Natural language first based development
navi.open() will open a prompt. Use this prompt to tell naVi what to do. For example, insert a React component outline.
navi.openRange() will open a prompt using the current selection as context. Use this prompt to get naVi to make
changes to the selected text. For example, remove a bug.
navi.openFile() will open a prompt using the current file as context. Use this prompt to get naVi to make
changes to the current file. For example, add a new function.
navi.requestReview() will request a review using the current selection as context. The final report will open in a separate window,
and will not alter the selected text.
navi.explainRange() will provide an explanation of code using the current selection as context. The explanation will
open in a separate window and will not alter the selected text.
navi.openChat() will open a chat interface where you can discuss your code with GPT.
Add the following to your Packer config
use({
'deifyed/naVi',
requires = {'jcdickinson/http.nvim', run = 'cargo build --workspace --release'},
})Add the following to your Lazy.vim config
{
'deifyed/naVi',
dependencies = {
{
"jcdickinson/http.nvim", build = "cargo build --workspace --release",
},
},
config = function()
require("navi").setup({ })
end,
keys = {
{ "<C-PageDown>", "<cmd>lua require('navi').open()<cr>", mode = "i", desc = "NaVI prompt" },
{ "<C-PageDown>", "<cmd>lua require('navi').openRange()<cr>", mode = "v", desc = "NaVI prompt with context" },
{ "<C-PageUp>", "<cmd>lua require('navi').requestReview()<cr>", mode = "v", desc = "NaVI request review" },
},
}
in ~/.config/nvim/after/plugin/navi.lua:
local navi = require('navi')
navi.setup({
-- OpenAI token. Required
openai_token = "<token>", -- Alternatively, use environment variable OPENAI_TOKEN=<token>
-- OpenAI model. Optional. Default is gpt-3.5-turbo
openai_model = "gpt-3.5-turbo",
-- OpenAI max tokens. Optional. Default is 512
openai_max_tokens = 512,
-- OpenAI temperature. Optional. Default is 0.6
openai_temperature = 0.6,
-- Debug mode. Optional. Default is false
debug = false, -- Alternatively, use environment variable NAVI_DEBUG=true
-- Setup for input window
prompt_window = {
border = "single",
style = "minimal",
relative = "editor",
},
-- Setup for window showing various reports
report_window = {
-- Specifies if the report will be shown in a vertical window or in a floating window.
window = "floating",
border = "single",
style = "minimal",
relative = "editor",
},
})
-- Set keybindings
vim.api.nvim_set_keymap('v', '<C-PageDown>', '', { callback = navi.openRange })
vim.api.nvim_set_keymap('i', '<C-PageDown>', '', { callback = navi.open })
vim.api.nvim_set_keymap('v', '<C-PageUp>', '', { callback = navi.requestReview })
vim.api.nvim_set_keymap('v', '<C-End>', '', { callback = navi.explainRange })
vim.api.nvim_set_keymap('n', '<C-End>', '', { callback = navi.openChat })- Make the current file context of the chat interface
- Make the chat toggleable
- Improve prompts
- Refactor
- Where can I get an OpenAI token?
- Can naVi close nVim for me?
AI has come far, but not that far. We'll have to wait for human alignment before attempting this.
- Neat! How can I contribute?
Great! Check out the contributing guide for more information.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent