https://github.com/horheynm/code-agent-beta

Coding agent that recognizes your code, bugs and errors
https://github.com/horheynm/code-agent-beta

Last synced: 6 days ago
JSON representation

Coding agent that recognizes your code, bugs and errors

• Host: GitHub
• URL: https://github.com/horheynm/code-agent-beta
• Owner: horheynm
• License: apache-2.0
• Created: 2024-06-08T19:11:37.000Z (7 months ago)
• Default Branch: main
• Last Pushed: 2024-06-30T05:49:38.000Z (6 months ago)
• Last Synced: 2024-10-29T14:40:42.297Z (about 2 months ago)
• Language: Python
• Homepage:
• Size: 78.2 MB
• Stars: 1
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Component generator

Simple component generator using simplified agents. Generate react/next.js code from input prompt! 

![Demo](./static/demo.gif)

## How to run

Create `.env` file and enter your `OPENAI_API_KEY`, where the value should start with `sk-...`. Also include your github access token to push your project. 

Create `src/api/constant.py` and populate your constants for agents.

### Verify your tokens

Run the below commands to verify your github and openai tokens. 

```bash

 src/bin/validation/[(github|openai).sh]

```

## Commands

To create projects from scratch:

```python3

python3 './example/create_bash_script.py'

```

^ Generates bash script executable from LLM

To generate components frome existing framework:

```python3

python3 './example/completion.py'

```

^ Generates a "plan" and "executes" the plan to generate code according to the plan

Note: constants are not included in this repo. For usage please contact [email protected]

## How it works

### `create_bash_script.py'

`create_bash_script.py' asks LLM to generate a bash script executable to create, install and modify code. Generated bash script and all dependecies are carried out in a fresh docker container that only has node. 

### './example/completion.py'

Asks agents to plan, execute the plan (layout the files and the folder structure), and generate code each of the file to be created.