AgentSkillsCN

writing-confirmation-dialogs

为用户操作设计清晰明了的确认对话框。适用于当用户触发不可逆操作、破坏性操作，或涉及重大状态变更时，需要用户明确同意时使用。

SKILL.md
--- frontmatter
name: writing-confirmation-dialogs
description: Write clear confirmation dialogs for user actions. Use when users trigger irreversible actions, destructive operations, or significant state changes that require explicit consent.

Writing Confirmation Dialogs

Quick start

Collect or infer:

  • •Action being confirmed
  • •Reversibility (reversible, hard to reverse, irreversible)
  • •Impact scope (single item, multiple items, account-wide)
  • •Consequences of proceeding

Then produce output using TEMPLATES.md. Validate with RUBRIC.md.

Workflow

  1. •Classify the action's severity and reversibility
  2. •Write the headline (what will happen)
  3. •Write the body (consequences, if not obvious)
  4. •Write confirm button (specific verb matching action)
  5. •Write cancel button (safe exit)
  6. •Run the rubric check. Revise until it passes.

Degrees of freedom

  • •Low: Structure (headline + body + confirm + cancel) is fixed
  • •Medium: Body length varies by complexity of consequences
  • •Allowed variation: Body can be omitted for simple reversible actions

Constraints

  • •Headline: max 60 characters
  • •Body: max 200 characters
  • •Confirm button: max 20 characters, must be specific verb
  • •Cancel button: max 15 characters
  • •Confirm button must NOT be generic "Yes" or "OK"
  • •Destructive actions use red/warning button styling (note in output)

References