Home

Awesome

ChatGPT.nvim

GitHub Workflow Status Lua

ChatGPT is a Neovim plugin that allows you to effortlessly utilize the OpenAI ChatGPT API, empowering you to generate natural language responses from OpenAI's ChatGPT directly within the editor in response to your inquiries.

preview image

Features

For a comprehensive understanding of the extension's functionality, you can watch a plugin showcase video

Installation

The OpenAI API key can be provided in one of the following two ways:

  1. In the configuration option api_key_cmd, provide the path and arguments to an executable that returns the API key via stdout.

  2. Setting it via an environment variable called $OPENAI_API_KEY.

Custom OpenAI API host with the configuration option api_host_cmd or environment variable called $OPENAI_API_HOST. It's useful if you can't access OpenAI directly

Custom cURL parameters can be passed using the configuration option extra_curl_params. It can be useful if you need to include additional headers for requests:

{
  ...,
  extra_curl_params = {
    "-H",
    "Origin: https://example.com"
  }
}

For Azure deployments, you need to specify the URL base, the engine, and the API type. You can accomplish this in one of two ways:

  1. Use the configuration options api_type_cmd, azure_api_base, azure_api_engine_cmd, and azure_api_version_cmd. Each of these should be an executable command that returns the corresponding value.

For example:

  local config = {
    api_host_cmd = 'echo -n ""',
    api_key_cmd = 'pass azure-openai-key',
    api_type_cmd = 'echo azure',
    azure_api_base_cmd = 'echo https://{your-resource-name}.openai.azure.com',
    azure_api_engine_cmd = 'echo chat',
    azure_api_version_cmd = 'echo 2023-05-15'
  }

  require("chatgpt").setup(config)
  1. Set the values via the environment variables $OPENAI_API_TYPE, $OPENAI_API_BASE, $OPENAI_API_AZURE_ENGINE, and $OPENAI_API_AZURE_VERSION.

For example:

export OPENAI_API_TYPE="azure"
export OPENAI_API_BASE="https://{your-resource-name}.openai.azure.com"
export OPENAI_API_AZURE_ENGINE="chat"
export OPENAI_API_AZURE_VERSION="2023-05-15"

Please note that edit models have been deprecated and may not function as expected.

If you are using packer.nvim as plugin manager:

-- Packer
use({
  "jackMort/ChatGPT.nvim",
    config = function()
      require("chatgpt").setup()
    end,
    requires = {
      "MunifTanjim/nui.nvim",
      "nvim-lua/plenary.nvim",
      "folke/trouble.nvim",
      "nvim-telescope/telescope.nvim"
    }
})

or if you are using lazy.nvim:

-- Lazy
{
  "jackMort/ChatGPT.nvim",
    event = "VeryLazy",
    config = function()
      require("chatgpt").setup()
    end,
    dependencies = {
      "MunifTanjim/nui.nvim",
      "nvim-lua/plenary.nvim",
      "folke/trouble.nvim", -- optional
      "nvim-telescope/telescope.nvim"
    }
}

Configuration

ChatGPT.nvim comes with the following defaults, you can override them by passing config as setup param

https://github.com/jackMort/ChatGPT.nvim/blob/main/lua/chatgpt/config.lua

Example Configuration

A simple configuration of the chat model could look something like this:

{
  "jackMort/ChatGPT.nvim",
  event = "VeryLazy",
  config = function()
    require("chatgpt").setup({
      -- this config assumes you have OPENAI_API_KEY environment variable set
      openai_params = {
        -- NOTE: model can be a function returning the model name
        -- this is useful if you want to change the model on the fly
        -- using commands
        -- Example:
        -- model = function()
        --     if some_condition() then
        --         return "gpt-4-1106-preview"
        --     else
        --         return "gpt-3.5-turbo"
        --     end
        -- end,
        model = "gpt-4-1106-preview",
        frequency_penalty = 0,
        presence_penalty = 0,
        max_tokens = 4095,
        temperature = 0.2,
        top_p = 0.1,
        n = 1,
      }
    })
  end,
  dependencies = {
    "MunifTanjim/nui.nvim",
    "nvim-lua/plenary.nvim",
      "folke/trouble.nvim", -- optional
    "nvim-telescope/telescope.nvim"
  }
}

Secrets Management

Providing the OpenAI API key via an environment variable is dangerous, as it leaves the API key easily readable by any process that can access the environment variables of other processes. In addition, it encourages the user to store the credential in clear-text in a configuration file.

As an alternative to providing the API key via the OPENAI_API_KEY environment variable, the user is encouraged to use the api_key_cmd configuration option. The api_key_cmd configuration option takes a string, which is executed at startup, and whose output is used as the API key.

The following configuration would use 1Passwords CLI, op, to fetch the API key from the credential field of the OpenAI entry.

require("chatgpt").setup({
    api_key_cmd = "op read op://private/OpenAI/credential --no-newline"
})

The following configuration would use GPG to decrypt a local file containing the API key

local home = vim.fn.expand("$HOME")
require("chatgpt").setup({
    api_key_cmd = "gpg --decrypt " .. home .. "/secret.txt.gpg"
})

Note that the api_key_cmd arguments are split by whitespace. If you need whitespace inside an argument (for example to reference a path with spaces), you can wrap it in a separate script.

Usage

Plugin exposes following commands:

ChatGPT

ChatGPT command which opens interactive window using the gpt-3.5-turbo model. (also known as ChatGPT)

ChatGPTActAs

ChatGPTActAs command which opens a prompt selection from Awesome ChatGPT Prompts to be used with the gpt-3.5-turbo model.

preview image

ChatGPTEditWithInstructions

ChatGPTEditWithInstructions command which opens interactive window to edit selected text or whole window using the code-davinci-edit-001 model (GPT 3.5 fine-tuned for coding).

You can map it using the Lua API, e.g. using which-key.nvim:

local chatgpt = require("chatgpt")
wk.register({
    p = {
        name = "ChatGPT",
        e = {
            function()
                chatgpt.edit_with_instructions()
            end,
            "Edit with instructions",
        },
    },
}, {
    prefix = "<leader>",
    mode = "v",
})

preview image

ChatGPTRun

ChatGPTRun [action] command which runs specific actions -- See actions.json file for a detailed list. Available actions are:

  1. grammar_correction
  2. translate
  3. keywords
  4. docstring
  5. add_tests
  6. optimize_code
  7. summarize
  8. fix_bugs
  9. explain_code
  10. roxygen_edit
  11. code_readability_analysis -- see demo

All the above actions are using gpt-3.5-turbo model.

It is possible to define custom actions with a JSON file. See actions.json for an example. The path of custom actions can be set in the config (see actions_paths field in the config example above).

An example of custom action may look like this: (# marks comments)

{
  "action_name": {
    "type": "chat", # or "completion" or "edit"
    "opts": {
      "template": "A template using possible variable: {{filetype}} (neovim filetype), {{input}} (the selected text) an {{argument}} (provided on the command line), {{filepath}} (the relative path to the file)",
      "strategy": "replace", # or "display" or "append" or "edit"
      "params": { # parameters according to the official OpenAI API
        "model": "gpt-3.5-turbo", # or any other model supported by `"type"` in the OpenAI API, use the playground for reference
        "stop": [
          "```" # a string used to stop the model
        ]
      }
    },
    "args": {
      "argument": {
          "type": "strig",
          "optional": "true",
          "default": "some value"
      }
    }
  }
}

The edit strategy consists in showing the output side by side with the input and available for further editing requests. For now, edit strategy is implemented for chat type only.

The display strategy shows the output in a float window.

append and replace modify the text directly in the buffer.

Interactive popup

When using ChatGPT and ChatGPTEditWithInstructions, the following keybindings are available:

When the setting window is opened (with <C-o>), settings can be modified by pressing Enter on the related config. Settings are saved across sections

Whichkey plugin mappings

Add these to your whichkey plugin mappings for convenient binds

c = {
  name = "ChatGPT",
    c = { "<cmd>ChatGPT<CR>", "ChatGPT" },
    e = { "<cmd>ChatGPTEditWithInstruction<CR>", "Edit with instruction", mode = { "n", "v" } },
    g = { "<cmd>ChatGPTRun grammar_correction<CR>", "Grammar Correction", mode = { "n", "v" } },
    t = { "<cmd>ChatGPTRun translate<CR>", "Translate", mode = { "n", "v" } },
    k = { "<cmd>ChatGPTRun keywords<CR>", "Keywords", mode = { "n", "v" } },
    d = { "<cmd>ChatGPTRun docstring<CR>", "Docstring", mode = { "n", "v" } },
    a = { "<cmd>ChatGPTRun add_tests<CR>", "Add Tests", mode = { "n", "v" } },
    o = { "<cmd>ChatGPTRun optimize_code<CR>", "Optimize Code", mode = { "n", "v" } },
    s = { "<cmd>ChatGPTRun summarize<CR>", "Summarize", mode = { "n", "v" } },
    f = { "<cmd>ChatGPTRun fix_bugs<CR>", "Fix Bugs", mode = { "n", "v" } },
    x = { "<cmd>ChatGPTRun explain_code<CR>", "Explain Code", mode = { "n", "v" } },
    r = { "<cmd>ChatGPTRun roxygen_edit<CR>", "Roxygen Edit", mode = { "n", "v" } },
    l = { "<cmd>ChatGPTRun code_readability_analysis<CR>", "Code Readability Analysis", mode = { "n", "v" } },
  },

"Buy Me A Coffee"