Alpha4

Alpha4 is an AI-powered programming assistant for Neovim. It enhances your coding experience with intelligent suggestions, explanations, and interactive assistance.

Features

• Code analysis and improvement suggestions
• Interactive problem-solving
• Code generation and completion
• Dedicated chat interface for extended discussions

alpha4-demo.mp4

Installation

Using LazyVim

1. Ensure you have Neovim (0.8 or later) and LazyVim installed.

2. Add the following to your LazyVim configuration file (usually ~/.config/nvim/lua/plugins/alpha4.lua):

   return {
     {
       "rwtnb/alpha4.nvim",
       dependencies = {
         "nvim-lua/plenary.nvim",
       },
     },
   }

3. Restart Neovim or run :Lazy sync to install Alpha4.

4. If you're using Ollama (the default provider), make sure it's installed and running. Visit Ollama's website for installation instructions.

Usage

Alpha4 provides several commands for easy access:

1. Open the chat interface:

   • Press <leader>lc in normal mode
   • Or run the command :Alpha4ToggleChat

2. Explain selected code:

   • Select code in visual mode, then press <leader>le
   • Or run the command :Alpha4Explain

3. Replace selected code with improvements:

   • Select code in visual mode, then press <leader>lr
   • Or run the command :Alpha4Replace

4. Get code suggestions:

   • Select code in visual mode, then press <leader>ls
   • Or run the command :Alpha4Suggest

Chat Interface

1. When the chat interface opens, type your message at the bottom of the window.
2. Press <leader><CR> to send your message (default key binding).
3. Alpha4 will respond in the chat window.
4. Continue the conversation as needed.
5. To close the chat, press <leader>lc again or use your normal Neovim window closing command (e.g., :q).

Configuration

You can customize Alpha4 by passing options to the setup function. Here's a detailed explanation of each configuration option:

require('alpha4').setup({
  -- The default AI provider to use
  default_provider = "ollama",

  -- Configuration for different AI providers
  providers = {
    ollama = {
      url = "http://localhost:11434/api/chat",
      model = "llama3.1",
    },
    groq = {
      url = "https://api.groq.com/openai/v1/chat/completions",
      model = "llama-3.1-70b-versatile",
      api_key_name = "GROQ_API_KEY",
    },
    togetherai = {
      url = "https://api.together.xyz/v1/chat/completions",
      model = "meta-llama/Meta-Llama-3.1-405B-Instruct-Turbo",
      api_key_name = "TOGETHERAI_API_KEY",
    },
    openrouter = {
      url = "https://api.openrouter.io/v1/chat/completions",
      model = "openai/gpt-4o-2024-08-06",
      api_key_name = "OPENROUTER_API_KEY",
    },
    anthropic = {
      url = "https://api.anthropic.com/v1/messages",
      model = "claude-3-5-sonnet-20240620",
      api_key_name = "ANTHROPIC_API_KEY",
    },
    openai = {
      url = "https://api.openai.com/v1/chat/completions",
      model = "gpt-4o",
      api_key_name = "OPENAI_API_KEY",
    },
  },

  -- Keybindings for Alpha4 commands
  keys = {
    { "<leader>lc", "<CMD>Alpha4ToggleChat<CR>", desc = "Toggle chat" },
    { "<leader>le", "<CMD>Alpha4Explain<CR>", desc = "Explain" },
    { "<leader>lr", "<CMD>Alpha4Replace<CR>", desc = "Replace" },
    { "<leader>ls", "<CMD>Alpha4Suggest<CR>", desc = "Suggest" },
  },

  -- Configuration for the chat interface
  chat = {
    keys = {
      submit = "<leader><CR>",  -- Key to submit a message in chat
    },
  },

  -- Parameters for different AI tasks
  params = {
    chat = {
      temperature = 0.9,
      top_p = 0.95,
      top_k = 50,
      repetition_penalty = 1.2,
      frequency_penalty = 0.2,
      presence_penalty = 0.2,
      max_tokens = 4096,
    },
    replace = {
      temperature = 0.3,
      top_p = 0.8,
      top_k = 20,
      repetition_penalty = 1.05,
      frequency_penalty = 0.1,
      presence_penalty = 0.1,
      max_tokens = 4096,
    },
    suggest = {
      temperature = 0.2,
      top_p = 0.7,
      top_k = 10,
      repetition_penalty = 1.0,
      frequency_penalty = 0.05,
      presence_penalty = 0.05,
      max_tokens = 4096,
    },
    explain = {
      temperature = 0.5,
      top_p = 0.9,
      top_k = 35,
      repetition_penalty = 1.1,
      frequency_penalty = 0.2,
      presence_penalty = 0.2,
      max_tokens = 4096,
    },
  },

  -- Customizable prompts for different tasks
  -- @see: lua/alpha4/prompts.lua
  prompts = {
    explain = {
      system = "Your custom system prompt for explanations",
      user = "Your custom user prompt for explanations",
    },
    replace = {
      system = "Your custom system prompt for code replacement",
      user = "Your custom user prompt for code replacement",
    },
    suggest = {
      system = "Your custom system prompt for code suggestions",
      user = "Your custom user prompt for code suggestions",
    },
    chat = {
      system = "Your custom system prompt for chat interactions",
      user = "Your custom user prompt for chat interactions",
    },
  },
})

Supported Providers

Alpha4 supports multiple AI providers. To use a provider other than Ollama, you'll need to set up API keys:

1. Ollama (default, local)
2. Groq: Set the GROQ_API_KEY environment variable
3. TogetherAI: Set the TOGETHERAI_API_KEY environment variable
4. OpenRouter: Set the OPENROUTER_API_KEY environment variable
5. Anthropic: Set the ANTHROPIC_API_KEY environment variable
6. OpenAI: Set the OPENAI_API_KEY environment variable

To change providers, update the default_provider option in the setup function.

Troubleshooting

1. If Alpha4 doesn't respond:

   • Ensure your chosen AI provider is properly configured
   • Check that you have an active internet connection (except for Ollama)
   • Verify that your API key is correct (for non-Ollama providers)

2. If keybindings don't work:

   • Make sure Alpha4 is properly installed and loaded
   • Check for conflicting keybindings in your Neovim configuration

3. For Ollama-specific issues:

   • Ensure Ollama is running (ollama serve in your terminal)
   • Verify that the Ollama URL in the configuration matches your Ollama setup

For more help, please open an issue on the GitHub repository.

Roadmap

• LSP diagnostic messages as context for better suggestions
• Local RAG (Retrieval-Augmented Generation)
• Advanced Prompting Techniques (ReAct, Tree of Thoughts, etc.)