ThePrimeagen/99

99

The AI client that Neovim deserves, built by those that still enjoy to code.

IF YOU ARE HERE FROM THE YT VIDEO

So many things have changed. So please be careful!

WARNING :: API CHANGES RIGHT NOW

It will happen that apis will disapear or be changed. Sorry, this is an BETA product.

Project Direction

This repo is meant to be my exploration grounds for using AI mixed with tradcoding.

I believe that hand coding is still very important and the best products i know of today still do that (see opencode vs claude code)

Warning

1. Prompts are temporary right now. they could be massively improved
2. Officially in beta, but api can still change. unlikely at this point

99

The AI Neovim experience

_99

99 is an agentic workflow that is meant to meld the current programmers ability with the amazing powers of LLMs. Instead of being a replacement, its meant to augment the programmer.

As of now, the direction of 99 is to progress into agentic programming and surfacing of information. In the beginning and the original youtube video was about replacing specific pieces of code. The more i use 99 the more i realize the better use is through search and work

Basic Setup

	{
		"ThePrimeagen/99",
		config = function()
			local _99 = require("99")

            -- For logging that is to a file if you wish to trace through requests
            -- for reporting bugs, i would not rely on this, but instead the provided
            -- logging mechanisms within 99.  This is for more debugging purposes
            local cwd = vim.uv.cwd()
            local basename = vim.fs.basename(cwd)
			_99.setup({
                -- provider = _99.Providers.ClaudeCodeProvider,  -- default: OpenCodeProvider
				logger = {
					level = _99.DEBUG,
					path = "/tmp/" .. basename .. ".99.debug",
					print_on_error = true,
				},
                -- When setting this to something that is not inside the CWD tools
                -- such as claude code or opencode will have permission issues
                -- and generation will fail refer to tool documentation to resolve
                -- https://opencode.ai/docs/permissions/#external-directories
                -- https://code.claude.com/docs/en/permissions#read-and-edit
                tmp_dir = "./tmp",

                --- Completions: #rules and @files in the prompt buffer
                completion = {
                    -- I am going to disable these until i understand the
                    -- problem better.  Inside of cursor rules there is also
                    -- application rules, which means i need to apply these
                    -- differently
                    -- cursor_rules = "<custom path to cursor rules>"

                    --- A list of folders where you have your own SKILL.md
                    --- Expected format:
                    --- /path/to/dir/<skill_name>/SKILL.md
                    ---
                    --- Example:
                    --- Input Path:
                    --- "scratch/custom_rules/"
                    ---
                    --- Output Rules:
                    --- {path = "scratch/custom_rules/vim/SKILL.md", name = "vim"},
                    --- ... the other rules in that dir ...
                    ---
                    custom_rules = {
                      "scratch/custom_rules/",
                    },

                    --- Configure @file completion (all fields optional, sensible defaults)
                    files = {
                        -- enabled = true,
                        -- max_file_size = 102400,     -- bytes, skip files larger than this
                        -- max_files = 5000,            -- cap on total discovered files
                        -- exclude = { ".env", ".env.*", "node_modules", ".git", ... },
                    },
                    --- File Discovery:
                    --- - In git repos: Uses `git ls-files` which automatically respects .gitignore
                    --- - Non-git repos: Falls back to filesystem scanning with manual excludes
                    --- - Both methods apply the configured `exclude` list on top of gitignore

                    --- What autocomplete engine to use. Defaults to native (built-in) if not specified.
                    source = "native", -- "native" (default), "cmp", or "blink"
                },

                --- WARNING: if you change cwd then this is likely broken
                --- ill likely fix this in a later change
                ---
                --- md_files is a list of files to look for and auto add based on the location
                --- of the originating request.  That means if you are at /foo/bar/baz.lua
                --- the system will automagically look for:
                --- /foo/bar/AGENT.md
                --- /foo/AGENT.md
                --- assuming that /foo is project root (based on cwd)
				md_files = {
					"AGENT.md",
				},
			})

            -- take extra note that i have visual selection only in v mode
            -- technically whatever your last visual selection is, will be used
            -- so i have this set to visual mode so i dont screw up and use an
            -- old visual selection
            --
            -- likely ill add a mode check and assert on required visual mode
            -- so just prepare for it now
			vim.keymap.set("v", "<leader>9v", function()
				_99.visual()
			end)

            --- if you have a request you dont want to make any changes, just cancel it
			vim.keymap.set("n", "<leader>9x", function()
				_99.stop_all_requests()
			end)

			vim.keymap.set("n", "<leader>9s", function()
				_99.search()
			end)
		end,
	},

Usage

I would highly recommend trying out search as its the direction the library is going

_99.search()

See search for more details

Description

API

setup

Sets up _99. Must be called for this library to work. This is how we setup in flight request spinners, set default values, get completion to work the way you want it to.

search

Performs a search across your project with the prompt you provide and return out a list of locations with notes that will be put into your quick fix list.

vibe

No description.

open

Opens a selection window for you to select the last interaction to open and display its contents in a way that makes sense for its type. For search and vibe, it will open the qfix window. For tutorial, it will open the tutorial window.

visual

takes your current selection and sends that along with the prompt provided and replaces your visual selection with the results

view_logs

views the most recent logs and setups the machine to view older and new logs this is still pretty rough and will change in the near future

stop_all_requests

stops all in flight requests. this means that the underlying process will be killed (OpenCode) and any result will be discared

clear_previous_requests

clears all previous search and visual operations

Extensions

check out Worker for cool abstraction on search and vibe

_99.Extensions.Worker

A persistent way to keep track of work.

this will likely be where the most change and focus goes into. I would like to take this into worktree territory and be able to swap between stuff super slick.

Until then, it is going to be a single bit of work that you can provide the description and then use search to find what is left that needs to be done.

Description

API

set_work

will set the work for the project. If opts provide a description then no input capture of work description will be required

search

will use _99.search to find what is left to be done for this work item to be considered done

_99.Options

No description.

Description

API

logger

No description.

model

No description.

in_flight_options

No description.

md_files

No description.

provider

No description.

display_errors

No description.

auto_add_skills

No description.

completion

No description.

tmp_dir

No description.

_99.ops.Opts

The options that are used throughout all the interations with 99. This includes search, visual, and others

Description

API

additional_prompt

by providing additional_prompt you will not be required to provide a prompt. this allows you to define actions based on remaps

remap("n", "<leader>9d", function()
  --- this function could be used to auto debug your project
  _99.search({
    additional_prompt = [[
run `make test` and debug the test failures and provide me a comprehensive set of steps where
the tests are breaking ]]
  })
end)

This would kick off a search job that will run your tests in the background. the resulting failures would be diagnosed and search results would be transfered into a quick fix list.

additional_rules

can be used to provide extra args. If you have a skill called "cloudflare" you could provide the rule for cloudflare and its context will be injected into your request

_99.ops.SearchOpts

See _99.opts.Opts for more information.

There are no properties yet. But i would like to tweek some behavior based on opts

Description

API

No properties.

_99.WorkOpts

No description.

Description

API

description

No description.

_99.Completion

No description.

Description

API

source

No description.

custom_rules

No description.

files

No description.

_99.InFlight.Opts

this is pure a class for testing. helps controls timings

Description

API

throbber_opts

options for the throbber in the top left

in_flight_interval

frequency in which the in-flight interval checks to see if it should be displayed / removed

enable

defaults to true

_99.Logger.Options

No description.

Description

API

level

No description.

type

No description.

path

No description.

print_on_error

No description.

max_requests_cached

No description.

_99.Agents.Rule

No description.

Description

API

name

No description.

path

No description.

absolute_path

No description.

Completions

When prompting, you can reference rules and files to add context to your request.

• # references rules — type # in the prompt to autocomplete rule files from your configured rule directories
• @ references files — type @ to fuzzy-search project files. This will exclude files that are in .gitignore.

Referenced content is automatically resolved and injected into the AI context. Native completions work by default. For nvim-cmp or blink.cmp, set source = "cmp" or source = "blink".

Providers

99 supports multiple AI CLI backends. Set provider in your setup to switch. If you don't set model, the provider's default is used.

_99.setup({
    provider = _99.Providers.ClaudeCodeProvider,
    -- model is optional, overrides the provider's default
    model = "claude-sonnet-4-5",
})

Extensions

Telescope Model Selector

If you have telescope.nvim installed, you can switch models on the fly via the Telescope picker:

vim.keymap.set("n", "<leader>9m", function()
  require("99.extensions.telescope").select_model()
end)

The selected model is used for all subsequent requests in the current session.

Telescope Provider Selector

Switch between providers (OpenCode, Claude, Cursor, Kiro) without restarting Neovim. Switching provider also resets the model to that provider's default.

vim.keymap.set("n", "<leader>9p", function()
  require("99.extensions.telescope").select_provider()
end)

fzf-lua

If you use fzf-lua instead of telescope, the same pickers are available:

vim.keymap.set("n", "<leader>9m", function()
  require("99.extensions.fzf_lua").select_model()
end)

vim.keymap.set("n", "<leader>9p", function()
  require("99.extensions.fzf_lua").select_provider()
end)

Reporting a bug

To report a bug, please provide the full running debug logs. This may require a bit of back and forth.

Please do not request features. We will hold a public discussion on Twitch about features, which will be a much better jumping point then a bunch of requests that i have to close down. If you do make a feature request ill just shut it down instantly.

The logs

To get the last run's logs execute :lua require("99").view_logs().

Dont forget

If there are secrets or other information in the logs you want to be removed make sure that you delete the query printing. This will likely contain information you may not want to share.