brainstorm

name

brainstorm

description

Vision refinement skill for this team. Invoke with /brainstorm. Reads memory/project/INDEX.md to learn the active project and either loads the active vision or starts a fresh one. Leads a structured interview to co-create or update vision.md — covering problem framing, target users, value proposition, key capabilities, and success signals. Use whenever the team is starting a new initiative or pressure-testing direction.

Brainstorm

Produces a vision.md artifact through a structured interview. The product domain is whatever the active session in INDEX.md describes — never assume a fixed domain.

Memory Read Contract

Required reads on entry:

1. memory/project/INDEX.md — canonical pointers + session lifecycle.
2. memory/task/state.md — current phase, active session, blockers.
3. The active session's vision.md (resolved from INDEX), if one exists.

Rules:

• Always resolve artifact paths through INDEX. Do not fuzzy-find via find.
• If INDEX is missing or points to a missing file, stop and tell the user to run /reindex. Do not silently fall back to globbing.
• The active session is whichever session has status: active in INDEX. Never use date-affinity heuristics.

Flow

RESOLVE ACTIVE VISION → INTERVIEW → SYNTHESIZE → SAVE ARTIFACT (transactional)

---

Step 1: Resolve active vision

Read INDEX.md. If a vision.md is listed under canonical artifacts:

• Read it, summarise what's there in 2–3 sentences, then ask: "Refine the existing vision, or start a fresh session?"

If none exists in INDEX:

• Proceed directly to the interview. A new session folder will be created on save.

If "fresh session" is chosen, the existing session is marked superseded in INDEX during the write transaction (Step 4).

---

Step 2: Interview

The goal is to surface the sharpest possible answer to: What does this product do, for whom, and why does it matter?

Work through these question clusters one at a time. Wait for the user's answer before moving to the next. Do not batch multiple clusters in one turn.

Cluster A — The Problem

1. "What is the core problem this product solves? Describe it in plain language, as if talking to that user."
2. "When does this problem hurt most — what's the specific moment or trigger that makes the user feel it?"

Cluster B — The User

3. "Who is the primary user?"
4. "What does this user currently do — spreadsheet, an existing app, nothing? What's broken about that?"

Cluster C — The Value

5. "Finish this sentence: 'This app is the only one that _______.'"
6. "What's the single most important thing a user should be able to do on day one?"

Cluster D — Success

7. "How will you know in 6 months that this product is working? What does a successful user look like?"

Cluster E — Scope guard

8. "What is explicitly out of scope — things you'll never build, even if users ask?"

After all clusters, do a brief synthesis pass: "Here's what I heard — [3-bullet summary]. Does this feel right, or is anything off?" Adjust until the user confirms.

---

Step 3: Synthesize

Draft vision.md from the confirmed interview. Rules:

• No fluff. Every sentence must be load-bearing.
• User's language. Use their words, not product-marketing words.
• Concrete over abstract. Specific behaviour beats generic "helps users save money."
• Decisions are decisions. Out-of-scope items belong in the Out of Scope section, not footnotes.

---

Step 4: Memory Write Transaction

Write order, no skipping:

1. Save the artifact:
   • For a refinement: overwrite the active session's vision.md.
   • For a fresh session: create exercise_one/YYYY-MM-DD-<scope-slug>/vision.md (today's date, 2–4 word kebab-case slug).
2. Verify the file exists and is non-empty.
3. Update memory/project/INDEX.md:
   • For a fresh session: add the new session row with status: active. Mark the previous active session superseded and record supersedes on the new vision pointer.
   • For a refinement: leave session lifecycle untouched; refresh the vision pointer's date if needed.
   • Recompute Canonical artifact set complete and update Last reindex if you reconciled disk state.
4. Replace memory/task/state.md wholesale with the new snapshot (phase = post-vision, next action = /research or /domain).
5. Run the Decision Gate. Append qualifying entries to memory/project/decisions.md. Common candidates here: out-of-scope items that foreclose a future product direction.
6. Output any glossary candidates as a ## Glossary Proposals section in the final response. Do not edit glossary.md.

If any step after step 1 fails or is skipped, the final response must say:

Memory drift possible. Run /reindex to reconcile.

---

Step 5: Decision candidates

End the session with this block:

Decision candidates from this session:
1. [candidate] — append: yes/no — reason
2. [candidate] — append: yes/no — reason

Only append: yes candidates are written to decisions.md. A scope-out decision qualifies when it forecloses a real alternative ("we will not do tax filing — chose financial flow over financial compliance"). Vague preferences do not qualify.

---

vision.md template

# Vision: [Short product name or initiative title]

_Date: YYYY-MM-DD_

---

## Problem

[The specific problem this product solves, in the user's language. The trigger moment. Why now.]

## User

[Who they are. What they currently use and why it fails them.]

## Value Proposition

[The one thing this product does that nothing else does. One paragraph max.]

## Day-One Capability

[The single most important thing a user can do on first use.]

## Success Signal

[What a successful user looks like in 6 months. Concrete, observable.]

## Out of Scope

[Explicit list of things this product will never do, even if users ask.]

## Open Questions

[Unresolved questions to carry into /research or /domain.]