AI Code Repository Assistant

A full-stack agentic AI application for understanding and navigating codebases, built with .NET 8 and React/TypeScript.

Features

• RAG Pipeline: Semantic code search using Azure Cosmos DB vector search
• Local LLM Support: Runs with Ollama (qwen2.5-coder) for local development
• Grounding: Source citations with file:line references
• Streaming Responses: Real-time streaming chat interface
• CI/CD: GitHub Actions pipeline for build and test

Screenshots

Local LLM with Ollama

Swagger API Documentation

Frontend Chat Interface

Azure Cosmos DB Data

GitHub Actions CI Pipeline

Architecture

┌─────────────────┐     ┌─────────────────────────────────────────┐
│  React Frontend │────▶│           .NET 8 Web API                │
│  (TypeScript)   │     │  ┌─────────────────────────────────┐   │
└─────────────────┘     │  │      Agent Orchestrator         │   │
                        │  │  (ReAct Loop with Tool Calls)   │   │
                        │  └──────────────┬──────────────────┘   │
                        │                 │                       │
                        │  ┌──────────────┴──────────────────┐   │
                        │  │            Tools                 │   │
                        │  │ • Code Search  • File Read       │   │
                        │  │ • Explain Code • Find References │   │
                        │  └──────────────┬──────────────────┘   │
                        │                 │                       │
                        │  ┌──────────────┴──────────────────┐   │
                        │  │         RAG Pipeline             │   │
                        │  │ • Document Chunker               │   │
                        │  │ • Embedding Service              │   │
                        │  │ • Vector Search Service          │   │
                        │  └─────────────────────────────────┘   │
                        └───────────────────┬─────────────────────┘
                                            │
                        ┌───────────────────┴───────────────────┐
                        │                                       │
              ┌─────────┴─────────┐               ┌─────────────┴─────────┐
              │  Azure Cosmos DB  │               │   Azure OpenAI        │
              │  (Vector Search)  │               │  (GPT-4 + Embeddings) │
              └───────────────────┘               └───────────────────────┘

Tech Stack

Backend

• .NET 8 Web API
• Azure.AI.OpenAI SDK
• Microsoft.Azure.Cosmos SDK
• Microsoft Semantic Kernel (optional)

Frontend

• React 18 + TypeScript
• Vite
• TailwindCSS
• React Query

Infrastructure

• Azure Bicep
• Docker Compose (local development)

Getting Started

Prerequisites

• .NET 8 SDK
• Node.js 20+
• Docker (for local Cosmos DB emulator)
• Azure CLI (for deployment)
• Azure subscription with OpenAI access

Local Development

1. Clone the repository

   git clone <repository-url>
   cd azure-ai-code-agent

2. Configure environment

   Update src/CodeAgent.Api/appsettings.json with your Azure OpenAI credentials:

   {
     "AzureOpenAI": {
       "Endpoint": "https://your-resource.openai.azure.com/",
       "ApiKey": "your-api-key",
       "ChatDeployment": "gpt-4",
       "EmbeddingDeployment": "text-embedding-ada-002"
     }
   }

3. Start local services with Docker Compose

   docker-compose up -d cosmosdb

4. Run the API

   cd src/CodeAgent.Api
   dotnet run

5. Run the frontend

   cd src/code-agent-ui
   npm install
   npm run dev

6. Open the application

   • Frontend: http://localhost:5173
   • API Swagger: http://localhost:5000/swagger

Indexing a Repository

Use the API to index a code repository:

curl -X POST http://localhost:5000/api/ingestion/repositories \
  -H "Content-Type: application/json" \
  -d '{
    "path": "/path/to/your/repository",
    "name": "My Project",
    "description": "Description of the project"
  }'

Azure Deployment

1. Login to Azure

   az login

2. Deploy infrastructure

   cd infra
   ./deploy.sh

3. Deploy the API

   cd src/CodeAgent.Api
   dotnet publish -c Release
   az webapp deploy --resource-group codeagent-rg \
     --name <app-service-name> \
     --src-path bin/Release/net8.0/publish

Project Structure

azure-ai-code-agent/
├── src/
│   ├── CodeAgent.Api/              # .NET 8 Web API
│   │   ├── Controllers/            # API endpoints
│   │   ├── Services/
│   │   │   ├── Agent/              # Agentic AI components
│   │   │   │   ├── Tools/          # Agent tools
│   │   │   │   └── AgentOrchestrator.cs
│   │   │   ├── Rag/                # RAG pipeline
│   │   │   └── Grounding/          # Citation service
│   │   ├── Models/                 # Data models
│   │   └── Infrastructure/         # Azure clients
│   │
│   └── code-agent-ui/              # React/TypeScript Frontend
│       └── src/
│           ├── components/         # UI components
│           ├── hooks/              # Custom hooks
│           └── services/           # API client
│
├── infra/                          # Azure Bicep templates
│   ├── main.bicep
│   └── modules/
│
├── tests/                          # Test projects
├── docker-compose.yml
└── README.md

API Endpoints

Agent

• POST /api/agent/chat - Send a message and get a response
• POST /api/agent/chat/stream - Send a message with streaming response
• GET /api/agent/conversations/{id} - Get conversation history
• DELETE /api/agent/conversations/{id} - Clear conversation

Ingestion

• GET /api/ingestion/repositories - List indexed repositories
• POST /api/ingestion/repositories - Index a new repository
• GET /api/ingestion/repositories/{id} - Get repository details
• DELETE /api/ingestion/repositories/{id} - Delete repository index

Running Tests

cd tests/CodeAgent.Api.Tests
dotnet test

Environment Variables

Variable	Description
AzureOpenAI__Endpoint	Azure OpenAI endpoint URL
AzureOpenAI__ApiKey	Azure OpenAI API key
AzureOpenAI__ChatDeployment	GPT-4 deployment name
AzureOpenAI__EmbeddingDeployment	Embedding model deployment name
CosmosDb__ConnectionString	Cosmos DB connection string
CosmosDb__DatabaseName	Database name (default: CodeAgentDb)
Frontend__Url	Frontend URL for CORS

License

MIT