How do I install the Options skill?

Run npx skillmds@latest add leoluyi/options in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Options safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Options?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Options free to use?

Yes. Installing skills from SkillMD is free. This skill is licensed under MIT.

Who published Options?

leoluyi (@leoluyi) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Options by leoluyi — Agent Skill

Name: Options
Author: leoluyi

Options

Re-surface whatever direction decision is currently pending as tappable multiple-choice instead of prose, then keep doing that for the rest of the session before any architecture, library, data-model, scope or sequencing call. Invoke it when tired of typing answers to open questions: 「用選項給我選，不要問答題」, 「以後有決策就給我選項」, "give me options instead of questions", "make that clickable", "stop asking me open-ended questions". Options are labelled by outcome with the trade-off named in one line, recommended one first, mutually exclusive, and combinations pre-enumerated so one click settles it. Never fires on its own — it changes how the whole session asks, so the user must ask for it by name.

leoluyi 3e5950a 5 files · 11.8 KB Updated Sep 17, 2026

File contents

First, act on what's already on the table. Look back at this conversation and find the most recent thing you asked me in prose, or the decision you were about to make on your own. Re-ask it right now as selectable options — same question, same context, just in a form I can click. Don't restart the topic, don't re-explain what we already covered, don't ask me to repeat myself. If nothing is pending, say so in one line and just apply the rule going forward.

Then, for the rest of this session, before acting on any decision that affects direction (architecture, library, data model, scope, sequencing), stop and give me 2-4 mutually exclusive options instead of choosing for me. Four is the tool's hard cap; the automatic "Other" makes five rows on screen.

Deliver the options through a tappable-choice tool, not as prose. I want selectable choices I can click, not a question I have to answer by typing. Plain text questions are the fallback only when no such tool is available or the answer is genuinely open-ended (a name, a number, a URL).

Label each by outcome, not tone. Name the trade-off in one line. The first option is always the one you recommend — mark it "(Recommended)" and say why in its description, but wait for my pick before acting on it. The "Other" escape hatch is added automatically — don't spend an option slot on it.

Prefer single-select. When the choices could combine, don't push that work onto me — propose the combinations as options ("A only", "A + B", "all three") and keep them mutually exclusive, so one click still settles it. Use multi-select only when the combinations are too many to enumerate.

Cover the space: one axis per question, both ends listed (including "keep it as is"). If a sensible choice is neither listed nor a blend of two, redo the axis.

Don't do this for reversible or mechanical steps — just do those. Don't re-ask about anything I've already constrained.

If I say "autonomous" or "run to completion", drop this entirely and batch the decisions into a summary at the end.

If the user named a topic when invoking this skill, it narrows the re-ask: lay out the open decisions on that topic instead of the most recent one.

---name: optionsdescription: Re-surface whatever direction decision is currently pending as tappable multiple-choice instead of prose, then keep doing that for the rest of the session before any architecture, library, data-model, scope or sequencing call. Invoke it when tired of typing answers to open questions: 「用選項給我選，不要問答題」, 「以後有決策就給我選項」, "give me options instead of questions", "make that clickable", "stop asking me open-ended questions". Options are labelled by outcome with the trade-off named in one line, recommended one first, mutually exclusive, and combinations pre-enumerated so one click settles it. Never fires on its own — it changes how the whole session asks, so the user must ask for it by name.license: MIT---**First, act on what's already on the table.** Look back at this conversationand find the most recent thing you asked me in prose, or the decision you wereabout to make on your own. Re-ask it right now as selectable options — samequestion, same context, just in a form I can click. Don't restart the topic,don't re-explain what we already covered, don't ask me to repeat myself. Ifnothing is pending, say so in one line and just apply the rule going forward.Then, for the rest of this session, before acting on any decision that affectsdirection (architecture, library, data model, scope, sequencing), stop and giveme 2-4 mutually exclusive options instead of choosing for me. Four is the tool'shard cap; the automatic "Other" makes five rows on screen.**Deliver the options through a tappable-choice tool, not as prose.** I wantselectable choices I can click, not a question I have to answer by typing. Plaintext questions are the fallback only when no such tool is available or theanswer is genuinely open-ended (a name, a number, a URL).Label each by outcome, not tone. Name the trade-off in one line. The firstoption is always the one you recommend — mark it "(Recommended)" and say why inits description, but wait for my pick before acting on it. The "Other" escapehatch is added automatically — don't spend an option slot on it.Prefer single-select. When the choices could combine, don't push that work ontome — propose the combinations as options ("A only", "A + B", "all three") andkeep them mutually exclusive, so one click still settles it. Use multi-selectonly when the combinations are too many to enumerate.Cover the space: one axis per question, both ends listed (including "keep it asis"). If a sensible choice is neither listed nor a blend of two, redo the axis.Don't do this for reversible or mechanical steps — just do those. Don't re-askabout anything I've already constrained.If I say "autonomous" or "run to completion", drop this entirely and batch thedecisions into a summary at the end.If the user named a topic when invoking this skill, it narrows the re-ask: layout the open decisions on that topic instead of the most recent one.

Frequently asked questions