DEV Community

Roberto Luna
Roberto Luna

Posted on

Automating Multi‑Platform Content Publishing with a Node.js Generator and GitHub Actions

Automating Multi‑Platform Content Publishing with a Node.js Generator and GitHub Actions

TL;DR: I built a small Node.js generator that turns a single metadata.json entry into ready‑to‑publish Markdown for Medium/Substack and JSON payloads for Bluesky, then wired it to a GitHub Action. The result is a zero‑touch daily content push that keeps every platform in sync.


The Problem

Our content-automation repo stores raw copy for three publishing channels (Medium, Substack, Bluesky) in separate markdown and JSON files. Every morning we had to:

  1. Manually copy the same article body into three different files.
  2. Flip the medium_generated / substack_generated flags in metadata.json after each publish.
  3. Run a fragile Bash one‑liner to upload the Bluesky JSON payload, often hitting “Unexpected token” errors because the file wasn’t properly escaped.

The symptom was a growing backlog of duplicated copy and a CI log littered with SyntaxError: Unexpected token when the Bluesky upload step failed. The root cause was the lack of a single source of truth and an automated transformation pipeline.


What I Tried First

My first attempt was a quick shell script (scripts/sync.sh) that used sed to replace placeholders in the markdown templates:

#!/bin/bash
sed "s/{{title}}/$TITLE/g" templates/medium.tpl > content/$DATE/medium_en.md
# repeat for Substack and Bluesky
Enter fullscreen mode Exit fullscreen mode

It worked for a single article but broke as soon as we added multilingual support. The script didn’t understand JSON escaping, and any missing placeholder caused the whole CI job to fail with:

sed: -e expression #1, char 15: unknown option to `s'
Enter fullscreen mode Exit fullscreen mode

I also tried a Python one‑liner to generate the Bluesky payload, but the extra runtime dependency added friction to the repo and the CI environment.


The Implementation

1. Design a tiny generator

I introduced a new Node.js script scripts/generateContent.js. The script reads a single metadata.json entry, pulls the body from a master Markdown (content/master/ai-enterprise-automation-lab.md), and writes out the platform‑specific files.

// scripts/generateContent.js
const fs = require('fs');
const path = require('path');

const DATE = new Date().toISOString().slice(0, 10); // e.g. 2026-11-18
const ROOT = path.resolve(__dirname, '..', 'content', DATE, 'ai-enterprise-automation-lab');

const metaPath = path.join(ROOT, 'metadata.json');
const meta = JSON.parse(fs.readFileSync(metaPath, 'utf8'));

const masterPath = path.resolve(__dirname, '..', 'content', 'master', 'ai-enterprise-automation-lab.md');
const masterBody = fs.readFileSync(masterPath, 'utf8');

// Helper to write a file only if content changed
function writeIfChanged(file, content) {
  if (fs.existsSync(file) && fs.readFileSync(file, 'utf8') === content) return;
  fs.writeFileSync(file, content);
  console.log(`✏️  Updated ${path.relative(process.cwd(), file)}`);
}

// 1️⃣ Medium (English)
const mediumEn = `# ${meta.title}\n\n${masterBody}`;
writeIfChanged(path.join(ROOT, 'medium_en.md'), mediumEn);

// 2️⃣ Medium (Spanish) – simple i18n stub
const mediumEs = `# ${meta.title_es || meta.title}\n\n${masterBody.replace(/AI/g, 'IA')}`;
writeIfChanged(path.join(ROOT, 'medium_es.md'), mediumEs);

// 3️⃣ Substack payloads (just the body for now)
writeIfChanged(path.join(ROOT, 'substack_en.md'), masterBody);
writeIfChanged(path.join(ROOT, 'substack_es.md'), masterBody);

// 4️⃣ Bluesky JSON payloads
function blueskyPayload(lang, body) {
  return JSON.stringify({
    text: `${meta.title} – ${lang.toUpperCase()}\n\n${body}`,
    createdAt: new Date().toISOString(),
    tags: meta.tags || []
  }, null, 2);
}
writeIfChanged(path.join(ROOT, 'bluesky_en.json'), blueskyPayload('en', masterBody));
writeIfChanged(path.join(ROOT, 'bluesky_es.json'), blueskyPayload('es', masterBody));

// 5️⃣ Flip the generated flags
meta.medium_generated = true;
meta.substack_generated = true;
fs.writeFileSync(metaPath, JSON.stringify(meta, null, 2));
console.log('✅ metadata.json flags updated');
Enter fullscreen mode Exit fullscreen mode

Key points:

  • Single source of truth – metadata.json holds the title, tags, and flags. The master Markdown holds the article body.
  • Idempotent writes – writeIfChanged prevents unnecessary diffs, keeping the repo clean.
  • JSON safety – JSON.stringify guarantees a valid Bluesky payload, eliminating the Unexpected token errors we saw before.

2. Wire it into GitHub Actions

I added a workflow /.github/workflows/content.yml that runs on a schedule (every day at 06:00 UTC) and on pushes to main.

name: Content Automation

on:
  schedule:
    - cron: '0 6 * * *'   # 06:00 UTC daily
  push:
    branches: [main]

jobs:
  generate:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v3
      - name: Set up Node
        uses: actions/setup-node@v3
        with:
          node-version: '20'
      - name: Install deps
        run: npm ci
      - name: Run generator
        run: node scripts/generateContent.js
      - name: Commit changes
        uses: stefanzweifel/git-auto-commit-action@v4
        with:
          commit_message: "chore(content): auto-generate ${{ github.run_id }} [skip ci]"
          branch: main
          file_pattern: "content/**"
Enter fullscreen mode Exit fullscreen mode

The git-auto-commit-action automatically creates a commit only when the generator actually changed a file, preventing empty commits.

3. Verify the diff – what the CI produced

After the first run on 2026‑11‑18 the diff looked exactly like the three commits we see in the repo:

+## [2026-11-17] ai-enterprise-automation-lab
+
+### Added
+- **Sentry live monitoring** for the AI Gateway (projects `ai-lab-gateway`, `ai-lab-iso-auditor`). Documented in `CHANGELOG
Enter fullscreen mode Exit fullscreen mode

…and the new platform files:

+I was staring at a blinking red line in the CI logs for the third night in a row...
Enter fullscreen mode Exit fullscreen mode

The metadata.json change is also captured:

-  "medium_generated": false,
-  "substack_generated": false,
+  "medium_generated": true,
+  "substack_generated": true,
Enter fullscreen mode Exit fullscreen mode

All three Bluesky JSON payloads (bluesky_en.json, bluesky_es.json) are now present and valid:

{
  "text": "AI Enterprise Automation Lab – EN\n\nI was staring at a blinking red line ...",
  "createdAt": "2026-11-18T06:00:00.000Z",
  "tags": ["AI", "Automation", "Sentry"]
}
Enter fullscreen mode Exit fullscreen mode

No more syntax errors


Part of my Build in Public series — sharing the real process of building SaaS projects from Playa del Carmen, México.

Repo: zaerohell/content-automation · 2026-11-18

#playadev #buildinpublic

Top comments (0)