README

Fully automated, backend-only AI blog/article system for Laravel.

Discover topics, write full articles, humanize them, run fact-checking and SEO/AEO/GEO optimization, generate images and FAQs, manage internal linking and duplicate detection, and control the full lifecycle through a Filament admin panel. You bring your own frontend (Blade, Livewire, Inertia, or headless).

MIT licensed. Multi-provider throughout: Groq, Gemini, OpenAI, Claude.

Table of Contents

• What This Package Does
• Feature List
• Requirements
• Installation
• Configuration
• Admin Panel Guide
• The Article Pipeline
• Day-to-Day Usage
• Artisan Commands
• Scheduled Jobs
• Publishing Modes
• Topic Discovery
• AI Providers
• Image Providers
• Frontend Integration
• Models & Querying
• Programmatic Usage
• Database Tables
• Troubleshooting
• Package Structure
• License

What This Package Does

Most AI writing tools generate text and stop. This package treats article creation as a full lifecycle:

1. Discover and score topics from RSS, News API, or manual entry
2. Generate structured articles section-by-section
3. Humanize, fact-check, and optimize for SEO, AEO, and GEO
4. Generate FAQs, images, and internal links
5. Flag duplicates and route to review (or auto-publish)
6. Monitor freshness and refresh aging content

Everything is self-hosted. Your data, your pipeline, your providers.

Feature List

Module	Description
Topic Discovery	Pluggable sources: Manual, RSS, News API — each topic scored for relevance & opportunity
Article Generation	Outline-first, section-by-section writing with configurable length, tone, and reading level
Humanization Pass	Second AI pass to reduce robotic phrasing; optional voice profile file
Fact Verification	Extracts claims, cross-checks against source material, flags unverified claims
SEO Layer	Meta title/description, slug, readability score, JSON-LD schema
AEO Layer	Featured-snippet answers, definition callouts, FAQPage schema
GEO Layer	Citation-friendly structure for AI answer engines (ChatGPT, Perplexity, etc.)
FAQ Generation	4–6 auto-generated Q&A pairs per article
Image Generation	Featured + optional inline images; Pollinations (free) default
Internal Linking	Semantic similarity finds related articles; AI inserts contextual links
Duplicate Detection	Embedding similarity check before publish
Review Workflow	Human approve / reject from admin; or self-writing auto-publish mode
Freshness Engine	Flags old or declining-traffic articles for refresh
Content Calendar	AI proposes weekly topics based on content gaps
Analytics Loop	Weights topic sources by historical article performance
Filament Admin	Dashboard, Topics, Articles, Sources, Freshness Queue, AI Settings

Requirements

• PHP 8.2+
• Laravel 11+ (tested on Laravel 13)
• Database MySQL, PostgreSQL, or SQLite
• Queue driver — database or redis recommended (QUEUE_CONNECTION=database)
• At least one AI text provider API key (Groq recommended — free tier available)
• Composer with ext-json and ext-mbstring

Installation

1. Install the package

composer require muzorix/ai-blog-engine

For local path development (monorepo):

// composer.json
"repositories": [
    {
        "type": "path",
        "url": "packages/ai-blog-engine",
        "options": { "symlink": true }
    }
],
"require": {
    "muzorix/ai-blog-engine": "@dev"
}

2. Run the install command

php artisan ai-blog-engine:install --migrate

This will:

• Publish config/ai-blog-engine.php
• Publish and run database migrations
• Seed default topic sources (Manual, RSS, News API)
• Create the public/storage symlink (for images)
• Publish Filament CSS/JS assets

3. Publish Filament assets (if admin looks unstyled)

If the admin login page has no styling, run:

php artisan filament:assets

Important: Set APP_URL in .env to match how you access the site (http://127.0.0.1:8000 vs http://localhost:8000).

4. Configure environment variables

Copy the block below into .env and fill in your keys:

# Core
ABE_MODE=review
ABE_TEXT_PROVIDER=groq
ABE_IMAGE_PROVIDER=pollinations

# Site identity (used in prompts & schema.org)
ABE_SITE_NAME=Muzorix
ABE_SITE_URL=https://muzorix.com
ABE_SITE_NICHE="tech news, software, AI, gadgets, and technology industry"
ABE_SITE_DESCRIPTION="Tech news and insights covering software, AI, gadgets, and industry trends"
ABE_AUTHOR_NAME=Muzorix
ABE_ORGANIZATION_NAME=Muzorix

# Generation defaults
ABE_DEFAULT_LENGTH=1200
ABE_DEFAULT_TONE=professional
ABE_INLINE_IMAGES=false

# Publishing
ABE_REQUIRE_APPROVAL=true
ABE_ADMIN_PATH=abe-admin
ABE_BLOG_PATH=blog

# Topic sources
ABE_RSS_ENABLED=true
ABE_RSS_FEEDS=https://feeds.example.com/rss,https://another.com/feed
ABE_NEWS_API_ENABLED=true
ABE_NEWS_QUERY="technology OR software OR artificial intelligence"

# API keys
GROQ_API_KEY=your-groq-key
NEWS_API_KEY=your-newsapi-key

5. Enable Filament panel access on your User model

Your App\Models\User must implement Filament\Models\Contracts\FilamentUser:

use Filament\Models\Contracts\FilamentUser;
use Filament\Panel;

class User extends Authenticatable implements FilamentUser
{
    public function canAccessPanel(Panel $panel): bool
    {
        return true; // restrict as needed
    }
}

6. Create an admin user

php artisan make:filament-user --panel=ai-blog-engine

7. Start the queue worker

Article generation runs as queued jobs. Start a worker:

php artisan queue:work

Or use the Laravel composer dev script if your project has it.

8. Start the scheduler (production)

Add to your server crontab:

* * * * * cd /path-to-your-project && php artisan schedule:run >> /dev/null 2>&1

The package registers these scheduled jobs automatically:

Job	Default schedule	Purpose
FetchTopicsJob	Every 6 hours	Pull topics from enabled sources
RunFreshnessScanJob	Mondays 3 AM	Flag articles needing updates
RunContentCalendarJob	Sundays 4 AM	Propose new topics for the week
PullAnalyticsJob	Daily 5 AM	Update source weights & queue topics

9. Open the admin panel

http://your-site.test/abe-admin

(Path is configurable via ABE_ADMIN_PATH.)

Configuration

Publish or re-publish config anytime:

php artisan vendor:publish --tag=ai-blog-engine-config --force

Key config sections

Section	Purpose
mode	review or self_writing
site	Niche, description, author, organization — fed into AI prompts and schema
text_provider	Default provider + per-step overrides (generate, humanize, fact_check, etc.)
image_provider	Default image driver
humanization	Enable/disable pass; optional voice_profile_path text file
fact_check	Enable/disable; block_unverified_claims for hard blocking
seo	Min word count, readability target, meta length limits
aeo	FAQ count, snippet optimization
geo	Citation-friendly restructuring
images	Disk, directory, inline_per_h2 toggle
duplicate_check	Similarity threshold (0.85 default), compare last N articles
internal_linking	Max links per article, related article count
freshness	Check interval days, traffic decline threshold
topic_discovery	Min scores, RSS feeds, News API query
generation	Default length, tone, reading level
frontend	blog_path for canonical URLs in schema
schedule	Cron expressions for automated jobs

Most settings can also be changed from AI Settings in the admin panel without editing files.

Admin Panel Guide

The Filament panel is at /abe-admin (panel ID: ai-blog-engine).

Dashboard

• Pipeline stats: queued topics, pending review, published count, freshness queue
• Recent articles table

Topics

Manage the topic pipeline:

Action	Description
Create	Manually add a topic (auto-queued for generation)
Fetch Topics	Pull from RSS / News API and score with AI
Score	Re-run AI relevance/opportunity scoring on a topic
Generate	Dispatch full article pipeline for a topic

Topic statuses: pending → scored → queued → processing → completed

Articles

Full CRUD for generated articles:

Action	Description
Edit	Modify title, body, meta tags, status
Approve	Publish article (sets published_at)
Reject	Mark as rejected

Review panel shows SEO score, readability, AEO score, duplicate flag, and fact-check flags.

Sources

Enable/disable and configure topic sources:

• Manual — topics entered by hand
• RSS — comma-separated feed URLs in config or config key-value in admin
• News API — requires NEWS_API_KEY and search query

Adjust performance_weight to boost sources that produce high-performing articles.

Freshness Queue

Articles flagged as needs_update appear here:

• Run Freshness Scan — scan all published articles
• Refresh — regenerate an updated draft through the full pipeline

AI Settings

Configure without touching .env:

• Site niche, description, author, URL
• Mode (review vs self-writing)
• Text and image providers
• Tone and target word count
• Humanization, fact-check, and inline image toggles

The Article Pipeline

When a topic is generated, it passes through these steps in order:

Topic
  ↓
1. ArticleGenerator     — outline + section-by-section HTML body
  ↓
2. HumanizerPass        — reduce AI tells (optional)
  ↓
3. FactVerifier         — extract & verify claims against source
  ↓
4. SeoProcessor         — meta tags, slug, schema.org JSON-LD
  ↓
5. AeoProcessor         — snippet answer + definition callouts
  ↓
6. GeoProcessor         — citation-friendly restructuring
  ↓
7. FaqGenerator         — 4–6 FAQ pairs + FAQPage schema
  ↓
8. ImageGenerator       — featured image (+ inline per H2 if enabled)
  ↓
9. InternalLinker       — semantic links to related published articles
  ↓
10. DuplicateChecker    — flag near-duplicates by embedding similarity
  ↓
11. Finalize            — pending_review OR auto-publish

Each step logs to abe_ai_generation_logs with provider, tokens, and cost estimate.

Day-to-Day Usage

Workflow A: Fully automated (with review)

# 1. Fetch topics from RSS / News API
php artisan ai-blog-engine:fetch-topics

# 2. Generate articles for all queued topics
php artisan ai-blog-engine:generate --queued

# 3. Review in admin → Articles → Approve or Reject

Workflow B: Manual topic → article

1. Go to Topics → Create
2. Enter title and summary (paste source content in summary for fact-checking)
3. Set status to queued
4. Click Generate on the topic row (or run php artisan ai-blog-engine:generate {id})
5. Review and approve in Articles

Workflow C: Self-writing (no human step)

ABE_MODE=self_writing
ABE_REQUIRE_APPROVAL=false

Articles publish automatically when fact-check and duplicate checks pass.

Workflow D: Refresh aging content

php artisan ai-blog-engine:freshness-scan

Then use Freshness Queue in admin to refresh flagged articles.

Artisan Commands

Command	Description
php artisan ai-blog-engine:install	Publish config, migrations, seed sources, Filament assets
php artisan ai-blog-engine:install --migrate	Same + run migrations
php artisan ai-blog-engine:fetch-topics	Fetch and AI-score topics from enabled sources
php artisan ai-blog-engine:generate {topic?}	Generate article for one topic ID
php artisan ai-blog-engine:generate --queued	Generate all queued topics
php artisan ai-blog-engine:freshness-scan	Scan published articles for freshness flags
php artisan make:filament-user --panel=ai-blog-engine	Create admin login
php artisan filament:assets	Re-publish Filament CSS/JS (fix unstyled admin)

Scheduled Jobs

Registered automatically when APP_ENV runs in console. Customize cron in config/ai-blog-engine.php under schedule:

'schedule' => [
    'fetch_topics'     => '0 */6 * * *',  // every 6 hours
    'freshness_scan'   => '0 3 * * 1',    // Mondays 3 AM
    'content_calendar' => '0 4 * * 0',    // Sundays 4 AM
    'analytics_pull'   => '0 5 * * *',    // daily 5 AM
],

Override via env: ABE_SCHEDULE_FETCH_TOPICS, ABE_SCHEDULE_FRESHNESS, etc.

Publishing Modes

Mode	Config	Behavior
Review	ABE_MODE=review	Articles land in pending_review; human approves in admin
Self-writing	ABE_MODE=self_writing	Auto-publish if no blocking fact-check flags and not duplicate-flagged

Article statuses: draft → humanizing → fact_checking → seo_processing → pending_review → published

Also: scheduled, needs_update, rejected

Topic Discovery

Manual

Create topics in admin. Best practice: paste source article text in summary or store full content in raw_payload so fact-checking can verify claims.

RSS

ABE_RSS_ENABLED=true
ABE_RSS_FEEDS=https://feeds.arstechnica.com/arstechnica/index,https://www.theverge.com/rss/index.xml

Or configure feeds per-source in Sources → RSS → config key feeds as JSON array.

News API

ABE_NEWS_API_ENABLED=true
NEWS_API_KEY=your-key
ABE_NEWS_QUERY="technology OR artificial intelligence"

Get a free key at newsapi.org.

AI Scoring

Each topic receives:

• Relevance score (0–100) — fit to your niche
• Opportunity score (0–100) — timeliness and content gap potential

Topics below ABE_MIN_RELEVANCE (50) or ABE_MIN_OPPORTUNITY (30) are not queued.

AI Providers

Text providers

Driver	Env key	Notes
groq	GROQ_API_KEY or GROQ_API_KEYS	Free tier, fast — supports multiple keys with auto-rotation on 429
openai	OPENAI_API_KEY	GPT-4o mini default
claude	ANTHROPIC_API_KEY	Claude Sonnet default
gemini	GEMINI_API_KEY	Gemini 2.0 Flash default

Use one provider for all steps:

ABE_TEXT_PROVIDER=groq
GROQ_API_KEY=gsk_your_first_key

Multiple Groq keys (recommended for production / heavy usage):

When one key hits its rate limit, the package automatically switches to the next key — no job failure, no manual intervention.

GROQ_API_KEYS=gsk_key_one,gsk_key_two,gsk_key_three

You can still set GROQ_API_KEY for a single key; GROQ_API_KEYS takes precedence when set (comma-separated, no spaces required).

Keys are rotated round-robin for even load. On HTTP 429, the current key is skipped and the next is tried immediately.

Or override per pipeline step:

ABE_TEXT_PROVIDER_GENERATE=claude
ABE_TEXT_PROVIDER_HUMANIZE=groq
ABE_TEXT_PROVIDER_FACT_CHECK=claude

Programmatic usage

use Muzorix\AiBlogEngine\AiText\AiTextManager;

$ai = app(AiTextManager::class);
$response = $ai->prompt('generate', 'Write a headline about Laravel 13');
$json = $ai->json('seo', 'Return JSON with meta_title and meta_description for...');

Image Providers

Driver	Env key	Cost
pollinations	none	Free
unsplash	UNSPLASH_ACCESS_KEY	Free (stock photos)
openai	OPENAI_API_KEY	Paid (DALL-E 3)
stability	STABILITY_API_KEY	Paid

ABE_IMAGE_PROVIDER=pollinations
ABE_INLINE_IMAGES=false   # true = generate image per H2 section

Images are stored on the disk configured in ABE_IMAGE_DISK (default: public) under abe-images/.

Frontend Integration

This package is backend-only. Query models from your own routes, controllers, or Livewire components.

Blade example

// routes/web.php
use Muzorix\AiBlogEngine\Models\Article;

Route::get('/blog', fn () => view('blog.index', [
    'articles' => Article::published()
        ->with(['faqs', 'images'])
        ->latest('published_at')
        ->paginate(12),
]));

Route::get('/blog/{slug}', function (string $slug) {
    $article = Article::published()
        ->where('slug', $slug)
        ->with(['faqs', 'images', 'outboundLinks.linkedArticle'])
        ->firstOrFail();

    return view('blog.show', compact('article'));
});

{{-- resources/views/blog/show.blade.php --}}
<head>
    <title>{{ $article->meta_title ?? $article->title }}</title>
    @if ($article->meta_description)
        <meta name="description" content="{{ $article->meta_description }}">
    @endif
    @if ($article->schema_markup)
        <script type="application/ld+json">{!! $article->schema_json !!}</script>
    @endif
</head>

<h1>{{ $article->title }}</h1>
{!! $article->body !!}

@foreach ($article->faqs as $faq)
    <details>
        <summary>{{ $faq->question }}</summary>
        <p>{{ $faq->answer }}</p>
    </details>
@endforeach

Livewire

use Livewire\Component;
use Livewire\WithPagination;
use Muzorix\AiBlogEngine\Models\Article;

class BlogIndex extends Component
{
    use WithPagination;

    public function render()
    {
        return view('livewire.blog-index', [
            'articles' => Article::published()
                ->latest('published_at')
                ->paginate(12),
        ]);
    }
}

Canonical URLs

Articles expose a url attribute using ABE_SITE_URL + ABE_BLOG_PATH:

https://muzorix.com/blog/your-article-slug

Related articles

use Muzorix\AiBlogEngine\Pipeline\InternalLinker;

$related = app(InternalLinker::class)->findSimilarArticles($article);

Models & Querying

Model	Table	Purpose
Topic	abe_topics	Discovered/scored topics
TopicSource	abe_topic_sources	Source configuration
Article	abe_articles	Generated articles
ArticleFaq	abe_article_faqs	FAQ pairs
ArticleImage	abe_article_images	Featured and inline images
ArticleLink	abe_article_links	Internal links between articles
FactCheckFlag	abe_fact_check_flags	Claim verification results
AiGenerationLog	abe_ai_generation_logs	Per-step AI usage logs
ArticleEdit	abe_article_edits	Human/AI edit audit trail
ArticleAnalytic	abe_article_analytics	Pageview data for freshness/analytics

Useful scopes

Article::published()->latest('published_at')->get();
Article::pendingReview()->get();
Article::needsUpdate()->get();
Topic::readyForGeneration()->get();
Topic::scheduled()->get();

Publish manually

$article->publish(); // sets status, published_at, last_updated_at

Programmatic Usage

Run the full pipeline

use Muzorix\AiBlogEngine\Facades\AiBlogEngine;
use Muzorix\AiBlogEngine\Models\Topic;

$topic = Topic::find(1);
$article = AiBlogEngine::run($topic);

Generate with custom options

use Muzorix\AiBlogEngine\Pipeline\ArticleGenerator;

$article = ArticleGenerator::make()
    ->provider('groq')
    ->targetLength(1500)
    ->tone('professional')
    ->readingLevel('grade_8')
    ->generate($topic);

Dispatch as background job

use Muzorix\AiBlogEngine\Jobs\GenerateArticleJob;

GenerateArticleJob::dispatch($topic);

Database Tables

abe_topic_sources     — source drivers and config
abe_topics            — discovered topics with scores
abe_articles          — article content, SEO, schema, embeddings
abe_article_faqs      — FAQ Q&A pairs
abe_article_images    — featured and inline images
abe_article_links     — internal link graph
abe_fact_check_flags  — claim verification results
abe_ai_generation_logs — AI step audit log
abe_article_edits     — edit diff history
abe_article_analytics — traffic data for freshness loop

Troubleshooting

Admin panel has no CSS / looks like plain HTML

php artisan filament:assets

Hard-refresh the browser (Ctrl+Shift+R).

Admin panel CSS loads but from wrong host

Set APP_URL in .env to exactly match your browser URL:

APP_URL=http://127.0.0.1:8000

Articles stuck in processing

Ensure a queue worker is running:

php artisan queue:work

Check storage/logs/laravel.log and the abe_ai_generation_logs table.

"API key not configured" errors

Verify the correct env key for your provider (GROQ_API_KEY, etc.) and run:

php artisan config:clear

Images not displaying

php artisan storage:link

Confirm ABE_IMAGE_DISK=public and files exist in storage/app/public/abe-images/.

No topics after fetch

• Check RSS feed URLs are valid and reachable
• Verify NEWS_API_KEY if using News API
• Lower ABE_MIN_RELEVANCE / ABE_MIN_OPPORTUNITY in config
• Topics with duplicate titles from the last 30 days are skipped

Fact-check flags everything as unverified

Paste the original source content into the topic summary or ensure raw_payload.content is populated when the topic is created.

Package Structure

packages/ai-blog-engine/
├── config/ai-blog-engine.php
├── database/migrations/
├── src/
│   ├── AiBlogEngineServiceProvider.php
│   ├── AiText/Drivers/          Groq, OpenAI, Claude, Gemini
│   ├── AiImage/Drivers/         Pollinations, Unsplash, OpenAI, Stability
│   ├── TopicSources/Drivers/    Manual, RSS, News API
│   ├── Pipeline/                All processing steps + ArticlePipeline
│   ├── Jobs/                    Fetch, Generate, Refresh, Calendar, Analytics
│   ├── Commands/                Install, fetch, generate, freshness-scan
│   ├── Models/
│   ├── Filament/                Admin panel, resources, widgets
│   └── Support/                 Slug, embedding, readability helpers
└── README.md

License

MIT — free for personal and commercial use.