LLMV v0.2.0: Provider System & Configuration

🎉 New Features

Provider System

• Added a new provider architecture to support multiple LLM backends
• Anthropic (Claude) is now configured as the default provider
• Providers can be configured with model-specific settings

Configuration Options

• Added comprehensive configuration options:

  require('llmv').setup({
      provider = "anthropic", -- Default provider
      providers = {
          anthropic = {
              model = "claude-3-5-sonnet-20241022", -- Default model
              max_tokens = 8192,
              temperature = 0.7,
          }
      }
  })

Better Documentation

• Updated documentation with examples for all major plugin managers
• Added configuration examples for different setups
• Improved explanations of command usage and security considerations

🔧 Improvements

Core Refactoring

• Completely refactored the core module to support the provider system
• Improved error handling and response processing
• Better management of running LLM requests

UI Enhancements

• Better status messages during LLM requests
• Improved error reporting when requests fail
• More consistent handling of streaming responses

🧹 Housekeeping

Repository Changes

• Added prompts/ directory to .gitignore for local testing
• Improved code organization with dedicated provider modules
• Cleaner separation of concerns between UI, core logic, and providers

🗺️ Roadmap

• Support for additional LLM providers (OpenAI, Ollama, etc.)
• Custom prompt templates
• Response formatting options

🚀 Getting Started

To use the new provider system, update your configuration:

require('llmv').setup({
    provider = "anthropic",
    providers = {
        anthropic = {
            model = "claude-3-5-sonnet-20241022",
            max_tokens = 8192,
            temperature = 0.7,
        }
    }
})

Don't forget to set your API key:

export ANTHROPIC_API_KEY='your-api-key-here'

v0.1.0-beta

LLMV v0.1.0-beta - Initial Release

LLMV brings AI assistance into Neovim using simple, composable tools you already know. No special buffers, no complex UI - just Markdown files and shell commands.

Features

• Simple Markdown-based interface using >>> for prompts and <<< for responses
• Shell command integration via @bash() syntax
• Real-time streaming responses from Claude AI
• Basic commands: :Run and :Stop
• File-relative command execution
• Secure command evaluation (only most recent prompt)

Requirements

• Neovim
• Anthropic API key (Claude AI)

Installation

-- Using lazy.nvim
{
    'David-Factor/llmv',
    cmd = { "Run", "Stop" },
    config = true,
}

Known Limitations

• Currently only supports Claude AI
• Basic configuration options