dogfood

name	dogfood
description	Validate Toy Box behavior after significant changes using Playwright CLI across multi-client sessions, streaming flows, and terminal reconnect paths. Use when changes touch session hooks/reducers/adapters/cache/server functions, SSE or streaming transport logic, list/detail data fetching and invalidation behavior, terminal WebSocket/PTY logic, or route lifecycle/page visibility handling.

Dogfooding

Run Toy Box dogfooding scenarios after significant changes and report pass/fail with evidence.

Execute Workflow

1. Confirm whether the full workflow is required:

• Run this skill when touched areas include session state, streaming transport, realtime invalidation, terminal WS/PTY, or visibility/lifecycle logic.
• If none of those areas changed, state that dogfooding was skipped and why.

2. Start local runtime and browser sessions:

• Run bun dev.
• Open at least two sessions:
  • playwright-cli --session tabA open http://localhost:3100
  • playwright-cli --session tabB open http://localhost:3100
• Add a mobile session when layout or visibility flows changed.

3. Run scenario checks:

• Use playwright-cli --session <name> snapshot before any click/fill/press actions.
• Follow all required checks in references/validation-checklist.md.
• Refresh snapshots after key transitions because refs are snapshot-specific.

4. Collect validation evidence:

• Capture snapshots or concrete UI observations before and after key actions.
• Check browser console for regressions.
• Check browser network for expected SSE/stream/WS traffic.
• Record a pass/fail result for each scenario group.

5. Produce report:

• Return a concise summary grouped by:
  • Sessions List
  • Session Detail
  • Multi-Client Realtime
  • Terminal (WS + PTY)
• Include reproduction steps for each failure and likely impacted area when obvious.

6. Clean up sessions:

• Run playwright-cli --session <name> session-stop for specific sessions or playwright-cli session-stop-all.
• Keep the dev server running only if user asked for continued manual testing.

Use Reference

• Read references/validation-checklist.md for command quick reference, must-pass scenarios, and evidence expectations.