Cursor rule

1---
2description: "Refine a rough idea into a clear, detailed specification through dialogue. Does not produce code. Use when user has a vague idea, wants to think through a feature before planning, or needs to turn \"I want X\" into a concrete spec."
3alwaysApply: false
4---
5 
6# Elaborate Spec
7 
8Turn a rough idea into a clear specification through focused dialogue. No code is written during this skill — the output is shared understanding and a refined problem statement.
9 
10> **HARD GATE** — Do NOT proceed with planning or implementation until the problem space is clearly understood. Success criteria, actors, and scope must be explicit before drafting a plan.
11 
12## Process
13 
14### 1. Listen first
15 
16Let the user describe their idea in their own words. Do not interrupt or redirect. Take notes on:
17- The core problem they're trying to solve
18- Who is affected (actors)
19- What success looks like to them
20- Any constraints they've already identified
21 
22### 2. Ask clarifying questions
23 
24Ask one question at a time. Work through these areas:
25 
26**Problem clarity**
27- What is the current behavior (or lack of behavior) that prompted this?
28- Who experiences this problem? How often?
29- What's the cost of not solving it?
30 
31**Solution boundaries**
32- What is explicitly IN scope?
33- What is explicitly OUT of scope?
34- Are there existing solutions (internal or external) this replaces or integrates with?
35 
36**Success criteria**
37- How will you know this is done?
38- What does the happy path look like end-to-end?
39- What are the key failure modes to handle?
40 
41**Constraints**
42- Any performance requirements?
43- Any compatibility constraints (existing APIs, data formats)?
44- Any non-negotiable implementation decisions already made?
45 
46### 2.5. Multiple Interpretations (HARD GATE)
47 
48> **HARD GATE** — If the request admits ≥2 valid interpretations, do NOT guess. You must list them and ask the user to choose before proceeding. Proceeding with unresolved ambiguity is a failure of integrity.
49 
50Present the options clearly:
51> "I see two ways to read this:
52> 1. [Interpretation A] — my recommendation because [reason]
53> 2. [Interpretation B]
54> Which is closer to what you mean?"
55 
56### 3. Surface hidden assumptions
57 
58Once the user has answered the main questions, probe for assumptions:
59- "You mentioned X — does that mean Y is also true?"
60- "What happens when Z fails?"
61- "Is this for internal users, external users, or both?"
62 
63### 4. Synthesize and confirm
64 
65Summarize your understanding in 3–5 bullet points aligned with [countable-story-format.md](../../docs/countable-story-format.md):
66- The problem (feeds into §1 Business narrative)
67- The solution and main flow (feeds into §5)
68- The key constraints and alternative flows (feeds into §6)
69- The success criteria (feeds into §17 Gherkin)
70- What's out of scope (feeds into §18)
71 
72Ask: "Is this an accurate summary? Anything missing or wrong?"
73 
74### 5. Write specs/planning-context.yaml
75 
76After the user confirms the summary in step 4, persist the key decisions:
77 
78```yaml
79# specs/planning-context.yaml — written by elaborate-spec; consumed by scope-work and slice-tasks
80feature_name: "<from step 1>"
81problem_statement: "<one paragraph>"
82constraints:
83  - "<constraint 1>"
84out_of_scope:
85  - "<excluded item 1>"
86key_decisions:
87  - decision: "<what was decided>"
88    rationale: "<why>"
89```
90 
91If `specs/planning-context.yaml` already exists, ask: `"Planning context from a prior session exists. Update it? [Y/n]"`. Overwrite on Y; leave unchanged on N.
92 
93### 6. Suggest next skill
94 
95Once the spec is clear, recommend the next step:
96- If domain model needs work → `model-domain`
97- If ready to plan → `plan-release` (creates epic capsules with `epic.yaml` + story `.md` + `-tasks.yaml`) then `plan-work` per story
98- If a spike is needed first → `spike-prototype`
99- If architecture decisions are needed → `deepen-architecture` or `grill-me`
100- If the plan depends on a specific library or API → `grill-me` in docs mode
101