aminroosta/openai.vim

openai.vim

A Neovim plugin that exposes OpenAI’s Chat Completion API.

Installation

3. Using packer.nvim, add the following code to your init.lua file:

require('packer').startup(function()
  use {
    "aminroosta/openai.vim",
    requires = {
      "nvim-lua/plenary.nvim",
      "MunifTanjim/nui.nvim",
    }
  }
end)

Installation Ollama

1. Install ollama and pull a model to be used: ollama pull llama3.2
2. Export the following variables to your ~/.bashrc

export OPENAI_API_KEY=''
export OPENAI_CHAT_MODEL='llama3.2'
export OPENAI_ENDPOINT='http://127.0.0.1:11434/api/chat'

Installation OpenAI

1. Add your OPENAI_API_KEY to the environment variables by running the following command in the terminal:

export OPENAI_API_KEY='your-api-key-goes-here'
export OPENAI_CHAT_MODEL='gpt-4o-mini'

Installation LMStudio

1. Install lmstudio, and download a model to be used: lmstudio-community/Llama-3.2-3B-Instruct-GGUF.
2. Start the server from the "Local Server" tab in the app.
3. Export the following variables to your ~/.bashrc

export OPENAI_API_KEY=''
export OPENAI_CHAT_MODEL='lmstudio-community/Llama-3.2-3B-Instruct-GGUF'
export OPENAI_ENDPOINT='http://127.0.0.1:1234/v1/chat/completions'

Configuration using JSON

To customize your commands, create or modify the ~/.config/nvim/commands.json file. This file should contain your command configurations in JSON format.
For information on structuring your commands, refer to OpenAI's Chat Completion API documentation.

Here is an example, the TEXT keyword is replaced by the visual selection in vim.

{
  "ask": [
    {
      "role": "user",
      "content": "TEXT"
    }
  ],
  "jsdoc": [
    {
      "role": "user",
      "content": "Convert this to JSDoc format:\nTEXT"
    }
  ],
  "rewrite": [
    {
      "role": "user",
      "content": "Fix grammatical mistakes and reorder sentences if needed:\n\nTEXT"
    }
  ],
  "eng": [
    {
      "role": "user",
      "content": "Rewrite this in natural english:\n\nTEXT"
    }
  ]
}

Usage

The plugin exposes a visual mode command :Openai that takes an argument, the special argument list shows a popup.

:'<,'>Openai list

The first argument references the key in your ~/.config/nvim/commands.json file.

:'<,'>Openai ask
:'<,'>Openai jsdoc
:'<,'>Openai rewrite
:'<,'>Openai eng

Configuration using yaml

If you find .yaml files easier to modify and maintain, create ~/.config/nvim/commands.yml file, and manually convert it to json.

# cat ~/.config/nvim/commands.yml | yq -o json > ~/.config/nvim/commands.json
jsdoc:
  - role: "user"
    content: "Convert this to JSDoc format:\nTEXT"
ask:
  - role: "user"
    content: "TEXT"
rewrite:
  - role: "user"
    content: "Fix grammatical mistakes and reorder sentences if needed:\n\nTEXT"
eng:
  - role: "user"
    content: "Rewrite this in natural english:\n\nTEXT"

You'd need to brew install yq for that conversion to work.