Kamosh

Your optimal AI-powered coding assistant for VS Code.

Kamosh brings next-generation AI features directly into your workflow, helping you write, refactor, analyze, and document code faster and smarter.

🚀 Features

• Native VS Code Chat: Use @kamosh directly in VS Code Chat with slash commands such as /explain, /fix, /tests, /doc, /refactor, /workspace, /new, and /clear.
• Always-On Context: Kamosh automatically collects relevant selection, active file, diagnostics, open/recent workspace context, explicit references, and small local memory. There is no context toggle.
• Local Memory: User preferences, workspace memory, and recent edit/completion signals are stored locally in VS Code storage and sent only as context.memory.
• Smart References: Attach precise context with #selection, #current, #workspace, #codebase, #errors, #file:path, and #terminal.
• Visible Context References: Native chat responses show file and range references used for the answer.
• Inline Helper: Get instant code suggestions right inside your editor with configurable debounce and automatic project context.
• Inline Chat Editing: Select code, describe the change, preview a diff, and apply edits without automatic file saves.
• Smart Apply Flow: API responses can include structured edits[], and markdown code blocks get Preview/Apply-style actions through native chat.
• Terminal Suggestions: Shell commands are inserted into the terminal first; execution stays under explicit user control.
• Open WebUI Routing: All AI flows go directly through an OpenAI-compatible Open WebUI gateway at https://labs.skerik.me.
• Native Model Picker Integration: Open WebUI models are exposed through VS Code's Language Model Provider API with tool-calling capability, so they can appear in the chat model picker as Kamosh Open WebUI models.
• Dynamic Model Selection: Load DeepSeek, OpenAI, and local Ollama-backed models from Open WebUI with Kamosh: Vybrat model.
• Legacy Sidebar View: The custom sidebar chat remains available as a fallback panel.
• Code Actions: Explain or fix diagnostics directly from VS Code quick actions.
• Intelligent Code Analysis: Contextual suggestions for improvements, optimizations, and best practices.
• Smart Suggestions: Context-aware tips and recommendations based on your project and file type.
• Advanced Developer Tools:
  • Smart Refactor: Multiple refactoring options with AI-powered suggestions, including performance, readability, design patterns, and more.
  • Test Generation: Automatic unit test creation for selected code.
  • Code Analysis: Comprehensive code quality analysis.

✨ UI/UX Improvements

• Native Chat First: The primary interface follows VS Code Chat patterns instead of a custom-only sidebar.
• Slash Commands: Common coding workflows are started with /fix, /tests, /doc, /refactor, and related commands.
• Enhanced Legacy Chat: The fallback webview keeps quick prompts, conversation history, model selection, and edit actions.
• Conversation Tools: Start a new chat or compact long conversations into a reusable summary.
• Loading Indicators: Clear feedback during AI processing.
• Responsive Design: Optimized for all screen sizes.
• Diff Highlighting: Color-coded highlights for added/removed lines in code previews.

🛠️ Developer Experience

• Context Menu Integration: Right-click for AI-powered code assistance, including Edit, Generate Tests, Smart Refactor, and Analyze Code.
• Command Palette: Quick access to all AI features:
  • Kamosh: Otevřít chat
  • Kamosh: Inline chat
  • Kamosh: Opravit
  • Kamosh: Vysvětlit
  • Kamosh: Vygenerovat testy
  • Kamosh: Vygenerovat dokumentaci
  • Kamosh: Chytrý refaktoring
  • Kamosh: Analyzovat kód
  • Kamosh: Vybrat model
  • Kamosh: Zadat Open WebUI API klíč
  • Kamosh: Přepnout inline návrhy
• Keyboard Shortcuts: Seamless workflow integration.
• Status Bar: Real-time status of AI features.

Open WebUI Gateway

Kamosh uses Open WebUI as a model gateway only. The extension keeps local context, local memory, chat history, edit parsing, and apply flows inside VS Code, then sends OpenAI-compatible chat completions to POST /api/chat/completions. Available models are discovered from GET /api/models. Open WebUI history, file upload, and RAG storage are not used by this iteration.

📦 Installation

Requires VS Code 1.120.0 or newer for native Language Model Provider integration.

1. Open VS Code Extensions (Ctrl+Shift+X).
2. Search for Kamosh.
3. Click Install.

🏁 Getting Started

• Open VS Code Chat and type @kamosh /explain #selection, @kamosh /fix #errors, or @kamosh /tests #selection.
• Use #current, #workspace, #codebase, #errors, or #file:path when you want to steer context explicitly.
• Run Kamosh: Vybrat model to choose any model exposed by your Open WebUI instance.
• Use the context menu or command palette for editing, refactoring, test generation, and analysis.
• Open the Kamosh sidebar only when you need the legacy fallback panel.

🧑‍💻 License

MIT