Skip to content

dvy1987/agent-loom / .agents/skills/brainstorming

brainstorming

Verified

Turn a rough idea into a fully approved design before any code is written. Load when the user wants to brainstorm, explore ideas, design a feature, think through approaches, plan a new capability, or figure out what to build. Also triggers on "let's think through", "help me design", "explore options", "what's the best approach for", "I have an idea for", "before we build", or any request to design something before implementation. Enforces a hard gate: no code, no implementation until user approves a design. For executable feature specs (FRs, NFRs, ACs as Given/When/Then), route to `feature-spec` instead — brainstorming owns approach and architecture, not machine-readable requirements.

FileKindSize
SKILL.mdskill md8.8 KB
references/examples.mdmarkdown5.5 KB

references/examples.md

View on GitHub
# Brainstorming — Full Session Examples

Read when you need a complete walkthrough beyond the inline teaser in SKILL.md.

---

## Example 1 — Feature design (dark mode)

**Input:** "I want to add dark mode to our React app"

**Phase 1 — Frame (How Might We)**
```
How Might We let users switch between light and dark themes without breaking
existing custom components or increasing bundle size significantly?
```

**Clarifying question (one at a time):**
> Do you have an existing design system or component library (Tailwind, MUI, styled-components), or are styles written manually?

**User:** "Tailwind v4 + a few custom CSS modules for charts."

**Phase 2 — Options (3 divergent approaches)**

| Option | Summary | Trade-off |
|---|---|---|
| A. CSS variables + `class="dark"` on `<html>` | Token swap at root | Fast; chart CSS modules need manual vars |
| B. Tailwind `dark:` variant only | No extra runtime | Verbose; easy to miss a component |
| C. User preference in localStorage + system sync | Best UX | More state logic; flash-of-wrong-theme risk |

**Phase 3 — Converge**
Recommend **A + C**: CSS variables for charts, Tailwind dark: for app shell, `prefers-color-scheme` default with override.

**Design doc sections produced:**
- How Might We statement
- Target user: power users who work evenings; success = readable charts in both modes
- **Not Doing:** Per-component theme props (scope creep); auto-detect without manual toggle (accessibility concern)
- Key assumptions: chart library supports CSS color vars (validate in spike)
- HYPOTHESIS: 80% of screens themable via tokens alone — CONFIDENCE: 65%

**Output path:** `docs/specs/YYYY-MM-DD-dark-mode-design.md` — **no code until user approves.**

---

## Example 2 — Route to venture-exploration

**Input:** "I'm thinking of building a SaaS for freelance invoicing"

**Step 1 detect:** Business idea, not feature design → invoke `venture-exploration` instead of continuing brainstorming.

**Handoff message:**
> This sounds like a business validation question, not a feature design session. I'll route to venture-exploration for idea cards, Lean Canvas, and Mom Test interviews. Want to continue there?

---

## Example 3 — Anti-pattern recovery

**Input:** "Just pick the best architecture and implement auth"

**Response (Hard Gate):**
> Brainstorming blocks implementation until an approved design doc exists. I can either (1) run a 10-minute design pass on auth approach options, or (2) route directly to `feature-spec` if you already have an approved spec. Which?

---

## Verification checklist (full session)

- [ ] How Might We statement is clear
- [ ] Target user and success criteria defined
- [ ] Not Doing list makes trade-offs explicit
- [ ] Key assumptions listed with validation approach
- [ ] User approved design before implementation handoff

---

## Example 4 — Venture idea (restaurant regulars) — from idea-refine

**Input:** "I want to help local restaurants compete with DoorDash"

**Phase 1 — Reframe:**
> How might we help restaurants **retain** existing customers who already order from them — instead of competing for discovery on delivery apps?

**Clarifying:** Who is the user — owner, diner, or both? What does "compete" mean — more orders, higher margin, or less platform dependency?

**User:** "Restaurant owners. Compete = keep regulars ordering direct so they skip 30% platform fees."

**Variations (sample):**
1. **SMS reorder for regulars** — owner uploads top 20 phones + usual order; one-tap confirm link.
2. **Inversion — loyalty without an app** — punch-card via SMS, not another install.
3. **Simplification — pickup-only v1** — no delivery logistics in MVP.

**Phase 3 one-pager highlights:**
- **MVP:** SMS reorder for self-identified regulars, pickup only.
- **Not Doing:** Delivery, marketplace discovery, branded apps, POS integration in v1.
- **Assumption to validate:** Regulars will reorder via SMS vs their usual app (5 restaurants × 20 regulars, 4 weeks).

---

## Example 5 — Codebase-aware feature (real-time collaboration)

**Input:** "Add real-time collaboration to our document editor"

**Context scan:** Block-based editor, REST persistence, no WebSocket layer yet.

**Clarifying:** Same-paragraph editing (Google Docs) vs different sections (Notion)? How many simultaneous users?

**User:** "2–5 people, different sections. Losing deals — buyers expect real-time."

**Variations:**
1. **Block-level locking** — `locked_by` + presence; fits existing block model.
2. **Async-first** — suggestions/comments as primary mode.
3. **Presence without co-edit** — cursors only; coordinate verbally.

**Recommendation:** Variation 1 or 3 to **close the competitive checkbox** quickly; defer character-level CRDT.

**Not Doing (Phase 3):** Offline sync, AI-mediated collab, per-character merge in v1.

---

## Example 6 — Process idea (stale retrospectives)

**Input:** "Fix our weekly retros — people dread them"

**Diagnosis:** 8 people, 3–4 talk; same WWW format; **action items pile up unchanged** — output layer is broken, not just format.

**Variations:**
1. **One action item max** — demo completion at next retro or that *is* the topic.
2. **Async retro** — written reflections; 15-min standup to agree one change.
3. **Anonymous dot-vote** — structure for quiet voices.

**Phase 3 recommendation:** Start variation 1 + anonymous submission (variation 3) — **$0 prep**, test next week.

**Not Doing:** New tools, elaborate facilitation kits, budget spend before proving follow-through works.

brainstormingobra/superpowersYou MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation).Stars: 322kUpdated Claude CodeCodexVerifiedInstall
brand-guidelinesanthropics/skillsApplies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
frontend-designanthropics/skillsGuidance for distinctive, intentional visual design when building new UI or reshaping an existing one. Helps with aesthetic direction, typography, and making choices that don't read as templated defaults.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
mcp-builderanthropics/skillsGuide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK).Stars: 198kUpdated Claude CodeCodexVerifiedInstall