Generate text using LLMs with customizable prompts

• Ollama
• Curl

Use command Ollama to generate text based on predefined and customizable prompts. If Ollama is not running, it is automatically started.

Example key maps:

vim.keymap.set('v', '<leader>]', ':Ollama<CR>')
vim.keymap.set('n', '<leader>]', ':Ollama<CR>')

You can also directly invoke it with a predefined or a custom prompt:

vim.keymap.set('v', '<leader>]', ':Ollama Enhance_Grammar_Spelling<CR>')

Use command OllamaModel to show locally available models and change the default model for this session.

{
    "sgelb/ollama.nvim"
	cmd = { "Ollama", "OllamaModel" },
	dependencies = {
		{ "nvim-lua/plenary.nvim" },
	opts = {
		default_model = "codellama:7b",
        -- custom prompts
		prompts = {
			UnitTest = {
				prompt = "Write unit tests for the following code. Output the result in the format ```$filetype\n...\n```:\n```$filetype\n$text\n```",
				replace = false,
				extract = "```$filetype\n(.-)```",
			},
		},
	},
}

Predefined prompts are available threquire('ollama').prompts, you can enhance or modify them.

Example:

require('ollama').prompts['Elaborate_Text'] = {
  prompt = "Elaborate the following text:\n$text",
  replace = true
}

You can use the following properties per prompt:

• prompt: Prompt which can use the following placeholders:
  • $text: Visually selected text
  • $filetype: Filetype of the buffer (e.g. javascript)
  • $input: Additional user input
• replace: true if the selected text shall be replaced with the generated output
• extract: Regular expression used to extract the generated result

You can change the model with

require('ollama').model = 'your_model' -- default 'mistal:instruct'

Here are all available models.

You can also change the complete command with

require('ollama').command = 'your command' -- default 'ollama run $model $prompt'

You can use the placeholders $model and $prompt.