable (bolt URI, username, password)
• Docker (optional for containerized deployment)

Local installation steps:

1. Clone the repository git clone https://github.com/SaptaDey/NexusMind.git cd NexusMind

2. Install Python dependencies (via Poetry) poetry install

3. Configure environment (example):

   • Copy example config if needed and edit settings.yaml under config/
   • Create a .env file for sensitive settings (optional)

4. Run the server (development):

   From project root

   poetry shell python src/asr_got_reimagined/main.py

5. Alternative development server (FastAPI/uvicorn): uvicorn asr_got_reimagined.main:app --reload --host 0.0.0.0 --port 8000

6. Docker deployment (optional): docker compose -f docker-compose.yml up -d

   Ensure Neo4j service in docker-compose has APOC enabled

Tips and troubleshooting:

• Ensure Neo4j APOC library is installed; many reasoning steps rely on APOC procedures.
• If the server fails to start, verify that NEO4J_URI, NEO4J_USERNAME, and NEO4J_PASSWORD are correctly configured and that the database is reachable from the host.
• The MCP integration expects a structured prompt/response flow; use the MCP client to submit queries and parse reasoning traces when debugging.
• For production, prefer Docker Compose with docker-compose.prod.yml, which includes a Neo4j service pre-configured for NexusMind.
• Environment variables such as LOG_LEVEL, API_HOST, and API_PORT can be tuned for deployment environments.
• If migrating from development to production, consider securing API endpoints and enabling authentication for MCP clients.