Indusagi VS Code Extension

A Visual Studio Code extension that uses Indusagi as an AI coding assistant via RPC mode.

Features

• Ask Indusagi - Select code and get AI assistance
• Clear Context - Reset conversation context
• Real-time streaming - See responses as they're generated
• Configuration - Customize Indusagi path, model, and API key
• Output Channel - View all Indusagi communications and debug

Installation

From Source

# Clone the repository
git clone https://github.com/yourusername/indusagi-vscode.git

# Navigate to the extension directory
cd indusagi-vscode

# Install dependencies
npm install

From VS Code Marketplace (Coming Soon)

# Package the extension
npm run package

# Install locally for testing
code --install-extension indusagi-vscode-*.vsix

Configuration

Extension Settings

Open VS Code settings (Cmd+,) and search for "Indusagi Settings":

Setting API Key

You have two options:

Option 1: Indusagi Login (Recommended)

# Open Indusagi auth (opens browser for OAuth)
indusagi login

# Then use the extension

Option 2: Manual API Key

# Set your OpenAI API key in VS Code settings
1. Open Settings (Cmd+,)
2. Search for "Indusagi Settings"
3. Enter your API key in "API Key" field

Usage

Basic Chat

1. Open a file in VS Code
2. Select the code you want to ask about
3. Press Cmd+Shift+A or use the Indusagi panel → "Ask Indusagi"
4. View the response in the Indusagi output channel

Clear Context

Press Cmd+Shift+C or use Indusagi panel → "Clear Context" to reset the conversation.

How It Works

The extension communicates with Indusagi via JSON-RPC protocol over stdin/stdout:

1. Starts Indusagi in RPC mode: indusagi --mode rpc --no-session
2. Sends commands as JSON objects (one per line)
3. Streams responses in real-time
4. Parses message_update events to display streaming text

RPC Protocol

Request format:

{
  "type": "prompt",
  "id": "req-1",
  "message": "Your code here"
}

Event stream:

{"type":"message_update","message":{...},"assistantMessageEvent":{"type":"text_delta","delta":"Hello"}}
{"type":"message_end","message":{...}}

Architecture

┌─────────────────────────────────┐
│     VS Code Extension      │
├────────────────────────────────┤
│  RPC Client (extension.ts)  │
├────────────────────────────────┤
│  Indusagi (RPC Server)      │
└─────────────────────────────────┘

The extension spawns Indusagi as a subprocess and communicates via JSON over stdin/stdout, following the RPC protocol documented at indusagi.com/cli/rpc.

Development

Building

# Compile TypeScript
npm run compile

# Create VSIX package
npm run package

Running in Debug Mode

# Launch VS Code with your extension loaded
code --extensionDevelopmentPath=./out

Project Structure

indusagi-vscode/
├── src/
│   └── extension.ts       # Main extension logic
├── out/                    # Compiled output
│   └── extension.js        # Bundled extension
├── package.json              # Extension manifest
├── tsconfig.json             # TypeScript config
├── vsc-extension-quickstart.md  # VS Code debugging guide
└── README.md                # This file

Requirements

• Node.js 20+
• Indusagi CLI installed and available in PATH
• (Optional) OpenAI API key (can be set via extension or indusagi login)

Troubleshooting

Indusagi not found

If you see an error "Indusagi not found", ensure Indusagi is installed and available in your system PATH.

To verify Indusagi is installed:

which indusagi

If not installed:

• Install Indusagi: npm install -g indusagi-coding-agent
• Restart VS Code

Connection issues

• Check the Indusagi Output Channel for error messages
• Ensure Indusagi RPC mode is working: Run indusagi --mode rpc manually in terminal
• Check that your API key is valid (if using manual key)

Contributing

Contributions welcome! Please open issues at https://github.com/yourusername/indusagi-vscode/issues.

License

MIT

Built with ❤️ using Indusagi SDK

Test change