Feature : Neo4j Search Tool Prototype

[Charan6924]

Overview

This PR adds a GraphRetriever class (backend/neo4j_search_tool.py) to enable the agent to query graph-structured data (NIFSTD ontology).

Why this is needed

As discussed in the roadmap, vector search often misses explicit relationships (e.g., Disease -> Affects -> Brain Region). This tool allows the agent to retrieve these direct connections to augment the context window.

Changes

• Added neo4j dependency.
• Implemented GraphRetriever with environment variable configuration.
• Added connection verification and error handling.
• Added Unit Tests: Included tests/test_neo4j_tool.py to ensure that the graph retrieval tool works.

[QuantumByte-01]

Clean prototype with good error handling and unit tests, and graph-based ontology search is a genuinely interesting direction for neuroscience dataset discovery. However there is no integration into the agent pipeline, no documentation on what Neo4j instance or dataset this targets (NIFSTD? Custom graph?), and neo4j is added as a core dependency when it should be optional. Please integrate it as an optional search tool in the agent, document the required graph schema/dataset, and make the dependency opt-in, then reopen.