What problem does it solve? Writing effective prompts for Claude Code is hard: vague prompts miss tech stack context, acceptance criteria, and the right commands or agents. This Skill diagnoses a raw prompt, detects project context, and produces a ready-to-paste optimized prompt without executing the task itself. ## Core Features & Use Cases - Six-phase analysis pipeline: Detects project tech stack from files like package.json, go.mod, or Cargo.toml, classifies intent (feature, bug fix, refactor, research), and assesses scope from TRIVIAL to EPIC. - ECC component matching: Maps intent and tech stack to specific commands (/plan, /tdd, /verify), skills (tdd-workflow, security-review), agents (code-reviewer), and model recommendations (Sonnet 4.6 vs Opus 4.6). - Gap detection and clarification: Identifies missing context such as acceptance criteria, security requirements, and testing expectations, asking up to 3 clarifying questions when needed. - Use Case: A user pastes "帮我写一个用户登录页面". The Skill detects a Next.js 15 + TypeScript project, then outputs a full optimized prompt with workflow steps, security requirements, acceptance criteria, and scope boundaries, plus a compact quick version. ## Quick Start Ask the assistant to optimize your draft prompt, for example by saying "optimize this prompt: add a REST API endpoint for user profile updates".