🧠 GraphMind

Semantic Graphs for Cross-System Code Understanding & Generation

Modern tech systems are sprawling — dozens of microservices, databases, cloud resources like File Storages, Key Vaults, and more. Even a tiny change often means updating multiple codebases and touching many interconnected systems.

This slows down dev team to build new ideas/products while catering to changes in the existing system. Changes to existing system takes away around 70% of Dev Effort in a large org.

GraphMind is designed to break that bottleneck.

---

🔍 What is GraphMind?

GraphMind is a semantic-aware code intelligence tool purpose-built for Golang/gRPC-based distributed systems.

It builds a unified semantic graph across all your code repositories, enabling LLMs to understand and modify your system from a single specification.

---

🧠 How It Works

1. Repo Analysis
   Parses multiple codebases to identify microservice entry points and APIs (especially gRPC in Golang).

2. Semantic ASTs via LLMs
   For each API/method, GraphMind builds an AST. Then it uses Claude to semantically annotate the AST — identifying not just what functions do, but why and what they touch.

   Example: AST may show an HTTP call — but Claude can infer the target service or resource from URLs or variable names, giving context that ASTs alone miss.

3. Semantic Graph Construction
   Merges all annotated ASTs into a unified Semantic Graph using rdflib. This cross-repo graph represents a complete view of your system: services, APIs, resources, and dependencies.

   ✅ The semantic graph construction has been successfully tested on the following real-world microservice repositories:

   • authGo
   • notificationGo
   • socialGo
   • localizationGo

4. LLM-Powered Code Generation
   With this graph, an LLM can:

   • Understand a natural language spec
   • Identify affected APIs/services/resources
   • Generate accurate multi-repo code changes
   • Output change summaries and Mermaid.js diagrams to depict code change path.

🛠️ Getting Started

git clone https://github.com/SaiNageswarS/GraphMind.git
cd GraphMind
# Set up Golang and Python environments
# Provide API keys for Claude/OpenAI in .env as per .env.template
# Run temporal server
temporal server start-dev
# build golang worker
.\build.ps1 
# Run worker
.\build\GraphMind
# Run python rdflib server
cd GraphMindPyAPIs
python main.py

📺 Demo

Click the image above to watch the full walkthrough!