Aii - AI-Powered Assistant for VSCode

Version 0.3.0 - Agentic AI assistant that reads, writes, and searches your codebase. Works with Anthropic Claude, OpenAI GPT, Google Gemini, DeepSeek, and Moonshot.

Agent Mode • Multi-Provider • Inline Diff Review

✨ What's New in v0.3.0

• ⏹️ Stop Button - Cancel agent execution anytime with a single click
• ▶️ Continue Button - Resume paused agent tasks to keep working on complex problems
• 🤖 DeepSeek Chat Support - Agent mode now works with DeepSeek Chat models (in addition to Claude and OpenAI GPT)

Features

💬 Interactive Chat with Agent Mode

Chat with Aii directly in VSCode, now with autonomous tool execution.

• Keyboard Shortcut: Cmd+Shift+A (Mac) / Ctrl+Shift+A (Windows/Linux)
• Command: Aii: Open Chat
• Features:
  • Chat Mode: Direct conversation with AI
  • Agent Mode: AI can read/write files, search code, and list directories
  • Real-time streaming responses
  • Beautiful Markdown rendering with syntax highlighting
  • Conversation history with context
  • Interactive inline diff review for file changes

🔧 Code Generation

Generate code from natural language descriptions with streaming responses.

• Keyboard Shortcut: Cmd+Shift+G (Mac) / Ctrl+Shift+G (Windows/Linux)
• Command: Aii: Generate Code

📝 Git Commit Messages

AI-powered commit message generation from staged changes.

• Keyboard Shortcut: Cmd+Shift+C (Mac) / Ctrl+Shift+C (Windows/Linux)
• Command: Aii: Generate Commit Message

💡 Code Explanation

Understand complex code in plain English.

• Command: Aii: Explain Code
• Select code and run the command

🌍 Translation

Translate text/comments between languages.

• Command: Aii: Translate Text
• Select text and specify target language

⚙️ Settings & Model Selection

• Command: Aii: Open Settings - Configure API keys and providers
• Command: Aii: Select AI Model - Switch models instantly
• Command: Aii: Test API Key - Verify your API keys work

📋 Requirements

Aii Server (Docker) - Required for v0.3.x

Install with one command:

curl -fsSL https://aiiware.com/install.sh | bash

The server handles all AI requests and supports multiple LLM providers.

🚀 Quick Start

1. Install Aii Server:

   curl -fsSL https://aiiware.com/install.sh | bash
   

2. Install this extension from VSCode Marketplace

3. Configure API Keys:

   • Open Command Palette: Cmd+Shift+P (Mac) / Ctrl+Shift+P (Windows/Linux)
   • Run Aii: Open Settings
   • Go to "LLM Providers" tab
   • Add your API key(s) for at least one provider:
     • Anthropic: Get key from console.anthropic.com
     • OpenAI: Get key from platform.openai.com/api-keys
     • Google: Get key from makersuite.google.com/app/apikey
     • DeepSeek: Get key from platform.deepseek.com
     • Moonshot: Get key from platform.moonshot.ai
   • Click "Test Connection" to verify

4. Select a Model:

   • Click the model selector in chat panel header
   • Or run Aii: Select AI Model from Command Palette

5. Start Using:

   • Click the Aii icon in Activity Bar
   • Or press Cmd+Shift+A (Mac) / Ctrl+Shift+A (Windows/Linux)

Configuration

Settings

Access settings via Aii: Open Settings command or VSCode Settings UI:

• aii.serverUrl - Aii Server URL (default: http://localhost:26169)
• aii.streaming - Enable streaming mode for response display (default: true)
• aii.apiKeys.anthropic - Anthropic API key
• aii.apiKeys.openai - OpenAI API key
• aii.apiKeys.google - Google API key
• aii.apiKeys.deepseek - DeepSeek API key
• aii.apiKeys.moonshot - Moonshot API key

Usage Examples

Use Agent Mode

1. Open Chat (Cmd+Shift+A)
2. Toggle to "Agent" mode in the chat header
3. Ask AI to explore or modify your codebase:
   • "Find all TODO comments in this project"
   • "Update the API endpoint in all config files"
   • "Add error handling to database.ts"
4. Review and accept/reject file changes in inline diffs

Generate Code

1. Open a file in your preferred language
2. Press Cmd+Shift+G
3. Describe the code you want (e.g., "function to calculate fibonacci numbers")
4. Watch as code appears and inserts at cursor

Generate Commit Message

1. Stage your changes: git add .
2. Press Cmd+Shift+C
3. Review AI-generated commit message
4. Confirm to commit

Switch Models

1. Click the model name in chat panel header
2. Select a different model from the list
3. Or use status bar model indicator (left side)

Features

✅ Agent Mode - AI autonomously explores and modifies your codebase ✅ Inline Diff Review - Accept or reject changes with CodeLens ✅ Multi-Provider Support - 5 LLM providers (Anthropic, OpenAI, Google, DeepSeek, Moonshot) ✅ Real-time Streaming - Token-by-token display with <100ms latency ✅ Keyboard Shortcuts - 1-keystroke access to common operations ✅ Conversation Context - AI remembers chat history ✅ Modern Settings UI - Tabbed interface with API key testing ✅ Model Switching - Change models instantly without restarting

Troubleshooting

Server Not Connected

• Ensure Aii Server is installed: docker ps | grep aii-server
• If not installed, run: curl -fsSL https://aiiware.com/install.sh | bash
• Check server is running: curl http://localhost:26169/health
• The extension will show a connection banner in chat if server is not detected

Commands Not Working

• Check connection status in chat panel banner
• Verify at least one API key is configured in Settings
• Select a model using the model selector
• Check VSCode Output panel (View → Output → Aii Server)

API Key Issues

• Use Aii: Open Settings to configure keys
• Click "Test Connection" to verify each key
• Ensure keys are not expired or rate-limited
• Check provider console for quota/billing issues

Streaming Not Working

• Enable streaming in settings: aii.streaming: true
• Check network connection to Aii Server
• Ensure server is up to date (re-run install script)

📄 License & Credits

License: Apache 2.0 • Copyright 2025-present aiiware.com

💬 Support & Community

• 🐦 Follow us: @aii_dev - Release notes, tips, and updates
• 📖 Documentation: Extension docs and troubleshooting guides in this README
• 💬 Questions: VSCode Marketplace Q&A section
• 🐛 Issues: Report bugs through the marketplace or extension feedback