The research website for Protocol Labs, live at https://www.plrd.org/. The canonical repository is https://github.com/protocol/plrd.org.
Built with Next.js 15.5, React 19, and Tailwind CSS 4. Research content is authored in Markdown with YAML frontmatter and compiled to JSON; page sections and opportunity overrides are repository JSON. All website editing goes through GitHub, not a website login or ATProto writer. See the content-source map.
npm install
npm run dev # builds content + starts dev serverOpen http://localhost:3000.
| Command | Description |
|---|---|
npm run dev |
Build content from Markdown, start Next.js dev server |
npm run build |
Production build (content + Next.js) |
npm start |
Start production server |
npx tsc --noEmit |
Type-check without emitting files |
npm run build-sheets |
Print the Sovereign AI sheets in src/data/fa2/sovereign-ai/ to public/sovereign-ai/*.pdf, with a cover plate each (needs local Chrome, poppler and ImageMagick) |
Run npm test, npx tsc --noEmit, and npm run build before submitting a PR. Node node:test is the standard test runner. Production installs use pnpm 10 with pnpm install --frozen-lockfile; keep package-lock.json synchronized for npm users.
plrd-v2/
content/ # Markdown source files
authors/<slug>/_index.md # Team member profiles (~74)
publications/<slug>/index.md # Research papers (~92)
talks/<slug>/index.md # Conference talks (~61)
tutorials/<slug>/_index.md # Tutorial series (3)
blog/<slug>/index.md # Blog posts
areas/<slug>/_index.md # Focus areas (4, with leads in frontmatter)
outreach/<name>.md # Outreach pages (6)
<section>/_index.md # Section index pages
scripts/
build-content.mjs # Prebuild: Markdown -> JSON + RSS + search index
src/
app/ # Next.js App Router pages
components/ # Shared React components
data/
generated/ # Auto-generated JSON from content/ (checked into git)
fa2/ # Hand-curated FA2 focus area data
lib/
content.ts # Typed exports from generated JSON
site-config.ts # Site metadata, navigation structure
public/
icons/ # SVG icons for social links, UI
images/ # Static images (authors, heroes, logos)
feed.xml # RSS feed (auto-generated)
search-index.json # Client-side search index (auto-generated)
Markdown-backed research content follows this flow; directly authored page/opportunity JSON follows the separate content-source map:
content/*.md --> build-content.mjs --> src/data/generated/*.json --> App Router pages
- Author -- Markdown files in
content/with YAML frontmatter define all research content (publications, authors, talks, etc.) - Build --
scripts/build-content.mjsruns at build time (prebuildhook). It reads every Markdown file, parses frontmatter withgray-matter, renders the body to HTML withremark, and writes typed JSON arrays tosrc/data/generated/. - Consume --
src/lib/content.tsimports the generated JSON and exports typed arrays (publications,authors,talks, etc.) used by page components. - Extras -- The build script also generates
public/feed.xml(RSS) andpublic/search-index.json(for Fuse.js client-side search).
The generated JSON files are checked into git because Vercel needs them at build time.
See CONTRIBUTING.md for how to add and edit content.
The Radar on /insights/ uses repository content only: the newest talks, publications, and listed blog posts from src/lib/content.ts, plus one hand-picked field signal from src/lib/radar-signals.ts. It selects up to six items by recency, displays PL R&D content before the field signal, and labels the edition with the newest selected item's month. The existing collapsible/swipeable UI is unchanged.
There is no external Radar feed fetch or Curator configuration. External story links in the local content remain links, not ingestion. Former CMS copy is repository-owned as part of the GitHub content migration; there is no website editor.
Run node --test scripts/radar-local.test.mjs for the Radar regression tests; they also run with npm test or pnpm test.
| Route | Description |
|---|---|
/ |
Homepage with hero, focus areas grid, recent updates |
/about/ |
About Protocol Labs R&D |
/areas/ |
Focus areas listing |
/areas/[slug]/ |
Individual focus area (Markdown-driven; leads defined in frontmatter) |
/areas/economies-governance/ |
FA2 focus area with sub-pages (impact, projects, opportunity spaces) — hardcoded, not Markdown |
/areas/economies-governance/sovereign-ai/[slug]/ |
The Sovereign AI program's one-page sheets, each with a printed A4 PDF at /sovereign-ai/<slug>.pdf; the FA2 page lists them in its Sovereign AI flagship box (#flagship-program, also reached by #sovereign-ai), and /sovereign-ai/ links there. Old /briefs/ and /programs/ links redirect here |
/research/ |
Research hub |
/publications/ |
All publications, sorted by date |
/publications/[slug]/ |
Publication detail |
/talks/ |
All talks, sorted by date |
/talks/[slug]/ |
Talk detail with YouTube embed |
/tutorials/ |
Tutorial series listing |
/team/ |
Team grid (/authors and /authors/ permanently redirect here) |
/authors/[slug]/ |
Author profile with bio, publications, talks |
/blog/ |
Blog posts |
/blog/[slug]/ |
Blog post detail |
All URLs use trailing slashes for Hugo compatibility.
The FA2 (Upgrade Economies & Governance) focus area has hand-curated JSON data in src/data/fa2/:
projects.json-- 242 ecosystem teams/projects with metadata (name, logo, website, tier, tags, etc.)impact.json-- Impact dashboard data with headline metrics and initiative detailsopportunityspaces.json-- local image precedence, route generation, fallback fields, and legacy opportunity-space views. Active EG overview/detail prose is insrc/data/site-content/opportunities.json, not this older fallback.sovereign-ai/<slug>.json-- the Sovereign AI program's one-page sheets for governments and funders: every word, figure and source. The web edition renders the JSON;npm run build-sheetsprints it topublic/sovereign-ai/<slug>.pdfand a cover plate, both committed. See AGENTS.md §7
These files are edited directly as JSON and consumed by hardcoded FA2 sub-pages under /areas/economies-governance/.
Website login, CMS editing, browser publishing, and local community deletion are retired. Use GitHub pull requests for Markdown/JSON content changes. Legacy editor links explain the workflow; old auth/write APIs return 410 with Cache-Control: no-store, and status is always anonymous. Public Hypercert, community, profile, DID, and metric reads remain.
See the exact legacy route and canonical-source map. This retirement must be integrated with the companion repository snapshot migration; it does not delete public records or source accounts.
No cookie secret, OAuth/JWK, or ATProto app-password configuration is needed to build or run the website. Analytics and optional read-only metric configuration are separate.
| Variable | Required | Description |
|---|---|---|
NEXT_PUBLIC_GA_ID |
No | Google Analytics 4 Measurement ID (empty disables analytics) |
NEXT_PUBLIC_COOKIE_CONSENT |
No | off hides the analytics consent banner + footer link and loads analytics directly; anything else (default) shows the consent banner |
Deployed to Vercel from the GitHub repo protocol/plrd.org. Pushes to main trigger auto-deployment.
The vercel.json config is minimal:
{
"framework": "nextjs",
"buildCommand": "npm run build"
}- Tailwind CSS 4 with
@tailwindcss/postcss - Theme defined in
src/app/globals.cssvia@themeblock - Key colors:
--color-blue: #1982F4,--color-pink: #E51A66,--color-black: #131316 - Font: Inter (Google Fonts)
- Responsive: mobile-first with
md:andlg:breakpoints - Common patterns:
max-w-4xl mx-auto px-6,text-gray-600,divide-y divide-gray-200
src/data/generated/is checked into git -- Vercel needs it at build time*.pdffiles are gitignored (removed to reduce repo size), except the Sovereign AI sheets inpublic/sovereign-ai/- Images must use
unoptimized: true(configured innext.config.ts) - Hugo shortcodes like
{{< youtube ID >}}in talk Markdown are rendered as iframes at the component level - After deleting pages, remove
.next/to clear stale type cache before type-checking - FA2 sub-pages (
impact/,projects/,opportunity-spaces/) are hardcoded routes, not generated from Markdown - The path alias
@/*maps to./src/*-- always use it instead of relative imports
Website editing and publishing are GitHub-only. Public read-only ATProto integrations remain separate from content ownership; do not reintroduce OAuth or credential-based website publishing.