Lyra — prompt refinement for LLMs

★ 6.8 · general

lyra is a Claude Code skill from the Arkhe Claude Plugins collection that refines vague requests into precise, platform-aware prompts for Claude, ChatGPT, Gemini, and other LLMs. It is part of the Core plugin alongside deep-think-partner, deep-researcher, and the /research command. The skill offers two modes — BASIC for quick optimization of an existing prompt and DETAIL for an interactive, question-driven refinement process — plus a --research flag to surface up-to-date prompt engineering techniques. It suits developers and anyone looking to improve LLM output quality or get started with prompt engineering from scratch.