PR #78 — BS-5 Phase 1: Standalone Website Scaffold

The Broader Goal at the End of the Previous PR

At the end of v0.9.8 (health pulse + lifecycle commands), the CLI was hardened for clean onboarding, exit, repair, and observability. The understood next step was shifting focus toward public launch preparation — making the project discoverable and understandable to new users beyond the GitHub repo and internal docs.

Strategic Shifts in This PR (if any)

BS-5 is the first concrete step toward a real synlynk.com. This moves synlynk from "CLI tool with project-docs" to "product with a face". The website/ directory is a standalone 11ty v3 site (parallel to legacy site/) explicitly scoped as design-first. Phase 1 (Grok) delivers the scaffold, tokens, structure, and handoff surface so that Phase 2 (Agy) can focus on content fidelity, full visual system, and subpages without fighting structural or specificity problems.

What This PR Shipped

Visual direction study
Page structure plan
  • website/ 11ty v3 project:
    • package.json (eleventy ^3, scripts for build/serve)
    • .eleventy.js with passthrough, dateFilter, stub collections
    • src/_includes/base.njk: fixed nav, font preconnects, footer stub, content slot
    • src/index.njk: 8 section shells with correct IDs (#top #carousel #relief #how #features #vision #docs #waitlist)
    • src/assets/css/main.css: design tokens from spec/hero-v4, reset, nav, primitive section rules, responsive
    • S-glyph logo (from brainstorm 03-terminal-s-glyph.svg) placed at src/assets/img/logo/s-glyph.svg and used in nav
  • Dark/light alternation primitives and typography stack (Inter + JetBrains Mono) matching the visual system
  • Clean agent role split demonstrated: Grok authored scaffold + structure + build; no content fidelity beyond shells
  • Addressed all R1 Claude review findings in this branch:
    • Scoped section p muted color only to dark sections (#top p, ...)
    • Converted ID rules (#top, #...) to .section-dark / .section-light classes + applied classes to sections in index.njk
    • Stripped every inline style= attribute from the entire #top hero section (hero-headline / hero-byline classes preserved as pure hooks)
    • Replaced onclick getElementById('waitlist') button with <a href="/#waitlist"> link (crash-proof on subpages)
    • Removed duplicate @import Google Fonts line from main.css (base.njk <link> + preconnect is sufficient)
    • Fixed dateFilter to force local-day parsing for ISO date strings (date + 'T00:00:00')
  • Blog post (this file) per CLAUDE.md convention

cd website && npm install && npm run build produces a working _site/ with nav + all 8 sections.

Brainstorm Visuals Used

  • docs/brainstorm/bs5-website-redesign/hero-v4.html (hero, layout, tokens, color/alternation direction)
  • Supporting logo and diagram assets referenced in docs/superpowers/specs/2026-06-27-bs5-website-redesign-design.md

What This Achieved on the Path to Autonomy

This PR puts the multi-agent workflow on public display: different agents (Grok scaffold/animations, Agy templates+CSS+copy, Claude reviewer) collaborate on a real product artifact with clear ownership boundaries and clean handoff contracts. The specificity and style-stripping fixes ensure Agy can own the design system in Phase 2 without cascade fights or !important. The site itself is the external manifestation of "the OS for multi-agent development" — the thing a new visitor lands on and immediately understands the value of workgroups, shared state, and model diversity.

Strategic Note: The Goal at the End of This PR

Phase 1 complete: runnable 11ty build, correct tokens and structure, Phase 2 handoff surface ready. New goalpost: Agy fills the 8 sections with full content and micro-copy, implements the complete CSS design system (cards, components, states, responsive details), adds the blog index + subpages (docs, features, changelog, about), wires any interactive shells (carousel etc.), and ensures visual parity with the hero-v4 reference. Then full review + merge to unblock the rest of the launch arc.