mmlong818/skillforge

Combines a structured 7-step LLM generation pipeline with a 3-step optimization workflow to produce production-grade Agent Skills that conform to Anthropic standards and industry best practices. Built as a full-stack React + Express web application with MySQL persistence, real-time progress tracking, and automatic packaging, while also available as a lightweight Manus Skill for direct conversational use. Supports any OpenAI-compatible LLM API (OpenAI, DeepSeek, Together AI, local Ollama) and includes a 10-dimensional quality audit system that automatically validates and remediates generated Skills against established benchmarks.