Scriptgpt NPM | npm.io

Why ScriptGPT? 🚀

Hand off feature development to an AI agent: Let it excel in what it does best while you focus on managing how everything works together.
Control assistance: Specify how much help you need. If a task is beyond the AI agents capability, you can take control and handle it yourself.
Built for TS/JS: Tailored to automatically install required libraries, test code, add comments, and more.
Overcome limitations of current AI-powered Coding tools:
- Tools like copilot struggles with effective code integration and lacks planning capabilities. You can use both tools together, copilot can help with writing code and you can offload specific project features to ScriptGPT
- GPT-Engineer and similar tools aren't yet suitable for large, complex projects, facing difficulties in accessing language-specific tools and bug fixing.

CLI Interaction: Command-line code generation with automated testing.
API Interaction: Currenty it utilizes OpenAI's API to generate code with plans to integrate additional AI providers.
Code Crafting: Transforms natural language into TS/JS code with provided specification.
File Output: Generated code is automatically saved within your project

To use scriptgpt in your project, install the package using npm:

npm install scriptgpt -d

Create a .env file in your project root and add your OpenAI API key:

OPENAI_API_KEY=your_api_key_here

Create the Configuration File
Create a file named scriptgpt.config.json in the root directory of your project.
Define an array of functions within this file, specifying the following properties for each function:
- name: The name of the function to be generated (string).
- description: A brief description of the function's purpose (string).
- parameters: An array of objects describing the function's input parameters, where each object has:
  - name: The name of the parameter (string).
  - type: The data type of the parameter (e.g., "number", "string", "array", "object") (string).
- return: An object describing the function's output value, with:
  - type: The data type of the output value (string).
Example scriptgpt.config.json:
```
{
    "functions": [
        {
            "name": "add",
            "description": "Adds two integers together and returns values",
            "parameters": [
                {
                    "name": "a",
                    "type": "int"
                },
                {
                    "name": "b",
                    "type": "int"
                }
            ],
            "return": "int"
        }
    ]
}
```
Generate the Functions
Run the following command in your terminal:
```
scriptgpt generate
```
This command will generate the JavaScript functions based on your gpt.json configuration and place them in the root/generated/functions.js file.

Automatic Library Install Functionality:
- Users can automatically install required libraries for generated code.
Agent Planning Before Writing Code:
- The agent plans its approach before generating code for improved organization.

RAG System Integration:
- Implement a RAG (Read, Ask, Generate) system for better library documentation utilization.
- Enable the agent to ask clarifying questions about the documentation.
Return Class in Generated Code:
- Modify code generation to include class information, not just functions.
User Feedback Mechanism:
- Implement a user feedback mechanism for code improvements.