Desktop App Overview

The GuideAI desktop application is a privacy-first, cross-platform tool designed to monitor and analyze your AI coding assistant sessions locally on your machine.

Core Philosophy

Privacy First

GuideAI Desktop is built with privacy at its core:

• Local-Only by Default: All session data is stored locally on your machine
• Bring Your Own Key (BYOK): Use your own AI API keys for analysis
• You Control What’s Shared: Choose exactly what data (if any) gets synced to the cloud
• No Tracking: GuideAI never tracks your usage without your explicit consent

Multi-Provider Support

GuideAI Desktop works with multiple AI coding assistants:

• Claude Code - Anthropic’s AI coding assistant
• GitHub Copilot - GitHub’s AI pair programmer
• OpenCode - Open source coding assistant
• OpenAI Codex - OpenAI’s code generation model
• Gemini - Google gemini coding assistant

Each provider can be enabled independently and configured to monitor specific projects.

How It Works

1. File Watching

When you enable a provider, GuideAI Desktop monitors specific directories on your system where the AI assistant stores session data:

• Claude Code: ~/.claude/projects/
• GitHub Copilot: ~/.copilot/
• OpenCode: ~/.local/share/opencode/storage/ (Linux/macOS), %LOCALAPPDATA%/opencode (Windows)
• Codex: ~/.codex/
• Gemini: ~/.gemini/

2. Local Storage

All session data is stored in a local SQLite database:

• Location: Platform-specific application data directory
• Structure: Normalized schema for efficient queries
• Privacy: Never transmitted without your permission

3. Session Processing

Sessions are processed locally to extract metrics:

• Performance: Completion time, token count, memory usage
• Usage: Commands executed, files read, tools used
• Quality: Code lines generated, test coverage, complexity
• Engagement: User interactions, feedback, retries