craft-promptlisted

# craft-prompt ## Purpose Turn a goal, scattered notes, or a raw ask into a clear, well-structured prompt that gets the best results from any LLM — a separate step from doing the task itself, so the result is a copy-paste-ready text block the user can drop into any AI interface. ## Use this when - the user asks to write, make, or build a prompt from scratch - scattered notes need to become a reusable prompt or template - a `/goal` condition or other reusable prompt template is needed - Korean prompt requests appear, such as "프롬프트 만들어" or "프롬프트 작성" ## Inputs - **Outcome** — what the target should accomplish or return - **Target** — the model, product surface, or agent interface when it changes the prompt - **Context** — facts the target cannot reliably infer or retrieve itself - **Reuse** — one-shot use, or a reusable template with placeholders - For coding-agent or worktree prompts: the execution context — default to paths relative to the current worktree root unless the user explicitly needs machine-specific absolute paths Don't over-ask. Infer what the request or available context already establishes, and ask only when the answer would materially change the prompt. ## Workflow 1. **Resolve the outcome.** State what should be true or delivered at the end. Preserve a shape, length, tone, or language only when the user named it or the consumer requires it. 2. **Gather only load-bearing context.** The delivered prompt usually travels without the current conversation, s