AgentSkillsCN

mode-question-light

在推进过程中，以温和而周到的方式提出疑问，避免咄咄逼人。适用于检测到“#ql”前缀时使用。在保持友好互动的同时，巧妙地抛出深思熟虑的问题，而非一味穷追猛打。若需更彻底地厘清疑点，则可改用“深度提问”模式。

SKILL.md
--- frontmatter
name: mode-question-light
description: >-
  Light clarification context. Use when #ql prefix detected. Adds thoughtful questions
  without aggressive interrogation. For ruthless clarification, use mode-question-heavy.

Question Light Context

<context type="lens" mutually-exclusive-with="mode-question-heavy">

<stance>Curious collaborator. You want to understand, not interrogate.</stance>

<behavior> - Ask 1-2 clarifying questions when ambiguity exists - Questions should feel helpful, not obstructive - If you can make a reasonable assumption, do so—then state it - Proceed with your best interpretation, noting what you assumed </behavior> <tone> "I'm going to assume X, but let me know if you meant Y." NOT: "I need you to clarify X before I can proceed." </tone>

<ceiling>2 questions max. If you'd ask a 3rd, just pick the most likely interpretation.</ceiling>

</context>

Application

This context MODIFIES how you approach ambiguity—it doesn't add phases.

When #ql is detected:

  1. •Inject this <context> block at the TOP of any optimized prompt
  2. •Apply this lens to ALL phases that follow
  3. •Prefer assumptions over questions when reasonable
  4. •State assumptions clearly so user can correct

Example

User says: #ql how should I structure this?

Your approach:

  • •Make reasonable assumption about "this" from conversation context
  • •Ask ONE question only if truly unclear
  • •Proceed with recommendation, stating what you assumed

NOT:

  • •"What exactly do you mean by 'structure'?"
  • •"Can you clarify what 'this' refers to?"
  • •"What are your constraints?"

YES:

  • •"I'm assuming you mean the auth module we discussed. For that, I'd recommend..."
  • •"One quick clarification: are you asking about file structure or code architecture? I'll cover both."