IntelligenceX

Common Questions

Get quick answers about IntelligenceX setup, trust, supported providers, platform support, and GitHub App ownership.

Frequently Asked Questions

Is my code safe? Do you see it?

IntelligenceX has no backend service. Your code diffs go directly from GitHub Actions to your chosen AI provider (ChatGPT or Copilot). We never see, store, or process your code.

Why do I use my own ChatGPT or Copilot login?

This is the zero-trust design. You authenticate with your own account. No shared API keys, no pooled quotas, no third-party billing. You control the credentials, stored in GitHub Actions secrets that only your workflows can access.

What is 'Bring Your Own GitHub App'?

Instead of trusting our GitHub App, you can create your own App under your organization. This gives you branded bot identity, fine-grained permissions, and full audit control. The CLI wizard supports this flow natively.

Which AI providers are supported?

ChatGPT (native transport via OAuth) and GitHub Copilot (via Copilot CLI). You configure the provider and model in .intelligencex/reviewer.json.

How do I get started?

Run 'intelligencex setup autodetect' first, then use 'intelligencex setup wizard' or 'intelligencex setup web'. Onboarding is path-first (new setup, fix auth, cleanup, maintenance) and runs entirely on your machine.

Can IntelligenceX detect what path I should take?

Yes. Auto-detect runs doctor-style checks (auth/config/GitHub access) and suggests a path before repo selection. It helps you choose between first-time setup, auth refresh, or maintenance.

How do cleanup and maintenance differ?

Cleanup removes workflow/config (and optionally secrets). Maintenance inspects current state first, then guides you to setup, update-secret, or cleanup depending on findings.

What platforms are supported?

Windows, macOS, and Linux. The CLI and library target .NET 8+ and .NET 10+, the PowerShell module supports netstandard2.0, net472, and net8.0. The reviewer runs on any GitHub Actions runner.

Can I customize the review behavior?

Yes. Choose from presets (balanced, picky, security, performance, tests, minimal) or write custom JSON config. Control review mode (inline, summary, hybrid), length, output style, and auto-resolve behavior.

Is this similar to Claude Code or Codex?

The reviewer produces structured, Codex-style output integrated into your GitHub PR workflow. It supports inline comments with code suggestions, summary reviews, and hybrid mode.