Projects collection routes and searchCopy Anchor Link

Discovery state is encoded in paths so a collection can be copied, indexed, and revisited without a query-string-only UI.

Project type segments are an explicit exhaustive mapping from the shared ProjectType enum. Category, topic, and search values are Unicode-normalized, trimmed, lowercased, converted to hyphen-separated segments, and rejected when empty or equivalent to . or ... Route pages reject non-canonical values and invalid page suffixes rather than creating duplicate URLs.

Page one never carries /page/1. Pagination contains 12 projects per page and clamps an in-memory requested page to the available range; server route parsing still rejects malformed or non-canonical path syntax.

Filtering and searchCopy Anchor Link

Type filters compare the exact schema enum. Category and topic filters compare canonicalized published values. Search splits the canonical path segment into terms and requires every term to occur in a normalized text corpus containing the title, descriptions, type, status, academic context, course, categories, topics, people fields, and links.

Search pages use canonical metadata but are marked noindex, follow. Browse, type, category, and topic pages can be indexed. The sitemap derives populated collections from published records and emits every required pagination path.

Old root query keys such as q, category, topic, type, and page are not translated into the path model. Their presence causes a redirect to the clean home URL while unrelated query parameters are retained. New discovery features must define and test a canonical path grammar before adding controls.