/Docs
View Markdown Sign in
For the full documentation index, see PageWeave Documentation or fetch llms.txt.

Core Concepts

Understanding PageWeave’s architecture and data model.

Websites

A website is the top-level entity. Each website has:

  • Subdomain — auto-generated (e.g., my-site.pageweave.site)
  • Custom domains — optional, via nameserver delegation or self-managed DNS; SSL automatic
  • Status — draft or published
  • Theme — daisyUI theme applied to all pages
  • Language — BCP 47 tag (e.g., en, de). For multilingual sites, see the Internationalization (i18n) guide.

Draft vs Published

  • Draft sites: Edits apply directly. Agent iterates freely. No review gate.
  • Published sites: Agent edits create pending versions. User must approve before going live.

Pages

Pages are versioned HTML content at URL paths:

  • Path — URL anchor (e.g., /, /about, /contact)
  • Versions — every edit creates a new version
  • Title — page title for <title> and metadata
  • HTML content — raw HTML body
  • Metadata — description, og_title, og_image (JSONB)

Versioning

Every edit creates a new version with:

  • Sequential version number
  • Timestamp
  • Published status (nil = draft, set = published)
  • Rollback reference

Components

Header & Footer

Instance-based HTML components with versioning. A website can have multiple header or footer instances. The default instance is used on all pages.

Each instance supports versioning (same model as pages). The default is set per website via default_header_id / default_footer_id.

HTML Head

Instance-based <head> content with versioning — meta tags, tracking scripts, analytics, favicons. Uses the same instance model as header/footer. The default instance is set via default_html_head_id.

Themes

daisyUI 5 + Tailwind CSS v4 theme system:

  • Built-in themes — 35+ themes (nord, sunset, dark, etc.)
  • Custom themes — OKLCH colors with [data-theme="name"] selector
  • Instant changes — CSS variables update, no re-render needed

Snippets

Versioned text files served at /snippets/<path> on the website’s domain:

  • Types — CSS, JS, JSON, HTML, SVG, XML, YAML, Markdown, CSV, plain text
  • Path — relative, no leading slash (css/style.css, js/app.js)
  • Versioning — every edit creates a new version; pin/revert like pages
  • Size limit — 256 KB per version
  • Content type — inferred from extension when omitted
  • Skip-unchanged — identical content creates no new version

Use snippets for reusable code/data referenced from multiple pages (<link>, <script src>, fetch()). For binary files (images, fonts, video, PDFs), use assets.

Assets

Binary file uploads with automatic optimization:

  • Images — WebP + AVIF variants generated
  • Fonts — WOFF2, TTF, OTF supported
  • Documents — PDF, DOCX, XLSX, PPTX
  • UUID-based URLs — prevents enumeration
  • Size limits — 5MB images, 2.5MB fonts, 10MB other

Image Rendering

Images are automatically enhanced at render time:

  • Responsive srcset — size variants (320w, 640w, 1024w) injected automatically
  • Width/height attributes — auto-injected from asset metadata to prevent layout shift (CLS)
  • Lazy loading — images below the fold get loading="lazy" automatically

Use standard <img src="..."> without srcset. If you need CSS-controlled sizing (e.g., class="w-full", class="h-8"), add data-no-dimensions to skip auto-injection of width/height and avoid aspect-ratio distortion.

Forms

Versioned form builder:

  • Email notifications
  • Webhook integration
  • Spam protection (Altcha)
  • Submission management

Tables

Structured data per website with three access methods:

  • MCP tools — Full CRUD for tables and rows (create_table, list_table_rows, etc.)
  • Liquid templating — site.tables.slug in page HTML with filtering, ordering, and pagination
  • Public JSON API — GET /t/{table_id} on the website’s domain. No auth required. Supports filtering, sorting, search, and pagination.

Each table has a schema with typed fields (string, text, number, boolean, datetime). Rows auto-generate slugs from a configured display_column_key.

Full documentation: Data Tables

Domains

Custom domain management:

  • Nameserver delegation (PageWeave manages DNS) or self-managed records
  • Automatic SSL via on-demand TLS
  • DNS propagation checking

What PageWeave Replaces

PageWeave is a complete platform. You don’t need external services for hosting, databases, or deployment.

Feature PageWeave Traditional Stack
Hosting Built-in, instant deploy Vercel, Netlify, AWS
Database Data tables with Liquid/JSON API Neon, PlanetScale, MongoDB
DNS & SSL Automatic, managed Cloudflare, Route53, Let’s Encrypt
Forms Built-in with email notifications Formspree, custom backend
Analytics Privacy-first, no cookies Google Analytics, Plausible
Assets Auto-optimized images, fonts Cloudinary, S3
Versioning Every change versioned Git + CI/CD pipeline
Deployment Instant, no build step GitHub Actions, Docker

No External Tools Required

When your AI agent creates a website on PageWeave, it’s immediately live. No deployment pipeline, no external hosting, no database setup. Everything runs on PageWeave’s infrastructure.

You don’t need:

  • GitHub repositories
  • Vercel/Netlify accounts
  • External database services
  • Separate DNS providers
  • Build tools or CI/CD

Your agent builds directly on PageWeave. The site is live at your-site.pageweave.site the moment it’s created. Custom domains can be added anytime with automatic SSL.