test(vibe): design stacked-sheet API comparison

[imdreamrunner]

Summary

Design a preregistered four-way vibe test for stacked bottom-sheet APIs:

1. Silk-style declarative stack membership
2. Drawer-style implicit sibling stacking
3. the original controlled ordered-ID proposal
4. ordered IDs with explicit modality, scrim, and focus policy

This PR contains the experiment design only—no model runs and no candidate implementation change.

Experiment shape

• 8 identical user-outcome prompts: 6 core state-model probes and 2 separately reported interaction-policy probes
• 4 condition-specific contract drafts with the same documentation sections and a shared probe checklist
• condition-blind rubric with TypeScript, browser behavior, accessibility, escape-hatch, friction, and change-request metrics
• 2 holdout prompts, random opaque condition labels, isolated package copies, and a preregistered decision rule
• pilot: 4 conditions × 4 prompts × 3 reps = 48 runs
• confirmation: all 8 prompts, K≥5, strong + mid-tier coding models

The test explicitly does not measure library implementation complexity or visual preference; Phase 1 will put thin public adapters over one shared private runtime.

Key fairness choices

• core and policy tracks never share one score
• exact same product prompt, model, runtime appearance, and evaluator across conditions
• no candidate API term appears in agent-visible prompt text
• evaluators score behavior, not whether code uses booleans or an ordered array
• compilation, runtime behavior, and accessibility gates override aggregate scores

Validation

• node internal/vibe-tests/stacked-sheet-api-test/smoke-test.mjs
• pnpm check:repo
• pre-commit checks

The no-model smoke test verifies prompt purity, holdout isolation, condition-blind scoring, equal contract coverage, unique IDs, and variant-doc length parity.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
astryx	Ready	Preview	Sep 2, 2026 7:00pm UTC

[github-actions]

PR Analysis Report

Preview availability: Storybook and Sandbox were not published for this CI run.

No new or modified components detected.

Bundle Size Summary

No component packages changed.

Accessibility Audit

Status: No accessibility violations detected.

---

_{Generated by PR Enrichment workflow | View full report}