An AI-powered project migration and refactoring backend engine. NeuroDev uses Anthropic's Claude models via AWS Bedrock and LangChain/LangGraph to automate complex software development tasks.
NeuroDev (Internal: claude-poc-backend) is designed to handle high-level AI tasks such as project-wide refactoring, migrating codebases between languages, and generating new project structures from scratch. The codebase has been modernized to follow industry-standard asynchronous patterns and clean architecture.
- Project Migration: Automatically convert codebases (e.g., PL/SQL to Java/Spring Boot).
- Intelligent Refactoring: Optimize and refactor existing projects via target-state prompts.
- Project Scaffolding: Build new projects from natural language descriptions.
- Model Benchmarking: Test and compare AI model accuracy for specific coding tasks.
- Standardized Utilities: Robust file and workspace management tools for AI processing.
- Runtime: Node.js (Express)
- AI Orchestration: LangChain & LangGraph
- LLM Connectivity: AWS Bedrock (Claude 3 Haiku/Sonnet)
- File Processing:
express-fileupload,unzipper,archiver
# Clone the repository
git clone <repository-url>
cd NeuroDev
# Install dependencies
npm installCreate a .env file in the root directory based on the provided template:
PORT=3002
WORKSPACE_DIR=./workspace
BEDROCK_REGION=ap-south-1
BEDROCK_MODEL_ID=anthropic.claude-3-haiku-20240307-v1:0
CLAUDE_TEMPERATURE=0.2
CLAUDE_TOP_P=0.9
CLAUDE_MAX_TOKENS=4000# Start the development server
npm start| Method | Endpoint | Description |
|---|---|---|
| POST | /api/send-prompt |
Primary endpoint for Migrate/Refactor/Build tasks |
| POST | /api/upload-zip |
Upload project ZIPs to the workspace |
| POST | /api/import-git |
Clone repositories into the workspace |
| GET | /api/list-files |
List files currently in the AI workspace |
| POST | /api/test-models |
Benchmarking tool for LLM accuracy |
AI operations occur within the ./workspace directory. Post-processing results (like migrated code) are typically saved to ./workspace/migrated-project.
Modernized for better performance and maintainability.