Open Source Content Engine

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.

MIT Licensed
GPT-4o Powered
GitHub Actions Ready
Config-Driven

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

Pulls AI compliance news from the FrontOfAI Briefing API and 14+ RSS feeds — FTC, EEOC, Google News alerts for EU AI Act, LL144, AI hiring bias, prompt injection, and more.

AI Categorize & Score

OpenAI GPT-4o classifies each story as news vs. enforcement, assigns impact scores (1-10), tags, and structured metadata — entity, violation, framework, penalty, region.

Generate MDX Articles

Produces 800-1200 word news articles, enforcement action summaries, weekly roundups, and monthly trend analyses — all with frontmatter, source attribution, and internal links.

Auto-Publish to Your Repo

Clones your target repo, copies MDX files, commits, and pushes. Works with any GitHub-hosted Next.js, Astro, or MDX-based site. Never overwrites existing articles.

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

Every 4 hours

News Articles

800-1200 words. Generated from FrontOfAI API and RSS feeds. Categorized with impact score, tags, and source attribution.
Every 4 hours

Enforcement Actions

Structured frontmatter: entity, violation, framework, penalty amount, region, status. Generated when stories are classified as enforcement.
Every Monday

Weekly Roundups

1000-1500 words. Summarizes top 15 stories from the past week, grouped by theme — regulatory, enforcement, industry.
1st of each month

Monthly Trends

1500-2500 words. Analyzes patterns across 30+ stories from the past month. Identifies 3-5 key trends with evidence.

Configuration

Everything is driven by config/sites.json. No code changes needed to add your site.

config/sites.json
{
  "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.

WorkflowScheduleDescription
news-tracker.ymlEvery 4 hoursFetch + categorize + generate news/enforcement articles
weekly-content.ymlEvery Monday 9:00 UTCGenerate weekly roundup blog post
monthly-trends.yml1st of month 10:00 UTCGenerate monthly trend analysis
test-dry-run.ymlManual onlyDry 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.

OpenAIRequires an OpenAI API key. Cost is ~$0.01-0.05 per article with GPT-4o. Not free to run at scale.
TopicsOut of the box, it fetches AI compliance and governance news. Adding other topics requires configuring new RSS feeds in sources.json.
ReviewAll articles are labeled autoGenerated: true. You should review before publishing — especially for enforcement actions with legal implications.
GitHubYour target site must be hosted on GitHub and accept MDX files. Not compatible with WordPress, Medium, or hosted CMS platforms.

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.

Get Started

MIT licensed. Self-service setup in 15 minutes. Or get custom integration done for you.

Let's Talk →