agentsclimarketplace

Clarifying requirements

Skill yuchi-chang/no-cape/skills/clarifying-requirements

no-cape keeps the knowledge and drops the cape.

Install
npx -y skills add yuchi-chang/no-cape --skill clarifying-requirements

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use when starting a new feature, component, or project whose requirements are underspecified or ambiguous — before designing or writing code

SKILL.md

2.9 KB, as published. Nobody here has run it

Clarifying Requirements

First: has the approach been chosen?

Clarifying details inside an approach assumes the approach is settled. For a new project or a significant feature, confirm that first — [[exploring-approaches]] runs before this skill, precisely to stop you from refining a branch nobody checked was the right one.

Before asking your first question, check: was the project's fundamental shape ever on the table, or did you walk straight into the user's (or your own) first idea? Signs it was skipped: the goal hides a fork into different projects (cloud API vs self-host vs SaaS; batch vs realtime), the user stated a preference you accepted without examining, or your questions all live inside one unexamined architecture. If so, stop and run [[exploring-approaches]] now — then come back here.

Skip this check when the approach is genuinely not in question: a bug fix, a small change to existing code, or a task whose shape has only one reasonable form. The point is to catch the unexamined fork, not to ceremonially loop through a sibling skill before every clarification.

Code built on guessed requirements is rework with extra steps. But interrogation is friction too: every question must earn its place.

Question discipline

  • Ask only questions whose answers would change what you build. If every plausible answer leads to the same design, don't ask — decide and state your assumption.
  • One question at a time. A five-question dump gets shallow answers to all of them; sequential questions let each answer shape the next one.
  • Prefer multiple choice: 2-4 concrete options with your recommendation first. "Which auth: (a) JWT — recommended because X, (b) server sessions, (c) OAuth-only?" beats "how should auth work?"
  • Ask what the code can't tell you: purpose, users, constraints, success criteria. Don't ask what it can (existing conventions, current structure) — read those instead.

Scope first

If the request bundles several independent subsystems ("a platform with chat, billing, and analytics"), don't refine details yet — surface the decomposition first. Each piece gets its own design cycle.

Before building

  • For non-trivial work, present a short design before code: chosen approach, key components and their boundaries, and what you deliberately left out (YAGNI).
  • Propose 2-3 approaches with trade-offs when the solution space is genuinely open; lead with your recommendation. When one approach is obviously right, skip the ceremony — say so and move on.
  • Design units that can be understood without reading their internals; if changing one unit's internals breaks its consumers, the boundary needs work.

Smell test

If your first artifact was code and the requirements had more than one reasonable interpretation, this skill was skipped.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.