There's no single right setup. Here's what Notre Dame developers actually use, what the university supports at each level, and how the wider ecosystem beyond our official list fits in.

Whatever you pick: ND-provisioned accounts for ND work, and mind your data's sensitivity level. Those two rules apply no matter which tool you choose.

The landscape

Setup What it is Availability
VS Code + GitHub Copilot or Codex The ND developer standard: VS Code as the editor, Copilot for inline completion and chat, Codex for agentic work. Backed by GitHub Advanced Security — static analysis, secret scanning, dependency review. Standard — ND-provisioned accounts
Gemini (Canvas mode) Coding right in the Gemini web app — prototype, iterate, and preview with nothing installed. A natural first stop if you're just getting started. Included with ND Google accounts
Google AI Studio Google's model playground and API workbench for building with Gemini models. Available on request
Claude Code harness + your model provider Install the Claude Code harness and point it at an approved model provider — an agentic, command-line workflow for developers who want one. Bring-your-own (approved providers and accounts)
Portkey (API gateway) OIT-offered gateway for direct model API access — routing, observability, and governance for apps that call models programmatically. Available with chargeback
Other agentic IDEs & harnesses Cursor, Windsurf, Cline, aider, and a fast-moving field of others. Many work well; few are officially offered. Case-by-case — tier assessment and data rules apply

How to read this table

Standard means the institution has invested support, governance, and a backstop: when something breaks, there's someone to call, and the contractual protections are in place. On request and chargeback options are supported paths with an extra step. Bring-your-own means the workflow is fine but you're assembling it yourself from approved pieces.

The last row is deliberate: we don't pretend the wider ecosystem doesn't exist, and we don't ban it. If an unlisted tool would make a real difference for you, talk to the AI Enablement team before installing it. Going off the supported path means taking on the support and governance work yourself.

This page changes faster than any other in the guide — tools move between tiers as assessments complete. When in doubt, the approved AI tools list is authoritative.