Fixgpt NPM | npm.io

fixgpt

Tool that helps you do mass changes across your codebase

Create new files
Mass refactoring
Remove files upon certain conditions
Code execution

Watch a demo

Code generation - link
Mass Update - link
Parameters - link

Installation

Install from NPM:

npm i fixgpt -g

Or just use NPX:

npx fixgpt <command> <arg1> <arg2>

Before use

fixgpt will go over your files and will edit them according your request.

It is recommeneded to use GIT (or other VCS) and check the diff of changes yourself, once fixgpt is done.

It is necessary as LLM models are not 100% safe at producing reliable code.

How to use?

1. Initialize using your OpenAI access token

fixgpt init <openai_token>

init command creates a local .fixgpt folder with config.

CAUTION: Make sure to add it to .gitignore file of your project in order to not leak your API keys

2. Create a template

fixgpt create-template ./template.md

This will create a Markdown file with a template which then can be used in run-template command

3. Run it

fixgpt run-template ./template.md

Then grab some coffee and see it working :)

NOTE: OpenAI doesn't allow running multiple requests in parallel, so if you have like 100 files, it will be slow. It can be cheated by using multiple accounts but currently it's not supported

PRO Stuff

Tokens

You can put specific tokens into a template in order to ask for some data or use previous responses.

[[args.input.ArgumentName]] - asks for ArgumentName before execution
[[ask.input.ArgumentName]] - asks for ArgumentName during execution (unlike the previous one, this one will be asked separately for each file in "Mass Update")
[[responses.0.code]] - code from task response (where 0 is an index of the task)
[[responses.0.file]] - filename from task (where 0 is an index of the task)