What problem does it solve? Raw prompts typed into AI assistants are often vague, incomplete, or missing critical technical context, which leads to generic or incorrect responses. This Skill rewrites those raw instructions into precise, context-rich prompts before execution, without ever answering the prompt itself. ## Core Features & Use Cases - Dual-Mode Enrichment: Mode A (default) produces a single cohesive prompt block for standard requests, while Mode B generates a structured multi-section prompt (Role, Task, Input, Output, Constraints, Success Criteria) for complex technical or AI/ML tasks. - Strict Language Mirroring: All output, including headings, notes, and clarification questions, matches 100% of the user's input language (Vietnamese, English, Japanese, etc.), while technical identifiers stay in standard English. - Gap Analysis with Smart Defaults: Audits prompts across 7 core dimensions and 7 engineering criteria (P1–P7), auto-filling low-risk gaps and asking at most 1–2 targeted questions for high-risk ambiguities. - Use Case: A user types "Sửa lỗi crash khi upload ảnh lớn" and receives a fully enriched Vietnamese prompt specifying root-cause analysis, streaming/chunking fixes, early validation, error handling, and verification steps. ## Quick Start Ask the assistant to enrich or optimize your raw prompt, for example by typing "enrich-prompt: add JWT authentication to my Express API" and receive a rewritten execution-ready prompt.
