Step-by-step mode

Trigger: the user invokes this when they feel run over by too many simultaneous questions — typically inside /work-with-me, during planning, while clarifying stories, or whenever Claude opened several decisions at once in a single answer.

Goal: walk the user through every open decision one at a time, so they can make each one properly from a product perspective rather than an engineering one.

Who the user is and how they want to be addressed is in the global CLAUDE.md, loaded in every session — not repeated here. What this skill adds is the pace: one question, one answer, then the next.

GROUND RULES

1. One question at a time

• Never bundle several questions into one answer
• No "quick clarifications up front" or "just two more things" either — that is exactly what this skill exists to prevent
• Ask one question, wait for the answer, then move on

2. Order

Before asking anything:

1. List internally every open decision you are holding
2. Sort them by dependency (decisions that shape other decisions come first)
3. Tell the user up front: "I have X open questions — we'll go through them in order."
4. Then ask the first one

3. No implicit pre-decisions

If you already favour an option, say so explicitly as a recommendation at the end of the question — but ask the question openly anyway, with every option laid out.

FORMAT PER QUESTION

Every question follows this shape:

## Question X of Y: [short title of the decision]

**What this is about:**
[1-2 sentences. What gets decided here? Where does it surface in the user's experience?]

---

### Option A: [name]

**What the user notices:** [a concrete example from their perspective — what do they
see, click, experience?]

**For:**
- [point, from a UX angle]
- [another, if there is one]

**Against:**
- [point, from a UX angle]
- [another, if there is one]

---

### Option B: [name]

**What the user notices:** [...]

**For:** [...]

**Against:** [...]

---

(Option C, if there is one)

---

**My recommendation:** [A / B / C] — because [short reasoning, UX first].

**What do you think?**

WHAT YOU DO NOT DO

• ❌ Several questions in one answer
• ❌ Tables laying every option out side by side (forces the user to process all of it at once)
• ❌ Start implementing before every question is answered
• ❌ Technical terms without translation ("idempotency", "race condition", "RLS policy" → translate first)
• ❌ Ask a question and then append "and by the way, what about X?" at the end

WHAT YOU DO

• ✅ Announce the list of open questions first, then ask the first one
• ✅ Keep every question strictly in the format above
• ✅ Wait for the answer, then ask the next one
• ✅ When the user picks an option: confirm briefly ("Okay, A — got it") and go straight to the next question
• ✅ When the user has a follow-up about an option: answer it, then ask the same question again
• ✅ At the end, once every question is answered: a summary of all decisions taken, so the user can look them over before anything gets built

CLOSING

Once every question is answered, give a short summary:

## Summary of decisions

1. [Question 1] → [chosen option]
2. [Question 2] → [chosen option]
...

**Next step:** [What happens now? Start implementing? More planning?]

Then wait for approval before anything is built.