Blog Publish Automation

Automation Levels

Level	Description	Status
Level 1: Manual	Writer → commit MDX → push → Vercel deploys	Current default
Level 2: Semi-automated	AI drafts (mos-seo-content) → human reviews → manual commit	Current for AI-written posts
Level 3: Git-triggered	Boxi CRM → GitHub API → auto-commit → Vercel deploys	Roadmap: Sprint 1
Level 4: Fully automated	CRM scheduled → GitHub API → deploy → Postiz social queue	Roadmap: Sprint 2

Agencies using advanced automation report 29% higher revenue impact (Averi AI, 2026). Building Level 3+4 puts Boxi in the top 1–2% of agencies by automation capability.

Level 1 — Current (Manual)

mos-seo-content → MDX file written + reviewed
    │
    └── Manual: commit to client repo → push → Vercel deploys (~60 sec)
                                                │
                                                └── Manual: create social posts in Postiz

Track each delivery as a Task in Boxi CRM.

Level 3 — GitHub API Publisher (to build)

Boxi CRM creates the MDX file directly in the client’s GitHub repo via the GitHub API. Vercel picks it up and deploys automatically.

// In Boxi CRM — publish blog post to client's Astro site
await fetch(
  `https://api.github.com/repos/Traderfuz/${clientRepo}/contents/src/content/blog/${slug}.mdx`,
  {
    method: 'PUT',
    headers: {
      'Authorization': `Bearer ${GITHUB_TOKEN}`,
      'Content-Type': 'application/json',
    },
    body: JSON.stringify({
      message: `Add blog post: ${title}`,
      content: btoa(mdxContent),  // base64 encoded
      branch: 'main',
    }),
  }
)

Store in Boxi provider config: GITHUB_TOKEN per client in lib/settings/provider-config-loader.ts (encryption infrastructure already built).

Level 4 — Full Pipeline (to build)

1. mos-seo-content → MDX written + reviewed
       │
       ↓
2. Stored in Boxi CRM (draft, with publishAt timestamp)
       │
       ↓
3. Boxi cron fires at publishAt:
       ├── GitHub API → commit MDX → Vercel deploys
       └── Postiz API → schedule social posts: publish day + day 3 + day 7

// Schedule a social post via Postiz API
await fetch('https://postiz.yourdomain.com/api/posts', {
  method: 'POST',
  headers: { 'Authorization': `Bearer ${POSTIZ_API_KEY}` },
  body: JSON.stringify({
    integration: clientIntegrationId,   // client's connected social account
    content: 'Check out our latest guide...',
    date: publishDate.toISOString(),
    settings: { link: `https://clientsite.com/blog/${slug}` },
  }),
})

Store in Boxi provider config: POSTIZ_API_KEY + per-client integration IDs.

Content-Only Clients (Existing Sites)

WordPress (WP REST API)

await fetch(`https://clientsite.com/wp-json/wp/v2/posts`, {
  method: 'POST',
  headers: {
    'Authorization': `Basic ${btoa(`${wpUser}:${wpAppPassword}`)}`,
    'Content-Type': 'application/json',
  },
  body: JSON.stringify({ title, content, status: 'publish', slug }),
})

Webflow (CMS API)

Same pattern — POST /collections/{collectionId}/items with Webflow API key.

No API / locked CMS

Write in Google Docs → track delivery as a Boxi Task → client pastes into CMS.

Implementation Priority

Now: Manual publish + Postiz manual scheduling. Track in Boxi Tasks.
At >10 posts/week across clients: Build GitHub API publisher (Level 3). ~1–2 day sprint.
Social automation first win: Boxi Workflow → Postiz API. One workflow template that auto-queues 3 social posts on every blog publish. Highest client-visible value.