Codebotica.ai : LangGraph-based Code with predefined Template Generation Workflow

Overview

This project demonstrates a basic workflow using LangGraph, where an LLM (Language Learning Model) assists in generating a project template based on user input. The workflow also ensures that the generated template and code files are valid, free from hallucinations, and optimized to meet user requirements.

Workflow Process

1. User Input:

   • The process starts with the user providing input about what the desired program is supposed to accomplish.

2. Template Extraction:

   • The AI agent extracts the template on which the project is supposed to be built.

3. Template Validation:

   • The LLM cross-checks whether the returned template is valid for the given input.
   • If the template is invalid or wrong, the LLM rephrases the query while maintaining the context and actual requirements.
   • A counter is in place to avoid getting stuck in an infinite loop of wrong template generation.

4. File Structure Skeleton Creation:

   • Once the template is validated, the LLM generates the skeleton of the project with the complete file structure, listing all extra files required inside the template.

5. File Content Generation:

   • Each file and its context are listed one by one, and the code generation LLM (e.g., Code Llama or fine-tuned Llama 3.1) is used to create the file contents.

6. Hallucination and Error Detection:

   • The content of each file is fed back to the LLM to cross-check for errors or hallucinations in the code or text.

7. Final Output:

   • Once validated, the content is written to the corresponding files within the generated project structure.

Project Structure

The following is a general example of the project structure that the LLM will generate:

root/
├── main_project/
│   ├── file1.py
│   ├── file2.py
│   └── ... (other project files)
├── README.md
└── requirements.txt

• main_project/: Contains the core Python files or other necessary project files.
• README.md: Provides an overview and steps to set up the project.
• requirements.txt: Contains dependencies for the project.

Installation

1. Clone this repository:

git clone https://github.com/sksarvesh007/codebotica.ai.git
cd codebotica.ai

2. Install the required packages:

pip install -r requirements.txt

Usage (nhi bana abhi tak :\ so ignore the below commands )

1. Provide Input: Start by describing the program you want the LLM to generate. The AI will extract the template for you.
2. Verify the Template: The LLM will validate the generated template based on your input. If it finds errors, it will rephrase the query and retry.
3. Generate File Structure: After validation, the file structure of the project is generated.
4. Generate and Validate File Content: The LLM generates content for each file and checks for errors or hallucinations.
5. Review and Run: Once the files are written, review the generated code and run the project.

Features

• AI-assisted template generation: Automates the process of creating a project skeleton based on user input.
• Error and Hallucination Checking: Ensures that the generated content is valid and free of hallucinations.
• Loop Prevention: Counter mechanism avoids infinite loops when the wrong template is repeatedly generated.

Future Work

• Integration with more advanced LLM models for better context understanding.
• Further automation of the deployment process.

Contributing

1. Fork the repository.
2. Create a new branch.
3. Make your changes.
4. Submit a pull request.