AI Article GeneratorContent Automation Engine
Automated AI content engine that fetches news from authoritative sources, categorizes with OpenAI GPT-4o, and publishes MDX articles to your site.
Open source. GitHub Actions ready. Fully configurable. Set it up once and your blog publishes itself.
The Problem
You run an AI compliance or governance blog. You need fresh, accurate, well-written content — but the standard options are:
- •Manual writing — slow, expensive, and inconsistent in quality
- •Hiring freelance writers — $200-500 per article, doesn't scale for daily publishing
- •Generic AI tools — produce content without real sources, risk hallucination, and lack domain-specific categorization
Most AI compliance blogs either publish infrequently or publish low-quality content. Neither builds authority.
What It Does
A complete content pipeline: fetch, categorize, generate, publish. Set it up once and your blog publishes itself on a schedule.
Fetch & Aggregate
AI Categorize & Score
Generate MDX Articles
Auto-Publish to Your Repo
How It Works
Step 1: Configure sites.json
Point to your GitHub repo, set your site name, author, logo, internal links, and content path.
Step 2: Fetch from authoritative sources
FrontOfAI Briefing API (39+ AI news sources) + 14 RSS feeds (FTC, EEOC, Google News alerts).
Step 3: OpenAI generates articles
GPT-4o categorizes, scores, and expands each story into a full MDX article with frontmatter.
Step 4: Push to your repo
Engine clones your repo, copies MDX files, commits, and pushes. Vercel/Netlify auto-deploys.
Content Types
News Articles
Enforcement Actions
Weekly Roundups
Monthly Trends
Configuration
Everything is driven by config/sites.json. No code changes needed to add your site.
{
"sites": [
{
"name": "my-blog",
"repo": "your-username/your-blog-repo",
"branch": "main",
"contentPath": "src/content",
"baseUrl": "https://yoursite.com",
"siteName": "Your Site Name",
"author": "Your Name",
"authorRole": "AI Analyst",
"logoPath": "/images/logo.png",
"internalLinks": ["/blog", "/guides", "/tools"],
"categories": ["news", "enforcement", "blog", "trends"],
"minImpactScore": 6,
"enabled": true
}
]
}Automated Schedule
Four GitHub Actions workflows run on a schedule. All can also be triggered manually.
| Workflow | Schedule | Description |
|---|---|---|
| news-tracker.yml | Every 4 hours | Fetch + categorize + generate news/enforcement articles |
| weekly-content.yml | Every Monday 9:00 UTC | Generate weekly roundup blog post |
| monthly-trends.yml | 1st of month 10:00 UTC | Generate monthly trend analysis |
| test-dry-run.yml | Manual only | Dry run — generates 2 articles without pushing |
Content Integrity Rules
What It Does
- • Uses ONLY source material from FrontOfAI API and RSS feeds
- • Labels all content with autoGenerated: true in frontmatter
- • Cites primary sources with links
- • Includes configurable internal links
- • Never overwrites existing articles
What It Never Does
- • Fabricate facts, quotes, or statistics
- • Invent case studies or hypothetical examples as real
- • Make claims about what companies plan to do
- • Use promotional or marketing language
- • Publish without source attribution
Limitations
This is a content pipeline, not a magic button. It automates the fetch-categorize-generate-publish workflow, but quality depends on your configuration and source material.
This is a developer tool for people who want to automate content publishing on their own MDX-based site. If you need WordPress, Ghost, or a managed solution, contact us for custom integration.
How It Compares
vs. Manual Writing / Freelance Writers
This engine wins: Runs 24/7, $0.01-0.05 per article vs $200-500 per freelance article. Consistent format and frontmatter. Never misses a deadline.
Manual wins: Human writers produce original analysis, interviews, and opinion pieces. This engine only expands existing source material — it doesn't create original reporting.
vs. Jasper / Copy.ai / ChatGPT
This engine wins: Fetches real sources automatically, categorizes with structured metadata, generates with source attribution, and auto-publishes to your repo. No manual prompting.
Generic AI wins: More flexible for any topic, no setup required, can generate social posts, emails, and ad copy. This engine is purpose-built for AI compliance news on MDX sites.
vs. WordPress Autoblog Plugins
This engine wins: Open source, no subscription, full control over prompts and categorization, generates original articles (not RSS republishing), structured frontmatter for SEO.
Plugins win: One-click install, works with WordPress, no code or GitHub required. This engine requires a GitHub-hosted MDX site.
FAQ
What do I need to set this up?
You need an OpenAI API key, a GitHub personal access token with repo scope, and a target repository that accepts MDX files (Next.js, Astro, or any MDX-based site). Edit config/sites.json to point to your repo, set two GitHub secrets, and the GitHub Actions workflows handle the rest.
Does it work with my existing site?
If your site is hosted on GitHub and uses MDX content files, it works. The contentPath in sites.json tells the engine where to put files in your repo. Common setups include Next.js with src/content, Astro with src/content, or any static site generator that reads MDX.
Can I customize the generated content?
Yes. Everything is config-driven. Set your site name, author, logo, internal links, sponsor link, and category mapping in sites.json. Prompts use placeholder tokens that get replaced with your values. You can also create a custom prompts directory for full control over tone and focus.
How much does it cost to run?
The repo is free and open source (MIT). Your only costs are OpenAI API usage (typically $0.01-0.05 per article with GPT-4o) and GitHub Actions minutes (free for public repos, 2000 min/month free for private repos). No subscription, no platform fee.
Does it fabricate content?
No. The engine uses ONLY source material from FrontOfAI API and RSS feeds. All articles are labeled with autoGenerated: true in frontmatter. The prompts explicitly prohibit fabricating facts, quotes, statistics, or case studies. If source material is thin, it writes a shorter article rather than padding with speculation.
Can I get help with custom setup or integration?
Yes. The tool is designed to be self-service, but if you need custom integration with your CMS, custom prompts tuned for your industry, additional RSS sources, or multi-site deployment at scale, contact Subodh KC at subodhkc.com for consulting and custom setup services.