description

AFP Engineering Prompt Generator - Master-Level System Architecture 🎯 Core Positioning: A meta-level prompt engineering system—the foundational architecture used to generate other AFP skills. The entity is both an independent product and the core of AFP skills across various domains. 📊 Core Capability Matrix: Table Capability Dimensions System Support Application Scenarios Task Adaptive Recognition ✅ Automatic recognition of 8+ types of tasks including classification, generation, analysis, and decision-making. No need to manually specify task type; automatic complexity trimming. ✅ Automatic adjustment of components based on task complexity (1 layer/multi-layer/conditional branch). Avoids over-design and omission of key elements. AFP design gene encapsulation. ✅ Meets engineering prompt word standards (input specifications → processing flow → output verification). Generated prompt words are naturally high-quality and support multiple modes. ✅ Seamless switching between independent use and being called. Can be deployed independently or used as a foundation for other skills; cross-domain reusability. ✅ Automatically splits general parts/domain-specific parts. The same framework can be applied across 5+ domains. Large model compatibility. ✅ Universal for ChatGPT/Claude/GPT-4/domestic large models. Generated prompt words are not bound to a single model. 🔧 Technical Architecture: First Layer: Task Analysis Engine Natural Language Understanding User-input task description Automatically classifies task type (generational/analytical/decision-making/creative, etc.) Calculates task complexity score. Second Layer: The component library management includes 50+ built-in engineered prompt components (role setting, input specifications, process design, exception handling, etc.), categorized by complexity level (L0 Simple/L1 Medium/L2 Advanced/L3 Expert). It supports selective assembly and custom expansion of components. The third layer, AFP framework generation, organizes prompt structures according to AFP design principles (functional layering → process orchestration → output formatting), automatically generating complete instruction chains. Built-in quality checks (coverage, redundancy, consistency checks) are also included. The fourth layer, output and integration, generates plain text prompts (ready to use directly) and structured configuration files (callable by programs). Version management and iterative optimization are supported. Expected performance indicators: Table Indicators | Improvement | Description Prompt Design Cycle | From 1-2 weeks → 10-30 minutes | From manual design to automatic generation | Output Quality Stability | From 70-80% → 85-92% | Engineered design is inherently more stable | Cross-domain Reusability | From 30% → 80%+ | Automatic separation of general and special parts | Team Learning Cost | From 3-6 months to master → 1-2 weeks to get started Newcomers can quickly reuse high-quality frameworks, reducing the migration cost of large models. The framework is stable, requiring no major modifications for model upgrades, moving from a complete rewrite to minor tweaks.