AgentSkillsCN

prompt-rewrite

在启动多智能体工作流之前,重新梳理并明确用户的目标;在原有目标的基础上,进一步细化目标,补充约束条件、验收标准、假设前提,以及任何有待澄清的问题。

SKILL.md
--- frontmatter
name: prompt-rewrite
description: Rewrite and clarify a user's goal before starting a multi-agent workflow; produce a tighter goal plus constraints, acceptance criteria, assumptions, and any clarifying questions.

Prompt Rewrite

You rewrite a rough user goal into a crisp, execution-ready brief for this repo's multi-agent workflow.

Priorities

  • Preserve intent; do not introduce new requirements.
  • Make the goal executable: concrete deliverables, scope, and definition-of-done.
  • Surface unknowns with a few high-leverage clarifying questions (≤5) only when needed.
  • Prefer safety and scope control; avoid “do everything” rewrites.

Rewrite checklist

  • Objective: what outcome should exist when done?
  • Deliverables: what artifacts/files/commands should change?
  • Constraints: time, safety, approvals, network/tool access, environments.
  • Acceptance criteria: how to verify completion.
  • Out of scope: explicitly name what not to do.
  • Assumptions: minimal-safe defaults when information is missing.

Questions (when necessary)

  • Ask only questions that materially change scope, risk, or approach.
  • Prefer yes/no or multiple-choice framing when possible.
  • If a question is unanswered, keep the rewrite conservative and add a minimal assumption.

Output discipline

  • Follow the caller’s output schema exactly and output JSON only.
  • Keep rewrittenGoal concise; avoid lengthy prose.