A multi-LLM code graph agent for reading GitHub repos

nexusGPT is a code analysis agent that takes a link, copies the full repository, turns files into , and stores code relationships for later search and reasoning. It uses multiple LLMs and also supports , so different users can share the system while keeping their work separated. The process starts when a user enters a repository link in the interface.

The system clones the repo, converts each file into a , and sends the code through tree-sitter to detect the and find code parts and relationships. It then runs SCIP to extract more exact code nodes and links. If SCIP fails, the tree-sitter output remains as a backup.

The final information is stored in both a vector store and a Neo4j . The system is tied together with LangGraph and connects several free with MCP tools.

Key points

  • A GitHub repo link starts the pipeline, and the system clones the full repository.
  • Each file is converted into a before deeper analysis.
  • tree-sitter detects the language and extracts code parts and relationships.
  • SCIP tries to extract more precise code nodes and links; tree-sitter data is kept if SCIP fails.
  • Results are stored in both a vector store and a Neo4j .
Read original