A sophisticated content generation system built with PocketFlow that creates platform-optimized, brand-consistent content while eliminating AI fingerprints. This system implements a complete workflow for generating authentic marketing content across multiple platforms.

• Multi-Platform Content Generation: Creates optimized content for email, LinkedIn, Instagram, Twitter, Reddit, and blog posts
• AI Fingerprint Elimination: Detects and removes all 7 "Deadly Sins" of AI-generated content
• Brand Voice Consistency: Maintains consistent brand personality across all platforms
• Authenticity Scoring: Measures content authenticity with comprehensive metrics
• Iterative Refinement: Interactive feedback loop for content improvement
• Version Control: Tracks content versions with rollback capability
• Performance Analytics: Provides insights into pattern reduction and efficiency gains
• Preset Learning: Automatically optimizes configurations based on usage patterns

The system implements a sophisticated workflow with 17 specialized nodes:

1. Engagement Manager - Handles user inputs and preset management
2. Brand Bible Processor - Converts descriptive brand text into structured parameters
3. Format Guidelines - Generates platform-specific formatting rules
4. Creative Inspiration - Creates high-engagement content examples
5. Pattern-Aware Prompt Engineering - Enhances prompts to prevent AI patterns
6. Content Generator - Creates initial platform-optimized drafts
7. Style Optimizer - Refines content while maintaining brand voice
8. Deadly Sins Scanner - Detects AI fingerprint patterns
9. Style Compliance - Validates content authenticity
10. Authenticity Scorer - Evaluates human-like qualities
11. Feedback Router - Manages user feedback and version control
12. Content Refinement - Applies user-requested changes
13. Pattern Learner - Improves future prompt engineering
14. Version Manager - Tracks content versions
15. Final Formatter - Outputs clean markdown with copy functionality
16. Preset Optimizer - Learns from user preferences
17. Version Analytics - Provides performance insights

• Python 3.8+
• OpenAI API key or compatible LLM service
• Anthropic API key (optional)

pip install -r requirements.txt

python main.py

Choose between:

1. Content Generation Mode: Full PR content workflow with brand voice and platform optimization
2. QA Mode: Simple question-answering functionality

├── main.py                 # Entry point
├── flow.py                # Workflow definitions
├── nodes/                 # Individual workflow nodes
│   ├── engagement_manager.py
│   ├── brand_bible_processor.py
│   ├── format_guidelines.py
│   └── ... (14 more nodes)
├── utils/                 # Utility functions
│   ├── call_llm.py
│   ├── brand_bible_parser.py
│   ├── deadly_sins_detector.py
│   └── ... (9 more utilities)
├── shared_store.py        # Shared state management
└── requirements.txt       # Dependencies

The system identifies and eliminates 7 specific AI fingerprint patterns:

1. Em Dash Usage (—)
2. Rhetorical Contrasts ("It's not just X; it's Y")
3. Antithesis (contrasting opposing ideas)
4. Paradiastole (rhetorical reclassification)
5. Reframing Contrasts (shifting meaning)
6. Chiasmus-like Contrast (mirrored structures)
7. Tagline Framing (elevating products)

Advanced metrics evaluate content across multiple dimensions:

• Pattern elimination effectiveness
• Natural flow and readability
• Engagement quality
• Brand alignment
• Human-like qualities

• Track content versions with change history
• Analyze pattern reduction rates
• Measure time savings and ROI
• Optimize presets based on performance

This project demonstrates the power of PocketFlow, a lightweight LLM framework that enables complex agentic workflows through simple Node and Flow abstractions.