W.A.R.A.N

Workspace Autonomous Repository Analysis & Navigator

W.A.R.A.N is a VS Code extension that brings local AI-powered software development directly into your editor. It works with locally hosted language models and provides repository-aware assistance, intelligent code editing, project analysis, and development workflows while keeping your source code on your own machine.

Features

Local AI Integration

• Works with locally hosted language models
• Supports multiple inference runtimes
• No cloud dependency required
• No API keys required

Supported runtimes:

• Ollama
• llama.cpp
• LM Studio
• LocalAI

Repository Intelligence

W.A.R.A.N understands your project by indexing the workspace and analyzing:

• Project structure
• Source code
• Symbols
• Imports
• Dependencies
• Frameworks
• Package managers

This allows the assistant to work with relevant project context instead of processing the entire repository.

AI Coding Assistant

W.A.R.A.N can assist with development tasks such as:

• Explaining code
• Refactoring code
• Updating multiple files
• Creating new files
• Modifying existing files
• Finding related code
• Updating imports
• Navigating large projects

Build & Diagnostics

The extension can assist with common development workflows including:

• Running project builds
• Viewing compiler output
• Detecting common build issues
• Reviewing diagnostics
• Helping resolve compilation errors

Supported ecosystems include:

• Node.js
• React
• Next.js
• TypeScript
• Python
• Go
• Rust
• .NET

Model Manager

Manage local AI models directly from VS Code.

Features include:

• Runtime detection
• Installed model management
• Model download
• Model selection
• Runtime health information

Repository Index

The repository engine provides:

• Incremental indexing
• Dependency graph analysis
• Symbol extraction
• Import resolution
• Persistent caching

Diagnostics

Built-in diagnostics provide information about:

• Hardware
• Runtime status
• Repository index
• Storage usage
• Logs

Installation

Clone the repository.

git clone <repository-url>
cd waran

Install dependencies.

npm install

Build the extension.

npm run build

Launch the Extension Development Host.

Press F5 inside VS Code.

Supported Local Runtimes

These endpoints can be configured from the W.A.R.A.N extension settings.

Commands

Open the Command Palette and search for:

W.A.R.A.N

Available commands include:

• Open Agent
• Welcome & Setup
• Model Manager
• Settings
• Diagnostics
• Logs
• Rebuild Repository Index
• Switch Model
• Switch Runtime
• Run Agent Task
• Cancel Agent Task
• Detect Hardware
• Find Related Files

Development

Build

npm run build

Watch extension

npm run watch:extension

Watch webview

npm run watch:webview

Type checking

npm run typecheck

Lint

npm run lint

Package extension

npm run package

Privacy

W.A.R.A.N is designed to work with locally hosted AI models.

Repository contents remain on your local machine unless you explicitly configure external services.

Model downloads are performed only when requested by the user.

Requirements

One of the following local inference runtimes:

• Ollama
• LM Studio
• llama.cpp
• LocalAI

License

MIT License

Author

Developed and maintained by Dhurgesh.