A regional sales engineer visits clients with only a browser and no cloned code: she demos repo Q&A from documentation, while a platform engineer in a locked-down terminal must generate shell commands without leaving the shell. Which interface distinction best guides Copilot Chat on github.com, the Copilot app, and Copilot CLI for these constraints?

The correct pairing aligns interface affordances to access constraints: Copilot Chat on github.com runs entirely in the browser with repository-aware context, allowing documentation, issues, pull requests and code search to be queried without cloning, which fits the traveling demo, while Copilot CLI runs natively inside the shell to explain and generate shell commands without context-switching, which fits the locked-down terminal. The VS Code integrated terminal pairing fails on environment constraint because it presupposes an installable IDE and workspace access that the locked-down host explicitly prohibits despite being excellent for everyday shell guidance. The autonomous coding agent pairing fails on interaction model because the agent works asynchronously on branches to draft scripts for later review rather than providing synchronous conversational assistance during a live demo or inside an active shell session. The standalone Copilot app pairing fails on repository grounding because without the github.com chat surface it lacks immediate, URL-addressable repo Q&A over documentation and code, which matters when no clone exists. Exam caveat: distinguish surface location from capability and always match browser-only work to github.com chat and shell-only work to CLI rather than assuming any Copilot answers equally anywhere. Practical check: verify browser access to repository chat and run a test explain query with Copilot CLI in the restricted shell before client work begins.