@ibalzam/codejitsu-core 0.1.0

package/CLAUDE.md

@@ -0,0 +1,67 @@
+# Master instructions — @ibalzam/codejitsu-core
+
+This is the shared core for all Codejitsu sites. When a Codejitsu site depends on this package and the user invokes a module by name (e.g. **"implement codejitsu/core/blog"**, **"add codejitsu/core/seo"**), this file is your starting point.
+
+## How to act on a module request
+
+1. Open `node_modules/@ibalzam/codejitsu-core/modules/<name>/CLAUDE.md` — it tells you what to do for that module specifically.
+2. Import code from the matching subpath (e.g. `@ibalzam/codejitsu-core/blog`). Do **not** copy-paste the source into the site.
+3. If the module has a `templates/` directory, copy those files into the site at the locations the module's CLAUDE.md specifies. Templates are starting points — adapt them for the site's content.
+4. After the change, run the module's `checklist.md` mentally, plus `checklist/core.md` (sitewide).
+
+## Principles that apply to every Codejitsu site
+
+These are non-negotiable unless the user explicitly opts out:
+
+### Stack
+- **Astro** (latest stable). Pure static output (`output: 'static'`).
+- **Tailwind v4** via `@tailwindcss/vite`. Theme via CSS variables on `:root`, not hardcoded color palettes.
+- **TypeScript** everywhere except config files where Astro/Vite expects `.mjs`.
+- **React** integration only if the site needs interactive client islands (Framer, charts). Otherwise pure Astro.
+
+### Deploy
+- **Cloudflare Pages**, static deploy. `wrangler.toml` at site root.
+- `npm run build && npx wrangler pages deploy dist` is the deploy command (or git-integration on Pages).
+- Daily GH Action (`.github/workflows/daily-deploy.yml`) pings a Cloudflare deploy hook to publish scheduled content. See `modules/deploy/`.
+
+### URLs + routing
+- `trailingSlash: 'always'` in Astro config. Every internal link ends with `/`.
+- Canonical URLs are absolute and trailing-slashed.
+
+### Images
+- Source images in `public/images/` (or `src/assets/` for Astro-processed). Originals can be PNG/JPG.
+- Every shipped image must be available as WebP. The Astro sharp service handles `<Image>` references automatically (`image.defaults: { quality: 82, format: 'webp' }`). For images referenced by URL in HTML/CSS, run the pre-pass: `npx codejitsu-optimize-images`.
+- No raw PNGs referenced from production HTML when a WebP equivalent exists.
+
+### SEO (must be on every page)
+- `<title>` and `<meta name="description">` set per page.
+- Canonical URL `<link rel="canonical">`.
+- OG meta (`og:title`, `og:description`, `og:image`, `og:url`, `og:type`).
+- Twitter card meta.
+- JSON-LD schema.org appropriate to the page type (Organization on home, LocalBusiness if applicable, BlogPosting on blog posts, FAQPage if FAQs present). Use builders from `@ibalzam/codejitsu-core/seo/schema`.
+- `sitemap.xml` generated via `@astrojs/sitemap` with helpers from `@ibalzam/codejitsu-core/seo/sitemap`.
+- `robots.txt` at site root.
+- `/llms.txt` + `/llms-full.txt` generated via `npx codejitsu-llms` in prebuild.
+
+### Content
+- Blog posts as Markdown in `content/blog/` with frontmatter (see `modules/blog/CLAUDE.md`).
+- Future-dated posts are hidden from public pages and sitemap but kept addressable for OG meta scrapers.
+- All copy in English unless the site uses the i18n module (workzen-only currently).
+
+### What NOT to do
+- Don't reinvent modules that exist here. If you're tempted to write a blog loader, image optimizer, sitemap config, schema builder, or daily-deploy workflow, **import or copy from this package instead**.
+- Don't hardcode brand colors in component files — always via Tailwind theme tokens / CSS variables.
+- Don't add a runtime database, server-rendered routes, or anything that breaks static export. These sites deploy as plain HTML.
+- Don't add page-level redirects in code; use Cloudflare Pages `_redirects` or `_headers`.
+- Don't introduce a new heavy dependency without checking if existing modules already cover it.
+
+## After any non-trivial change
+
+Run through `checklist/core.md` before reporting work as done. For UI changes, build the site and view the pages in a browser — type-checking is not visual verification.
+
+## Updating to a new version of `@ibalzam/codejitsu-core`
+
+1. `npm update @ibalzam/codejitsu-core`
+2. Read `node_modules/@ibalzam/codejitsu-core/MIGRATIONS/` for any version notes newer than the previous installed version.
+3. Apply migration steps in order. They're prose, not codemods — judgment is fine; ask the user when ambiguous.
+4. Run `checklist/core.md` to verify nothing regressed.

package/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Ika Balzam
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/MIGRATIONS/README.md

@@ -0,0 +1,30 @@
+# Migrations
+
+When `@ibalzam/codejitsu-core` ships a change that requires touching site-owned files (page routes, config, content), the change is described here as prose Claude reads and applies.
+
+## Format
+
+One file per version: `MIGRATIONS/<version>.md`. Example: `MIGRATIONS/0.2.0.md`.
+
+Each file follows this structure:
+
+```markdown
+# 0.2.0
+
+## Summary
+One paragraph: what changed and why.
+
+## Required actions
+Numbered list of concrete things to do in the site. Reference file paths and code snippets. Be specific enough that Claude can apply them without ambiguity.
+
+## Verify
+What to check after applying.
+```
+
+## Why prose, not codemods
+
+We do migrations through Claude reading the notes, not jscodeshift. Trade-off: prose is more flexible (handles edge cases, branding variations, half-applied previous migrations) but requires Claude to actually be in the loop on the upgrade. That matches how this package is meant to be used — every Codejitsu site is maintained with Claude, so Claude can read and apply.
+
+## When NOT to write a migration
+
+Most upgrades don't need one. If a change can live entirely in library code (a new component, a fixed function, an Astro integration hook that injects something automatically), ship it as a package update and let `npm update` propagate. Only write a migration when site-owned files must change.

package/README.md

@@ -0,0 +1,35 @@
+# @ibalzam/codejitsu-core
+
+Shared core for all Codejitsu sites. Two channels per module:
+
+- **Code** — importable via `@ibalzam/codejitsu-core/<module>` (Astro/TS).
+- **Instructions** — a `CLAUDE.md` per module that tells Claude how to wire it into a site, what to do, what to avoid.
+
+Sites stay thin: configuration + content + brand. Everything else lives here.
+
+## Modules
+
+| Module | Subpath | What it provides |
+|---|---|---|
+| `blog` | `@ibalzam/codejitsu-core/blog` | Markdown-based blog system with scheduled publishing, FAQs, tags, categories |
+| `seo` | `@ibalzam/codejitsu-core/seo` | Sitemap helpers, schema.org JSON-LD builders, meta tag patterns |
+| `images` | `@ibalzam/codejitsu-core/images` + `codejitsu-optimize-images` CLI | PNG/JPG→WebP pre-pass + Astro sharp defaults |
+| `deploy` | (templates only) | GH Action daily-deploy + Cloudflare wrangler templates |
+| `llms` | `codejitsu-llms` CLI | Generates `/llms.txt` + `/llms-full.txt` from site content |
+
+Plus `checklist/` — sitewide invariants Claude verifies after any non-trivial change.
+
+## How Claude uses this package
+
+In a site that depends on `@ibalzam/codejitsu-core`, when the user says **"implement codejitsu/core/blog"** (or animations, or deploy, etc.), Claude reads `node_modules/@ibalzam/codejitsu-core/modules/<name>/CLAUDE.md` for instructions, imports code from the matching subpath, copies any templates, and runs the module's `checklist.md` to verify the result.
+
+Start at `CLAUDE.md` (the master entry).
+
+## Updating a site after a core release
+
+```bash
+npm update @ibalzam/codejitsu-core
+# Then tell Claude: "we just upgraded codejitsu/core, check MIGRATIONS for anything that needs applying"
+```
+
+Migration notes live in `MIGRATIONS/<version>.md` as prose Claude reads and applies (not jscodeshift codemods — Claude does the work).

package/checklist/bin/run.mjs

@@ -0,0 +1,189 @@
+#!/usr/bin/env node
+/**
+ * Smoke checker for Codejitsu sites. Run from a site repo root after
+ * `npm run build`. Verifies sitewide invariants from `checklist/core.md`
+ * that can be checked programmatically.
+ *
+ * Exit code 0 = all checks pass.
+ * Exit code 1 = at least one check failed (warnings still exit 0).
+ *
+ * Not exhaustive — visual/UX/content checks need human + Claude review.
+ */
+import fs from 'fs';
+import path from 'path';
+
+const cwd = process.cwd();
+const distDir = path.join(cwd, 'dist');
+
+const checks = [];
+
+function pass(name) {
+  checks.push({ status: 'pass', name });
+}
+function warn(name, detail) {
+  checks.push({ status: 'warn', name, detail });
+}
+function fail(name, detail) {
+  checks.push({ status: 'fail', name, detail });
+}
+
+function exists(p) {
+  return fs.existsSync(path.join(cwd, p));
+}
+
+function distHtmlFiles() {
+  if (!fs.existsSync(distDir)) return [];
+  const out = [];
+  (function walk(dir) {
+    for (const entry of fs.readdirSync(dir, { withFileTypes: true })) {
+      const full = path.join(dir, entry.name);
+      if (entry.isDirectory()) walk(full);
+      else if (entry.name.endsWith('.html')) out.push(full);
+    }
+  })(distDir);
+  return out;
+}
+
+function readFile(p) {
+  return fs.readFileSync(p, 'utf8');
+}
+
+// ─── Pre-build files ─────────────────────────────────────────────────────
+
+if (!exists('wrangler.toml')) fail('wrangler.toml present');
+else pass('wrangler.toml present');
+
+if (!exists('.github/workflows/daily-deploy.yml'))
+  warn('.github/workflows/daily-deploy.yml present', 'Skip only if site has no scheduled content.');
+else pass('.github/workflows/daily-deploy.yml present');
+
+if (!exists('astro.config.mjs') && !exists('astro.config.ts'))
+  fail('astro.config.{mjs,ts} present');
+else {
+  const cfg = readFile(
+    path.join(cwd, exists('astro.config.ts') ? 'astro.config.ts' : 'astro.config.mjs')
+  );
+  if (!/trailingSlash:\s*['"]always['"]/.test(cfg))
+    fail("trailingSlash: 'always' in astro.config", 'Required for canonical URL policy.');
+  else pass("trailingSlash: 'always' in astro.config");
+
+  if (!/output:\s*['"]static['"]/.test(cfg))
+    fail("output: 'static' in astro.config");
+  else pass("output: 'static' in astro.config");
+
+  if (!/format:\s*['"]webp['"]/.test(cfg))
+    warn("image.defaults.format: 'webp' in astro.config");
+  else pass("image.defaults.format: 'webp' in astro.config");
+}
+
+// ─── Build artifacts ─────────────────────────────────────────────────────
+
+if (!fs.existsSync(distDir)) {
+  fail('dist/ exists', 'Run `npm run build` first.');
+  printAndExit();
+}
+pass('dist/ exists');
+
+const htmlFiles = distHtmlFiles();
+if (htmlFiles.length === 0) fail('dist/ contains HTML files');
+else pass(`dist/ contains ${htmlFiles.length} HTML files`);
+
+const hasSitemap = fs.existsSync(path.join(distDir, 'sitemap-index.xml')) ||
+  fs.existsSync(path.join(distDir, 'sitemap-0.xml'));
+if (!hasSitemap) fail('sitemap-(index|0).xml in dist/');
+else pass('sitemap in dist/');
+
+if (!fs.existsSync(path.join(distDir, 'robots.txt'))) fail('dist/robots.txt');
+else pass('dist/robots.txt');
+
+if (!fs.existsSync(path.join(distDir, 'llms.txt'))) warn('dist/llms.txt');
+else pass('dist/llms.txt');
+
+if (!fs.existsSync(path.join(distDir, 'llms-full.txt'))) warn('dist/llms-full.txt');
+else pass('dist/llms-full.txt');
+
+// ─── Per-page checks ─────────────────────────────────────────────────────
+
+const missingTitle = [];
+const missingDescription = [];
+const missingCanonical = [];
+const missingOgImage = [];
+const missingJsonLd = [];
+const pngWhereWebp = [];
+const placeholderText = [];
+
+const webpSet = new Set();
+(function walkAssets(dir) {
+  if (!fs.existsSync(dir)) return;
+  for (const entry of fs.readdirSync(dir, { withFileTypes: true })) {
+    const full = path.join(dir, entry.name);
+    if (entry.isDirectory()) walkAssets(full);
+    else if (entry.name.toLowerCase().endsWith('.webp')) {
+      webpSet.add(path.relative(distDir, full).replace(/\.webp$/i, ''));
+    }
+  }
+})(distDir);
+
+const PLACEHOLDER_RE = /\b(lorem ipsum|TODO|FIXME|XXX:|placeholder)\b/i;
+
+for (const file of htmlFiles) {
+  const rel = path.relative(distDir, file);
+  const html = readFile(file);
+
+  if (!/<title>[^<]+<\/title>/.test(html)) missingTitle.push(rel);
+  if (!/<meta\s+name=["']description["']\s+content=["'][^"']+["']/i.test(html))
+    missingDescription.push(rel);
+  if (!/<link\s+rel=["']canonical["']\s+href=["'][^"']+["']/i.test(html))
+    missingCanonical.push(rel);
+  if (!/<meta\s+property=["']og:image["']\s+content=["'][^"']+["']/i.test(html))
+    missingOgImage.push(rel);
+  if (!/<script\s+type=["']application\/ld\+json["']/i.test(html))
+    missingJsonLd.push(rel);
+
+  // Find <img src="*.png|*.jpg"> and check if a .webp equivalent exists.
+  for (const m of html.matchAll(/<img[^>]+src=["']([^"']+\.(?:png|jpe?g))["']/gi)) {
+    const src = m[1].replace(/^\//, '');
+    const noExt = src.replace(/\.(?:png|jpe?g)$/i, '');
+    if (webpSet.has(noExt)) pngWhereWebp.push(`${rel} → ${m[1]}`);
+  }
+
+  if (PLACEHOLDER_RE.test(html)) placeholderText.push(rel);
+}
+
+function reportList(name, list, severity = 'fail') {
+  if (list.length === 0) {
+    pass(name);
+    return;
+  }
+  const detail =
+    list.length <= 5
+      ? list.join(', ')
+      : `${list.slice(0, 5).join(', ')} … (+${list.length - 5} more)`;
+  (severity === 'fail' ? fail : warn)(name, detail);
+}
+
+reportList('Every page has <title>', missingTitle);
+reportList('Every page has <meta description>', missingDescription);
+reportList('Every page has canonical link', missingCanonical);
+reportList('Every page has og:image', missingOgImage, 'warn');
+reportList('Every page has JSON-LD schema', missingJsonLd);
+reportList('No <img> references raw PNG/JPG where WebP exists', pngWhereWebp);
+reportList('No placeholder text in production HTML', placeholderText);
+
+// ─── Output ──────────────────────────────────────────────────────────────
+
+printAndExit();
+
+function printAndExit() {
+  const passes = checks.filter((c) => c.status === 'pass').length;
+  const warns = checks.filter((c) => c.status === 'warn').length;
+  const fails = checks.filter((c) => c.status === 'fail').length;
+
+  for (const c of checks) {
+    const icon = c.status === 'pass' ? '✓' : c.status === 'warn' ? '!' : '✗';
+    const line = `${icon} ${c.name}`;
+    console.log(c.detail ? `${line}\n  ${c.detail}` : line);
+  }
+  console.log(`\n${passes} pass · ${warns} warn · ${fails} fail`);
+  process.exit(fails > 0 ? 1 : 0);
+}

package/checklist/core.md

@@ -0,0 +1,55 @@
+# Sitewide checklist — every Codejitsu site
+
+Walk through this list after any non-trivial change. Programmatic checks are runnable via `npx codejitsu-check` (runs from the site repo root after `npm run build`).
+
+The runner covers the easy stuff (file presence, meta tags, canonical, schema script tags, PNG-where-WebP-exists, placeholder text). The rest needs human + Claude judgment.
+
+## Build + deploy
+
+- [ ] `npm run build` exits 0 with no warnings about missing pages, unresolved imports, or broken links.
+- [ ] `dist/` contains static HTML for every expected route. No `.html` route is missing.
+- [ ] `wrangler.toml` is present and points to `dist`.
+- [ ] `.github/workflows/daily-deploy.yml` exists; the `CLOUDFLARE_DEPLOY_HOOK_URL` secret is set in the repo (skip if site has no scheduled content).
+
+## URLs + routing
+
+- [ ] Astro config has `trailingSlash: 'always'` and `output: 'static'`.
+- [ ] No internal link in any `.astro`/`.tsx` file points to a URL without a trailing slash.
+- [ ] Canonical URL appears on every page in `<head>` and matches the trailing-slash policy.
+
+## SEO
+
+- [ ] Every page has `<title>` and `<meta name="description">`. No duplicates across pages.
+- [ ] OG meta on every page: `og:title`, `og:description`, `og:image`, `og:url`, `og:type`.
+- [ ] Twitter card meta on every page.
+- [ ] JSON-LD schema appropriate per page type (Organization, LocalBusiness, BlogPosting, FAQPage). Use `@ibalzam/codejitsu-core/seo/schema` builders.
+- [ ] `sitemap-index.xml` generated and lists every public page. Future-dated blog posts are excluded.
+- [ ] `robots.txt` at root and references the sitemap.
+- [ ] `llms.txt` and `llms-full.txt` at root, regenerated this build.
+
+## Images
+
+- [ ] No `<img src="*.png">` or `<img src="*.jpg">` references in built HTML where a WebP equivalent exists.
+- [ ] Astro `image.defaults` includes `format: 'webp'`.
+- [ ] All images in `public/images/` over 500KB have been run through `codejitsu-optimize-images` or have a documented exception.
+
+## Performance
+
+- [ ] No client-side JS on pages that don't need it (Astro should ship zero JS by default).
+- [ ] `inlineStylesheets: 'always'` in build config (or justified opt-out).
+- [ ] Hero image uses `loading="eager"` + `fetchpriority="high"`; everything else lazy-loaded.
+
+## Accessibility
+
+- [ ] Every `<img>` has alt text (decorative images: `alt=""`).
+- [ ] Every interactive element is keyboard-reachable and has a visible focus style.
+- [ ] Color contrast meets WCAG AA on body text.
+
+## Content
+
+- [ ] No placeholder text (`Lorem ipsum`, `TODO`, `FIXME`) in shipped routes.
+- [ ] Every blog post in `content/blog/` has required frontmatter (see `modules/blog/checklist.md`).
+
+## When something on this list changes
+
+If a new invariant is added to a Codejitsu site, add it here AND bump the package version with a `MIGRATIONS/<version>.md` note so existing sites can adopt it.

package/modules/blog/CLAUDE.md

@@ -0,0 +1,87 @@
+# Blog module — instructions for Claude
+
+When the user asks to **implement codejitsu/core/blog** (or "add the blog system"), do the following.
+
+## What this module provides
+
+A markdown-based blog with:
+- File-based posts in `content/blog/*.md` (gray-matter frontmatter)
+- Scheduled publishing — future-dated posts are hidden from public pages and sitemap, but their slugs stay buildable so OG meta scrapers (Hootsuite, etc.) can hit them
+- Dual-slug resolution — filename slug (`2026-02-08-foo`) and canonical frontmatter slug (`foo`) both resolve to the same post; the frontmatter slug is canonical for SEO
+- Reading time, tags, FAQs, categories
+- Listing, tag, category pages
+
+## Wiring it into a site
+
+### 1. Install peer deps in the site (one-time)
+
+```bash
+npm install gray-matter reading-time
+```
+(They're transitive deps of `@ibalzam/codejitsu-core`, but Astro's bundler resolves them from the site's `node_modules`.)
+
+### 2. Create the site's blog instance
+
+Copy `templates/lib/blog.ts` → `src/lib/blog.ts` in the site. Edit the config to set the site's default author and (optionally) its category list. The whole file is ~10 lines.
+
+### 3. Add page routes
+
+Copy these from `templates/pages/` → `src/pages/` in the site:
+- `blog/index.astro` — listing
+- `blog/[...slug].astro` — detail (handles both filename and canonical slug forms)
+- `blog/tag/[tag].astro` — tag pages
+- `blog/category/[category].astro` — category pages (skip if site has no categories)
+
+Adapt the markup to the site's design system. The page logic (data fetching, getStaticPaths) is the part that must stay correct — styling is the site's job.
+
+### 4. Add the first post
+
+Copy `templates/content/_sample-post.md` → `content/blog/<today>-<slug>.md` and edit.
+
+### 5. Wire scheduled-post filter into the sitemap
+
+In `astro.config.mjs`, import the site's blog instance and exclude future-dated slugs from the sitemap:
+
+```ts
+import { blog } from './src/lib/blog';
+const futureSlugs = await blog.getFutureBlogSlugs();
+
+// in sitemap integration:
+filter: (page) => {
+  const m = page.match(/\/blog\/([^/]+)\/?$/);
+  return !(m && futureSlugs.includes(m[1]));
+}
+```
+
+### 6. Wire the daily-deploy GH Action
+
+See `modules/deploy/CLAUDE.md`. The cron rebuilds the site so scheduled posts graduate from hidden to public on their publish date.
+
+## Post frontmatter shape
+
+```yaml
+---
+title: "How to size a furnace"            # required
+description: "Quick guide to BTU sizing"  # required (used as meta description)
+date: 2026-03-15                          # required; future date = hidden until that day
+slug: how-to-size-a-furnace               # optional; if set, this is the canonical URL
+author: "Pearl Remodeling"                # optional; falls back to defaultAuthor in config
+image: /images/blog/furnace-sizing.webp   # optional; used for OG + listing card
+tags: [HVAC, Heating, Guides]             # optional
+faqs:                                      # optional; rendered as FAQ schema + section
+  - question: "What BTU do I need?"
+    answer: "Roughly 30 BTU per sq ft as a starting point..."
+---
+```
+
+## What must NOT be done
+
+- **Don't reimplement the loader.** Always import from `@ibalzam/codejitsu-core/blog` via the site's `src/lib/blog.ts`.
+- **Don't bypass `getAllPosts()` for the listing.** It's already filtering future-dated posts; bypassing means drafts leak.
+- **Don't add `getStaticPaths` that calls `getAllPosts()` alone for the detail page** — it'll exclude future-dated posts and break OG scraping for scheduled releases. Use `getAllPostSlugs()` for path generation.
+- **Don't put `.mdx` files in `content/blog/`.** The loader is `.md` only. If MDX support is needed, raise it as a feature request — it's a deliberate scope decision.
+- **Don't change the dual-slug resolution.** Old date-prefixed URLs must keep working alongside short canonical slugs.
+
+## Verify
+
+Run `modules/blog/checklist.md` after wiring. Run `checklist/core.md` (sitewide).

package/modules/blog/checklist.md

@@ -0,0 +1,36 @@
+# Blog module — checklist
+
+## Setup
+
+- [ ] `src/lib/blog.ts` exists and calls `createBlog(...)` from `@ibalzam/codejitsu-core/blog`.
+- [ ] `content/blog/` directory exists; at least one `.md` post is present.
+- [ ] Page routes exist: `src/pages/blog/index.astro`, `src/pages/blog/[...slug].astro`.
+- [ ] If site uses tags: `src/pages/blog/tag/[tag].astro` exists.
+- [ ] If site uses categories: `src/pages/blog/category/[category].astro` exists and `categories` is passed to `createBlog`.
+
+## Post frontmatter
+
+For every post in `content/blog/`:
+- [ ] `title`, `description`, `date` are present.
+- [ ] `description` is < 160 chars (meta description budget).
+- [ ] `date` is ISO `YYYY-MM-DD` format.
+- [ ] If `image` is set, the file exists in `public/`.
+- [ ] If `faqs` is set, every entry has both `question` and `answer`.
+
+## Build behaviour
+
+- [ ] Future-dated posts are absent from `dist/blog/index.html` (listing).
+- [ ] Future-dated posts ARE built at `dist/blog/<slug>/index.html` (so OG scrapers can reach them).
+- [ ] `sitemap-index.xml` does NOT contain URLs for future-dated posts.
+
+## SEO
+
+- [ ] Each blog post page has `BlogPosting` JSON-LD schema (use `@ibalzam/codejitsu-core/seo/schema`).
+- [ ] If post has `faqs`, the page also has `FAQPage` JSON-LD.
+- [ ] Post pages have OG image set to `frontmatter.image` (absolute URL).
+- [ ] Canonical URL on a post points to the canonical (frontmatter) slug, not the filename slug.
+
+## Daily deploy
+
+- [ ] `.github/workflows/daily-deploy.yml` is present (see `modules/deploy/`).
+- [ ] `CLOUDFLARE_DEPLOY_HOOK_URL` secret is configured in the repo.

package/modules/blog/src/components/index.ts

@@ -0,0 +1 @@
+export type { BlogPost, BlogPostMetadata, BlogCategory, FAQItem } from '../index.js';