Agent skill

clarifying-intent

Use when user asks to create, build, design, add, or generate something — especially when the request is vague, underspecified, or the scope is unclear. Use before starting creative work, features, components, or any new functionality.

Stars 16
Forks 2

Install this agent skill to your Project

npx add-skill https://github.com/l-lin/dotfiles/tree/main/home-manager/modules/share/ai/.config/ai/skills/clarifying-intent

SKILL.md

Clarifying Intent

Overview

Don't build what wasn't asked for. Before starting any creative or generative work, use Socratic questioning to surface the user's real intent, constraints, and success criteria.

Core principle: A well-understood problem is half-solved. Questions are cheaper than rework.

Core Pattern

STOP. Ask before coding.

Use ask-user-question tool to probe only the dimensions that are actually missing. Don't ask everything — ask what matters.

Dimensions to Consider

  • Intent / Why — ask when purpose isn't stated → "What problem does this solve?"
  • Audience / Who — ask when it affects UX or tone → "Who will use or see this?"
  • Constraints — ask when tech, time, or style limits are unknown → "Any constraints I should know about?"
  • Success criteria — ask when "done" is unclear → "What does good look like to you?"
  • Scope — ask when it could be large or small → "Is this a quick prototype or production-ready?"

Decision Heuristic

Ask yourself: "If I'm wrong about X, does the whole thing need to be redone?"

  • Yes → Ask about X
  • No → Assume and note assumption

Target: 2–4 questions max. More than that is interrogation, not clarification.

Implementation

User: "Add a dashboard component"

Agent thinks:
- Why? (unclear — metrics? navigation? admin?)
- Who? (unknown — end user? internal?)
- Scope? (unknown — full page? widget?)
- Constraints? (assume existing stack unless asked)

Agent asks (using ask-user-question):
1. "What's the main purpose — monitoring data, navigation, or something else?"
2. "Who's the audience — end users, admins, internal team?"
3. "Is this a quick prototype or production-ready?"

→ Block until answered. Then build with full context.

Blocking Rule

Do not start implementation until answers are in.

  • Don't "start on the obvious parts" while waiting
  • Don't make assumptions silently — if you must assume, state them and ask for confirmation
  • Don't write placeholder code "to save time"

Common Mistakes

  • Asking all 5 dimensions every time — pick only what's missing, read the request
  • Proceeding with assumptions silently — state assumptions explicitly, ask for confirmation
  • One giant multi-part question — use ask-user-question with separate, focused options
  • Re-asking after user answered — track answers; don't loop
  • Skipping this for "simple" requests — simple requests have the most wrong assumptions

Expand your agent's capabilities with these related and highly-rated skills.

Didn't find tool you were looking for?

Be as detailed as possible for better results