GH-300 decision sheets

• Content exclusion: what it protects and what still leaks
• Copilot CLI: platforms, install methods and session commands
• Copilot safeguards: human-centred controls vs technical mitigations
• Which Copilot surface: coding agent vs code review vs Spaces vs CLI
• Using inline suggestions well: intent, review, accept
• Supplying context to Copilot: comments vs few-shot examples vs selection
• Structuring a Copilot prompt: general to specific
• Copilot risk categories: hallucination vs public code matching
• Copilot seat REST API: read subscription vs grant seats vs check one user
• Copilot for unit tests: how to ask and how to reuse