Welcome — Plugin Onboarding — Design Doc

Status: Early prototype · Last updated: 2026-07-08

Problem

New VS Code users land in an empty editor with no clear first move. The extension marketplace is powerful but overwhelming — thousands of options, no guidance on what a person building their kind of software actually needs. The first-run experience should turn that blank slate into momentum: help someone get a relevant, working toolset installed in under a minute, without decision paralysis.

Solution

A Welcome editor tab that runs a three-step, guided onboarding:

  1. Focus — the user picks what they're building (Frontend, Backend, AI, or DevOps). This is a single low-stakes choice that curates everything after it.
  2. Extensions — a curated shortlist filtered to that focus, with the essentials pre-selected. The user can toggle any item, then installs them all at once. Each install animates through queued → installing (progress bar) → installed, so the batch feels alive and trustworthy rather than instant/fake.
  3. Done — a celebratory summary and three concrete next actions (open a folder, ask Copilot, tune settings) so momentum carries past onboarding.

A persistent stepper anchors the flow, and the surface is framed inside a real editor tab strip so it reads as an actual VS Code Welcome tab, not a mockup.

Validated Patterns

  • Pre-selection over blank choice — defaulting the essentials to "checked" removes decision paralysis while still leaving the user in control.
  • Batch install with staggered progress — installing sequentially with a per-item progress bar communicates real work and makes the moment feel earned, versus an instant "all installed" flip.
  • Focus-first curation — one upfront choice keeps every later screen short and relevant instead of showing the full marketplace.

Open Questions

  1. Should the focus choice allow multiple selections (e.g. someone who is both frontend and AI), and how does that affect the curated list length?
  2. Is a skip/"I'll do this later" affordance needed, or does forcing one focus choice produce better activation?
  3. How should already-installed extensions surface if the user re-runs onboarding — hidden, or shown as "installed" and skipped?