content-pipeline

What problem does it solve?

The content-pipeline documents and standardizes the selection, orchestration, and secure usage of large language models and providers so teams can generate consistent, high-quality social media content without guessing which model or key to use.

Core Features & Use Cases

• Model Registry & Manager: Central catalog and health-managed selection of provider models to avoid hardcoded IDs and enable safe auto-failover.
• Provider Abstraction: Uniform provider interface (OpenRouter, Anthropic, Google) so code can switch providers without changing call semantics.
• BYOK Key Injection: Secure lookup, decryption, and short-lived usage of user-provided API keys with AES-256-GCM encryption and no logging of plaintext keys.
• Content Generator & Persona Support: Persona-driven prompt construction for platform-optimized posts, captions, threads, and long-form content.
• Content Scoring (Zero AI Calls): Deterministic, real-time scoring across engagement, readability, platform fit, clarity, and emotional impact.
• Repurposing: Automated transformation of a single content asset into platform-specific variants and formats.
• Use Case: Generate a persona-driven Instagram caption, score it for engagement and readability, and produce a repurposed LinkedIn post and Twitter thread.

Quick Start

Generate a LinkedIn post optimized for a professional tone on the topic "product launch", use the user's BYOK if available, and return the content plus a full five-dimension score.