Rebuild the patterns catalogue website — replace Docusaurus

[russelltrow]

Context

The current patterns catalogue runs on Docusaurus, which converts Markdown files into a static HTML site. This was adequate when the catalogue was first created, but the platform now constrains the project in several ways: it limits navigation to a single hierarchy, doesn't support filtering or tagging, can't offer persona-based browsing, and makes the contribution process dependent on GitHub pull requests. These limitations are blocking the delivery of the revised Patterns 2.0 catalogue structure.

With AI-assisted development now practical (Francesca has vibe-coded a prototype; Russ rebuilt one using Claude Code), the team agreed there's no longer a need to be constrained by Docusaurus. The goal is to build a new site purpose-built for the patterns catalogue that supports the navigation, filtering, and contribution workflows the project needs.

What needs to happen

Phase 1 — Prototype (immediate)

• Follow up with Francesca on her existing prototype — understand the tech stack, what it covers, and whether it can be extended
• Evaluate build options: (a) extend Francesca's prototype, (b) vibe-code a new site, or (c) use a Notion database as a rapid prototype
• Build a working prototype populated with a representative sample of patterns
• Share the prototype with the team for feedback before the next meeting

Phase 2 — Production site

• Based on prototype feedback, agree the final tech stack and hosting approach
• The production site must support:
  • Category-based navigation with subcategories
  • Persona/role-based browsing (homepage entry point)
  • Tag-based filtering (cross-cutting topics)
  • Individual pattern pages with full content and new properties (author, impact indicators)
  • Form-based pattern submission with review/approval workflow (modelled on Policy Radar)
  • Contributor recognition (author profiles, badges)
• Migrate all existing patterns from the Docusaurus site to the new platform
• Populate the backend data store (Notion database or equivalent) with all pattern metadata
• Launch the new site ahead of the SCI for AI publication (Q2 2026)

Phase 3 — Decommission Docusaurus

• Redirect the old patterns site URL to the new site
• Archive the Docusaurus repository

Acceptance criteria

• New site is live, publicly accessible, and contains all existing patterns
• Users can browse patterns by category, by persona, and by tag
• Contributors can submit new patterns via a form (no GitHub PR required)
• Submissions go through a review/approval workflow before publishing
• Old Docusaurus site is redirected to the new site
• Launched ahead of SCI for AI publication

Notes

• The prototype phase is about moving from spreadsheet discussions to tangible evaluation — it doesn't need to be production-ready.
• If using Notion as the backend, consider whether it can also serve as the long-term data store for pattern management (submission, review, approval).
• Removing the dependency on Docusaurus also removes the need for contributors to understand Markdown and Git, which should lower the barrier to contribution significantly.