Brainstorm

Run structured ideation sessions with named frameworks, approval gates at every phase boundary, and a clear divergent-to-convergent arc. Output follows ${CLAUDE_SKILL_DIR}/template.md and saves to ~/Tandem/creative/brainstorms/.

For framework operational details, load ${CLAUDE_SKILL_DIR}/references/frameworks.md during Step 2 (framework selection).

---

Language

See METHODOLOGY.md language mirror rule. Reply in the user's language.

---

Step 1: Understand the Problem (Socratic Phase)

Before generating any ideas, understand the problem through targeted questions. Ask one question at a time via AskUserQuestion. Do NOT batch questions.

Q1: Core problem or opportunity

AskUserQuestion: "What's the core problem or opportunity you're exploring?"

Free text response. Listen for signals about problem type (improving something, choosing between options, multi-angle analysis) to inform framework selection.

Q2: Constraints and must-haves

AskUserQuestion: "What constraints or must-haves should I know about?"
Options: ["None -- wide open", "I'll list them now"]

If user lists constraints, capture them for the output file and to scope ideation.

Q3: Audience or stakeholder

AskUserQuestion: "Who's the audience or stakeholder for this?"
Options: ["Just me -- personal decision", "My team", "External clients/users", "I'll specify"]

This shapes idea framing and evaluation criteria.

After all three questions, synthesize the problem statement internally before proceeding. Do NOT generate ideas yet.

---

Step 2: Framework Selection (Gate 1)

Auto-select a framework based on problem type mapping:

Problem Signal	Recommended Framework
Improving an existing product, process, or workflow	SCAMPER
Need to analyze from multiple perspectives, balanced view	Six Thinking Hats
Choosing between 2+ defined options, evaluating tradeoffs	Pros/Cons/Risks

Load framework details from ${CLAUDE_SKILL_DIR}/references/frameworks.md.

Present the recommendation with reasoning via AskUserQuestion:

AskUserQuestion: "Based on your [problem type], I recommend [Framework] because [reason]. Ready to start?"
Options: ["Yes, use [Framework]", "Switch to SCAMPER", "Switch to Six Thinking Hats", "Switch to Pros/Cons/Risks"]

This is Gate 1 -- do NOT proceed without explicit confirmation.

If the user switches frameworks, load the new framework's details from references/frameworks.md before continuing.

---

Step 3: Divergent Ideation

Generate 5-7 ideas using the selected framework's structure.

SCAMPER: Generate 1 idea per lens (7 lenses = 7 ideas). Each idea applies one SCAMPER lens to the problem. See references/frameworks.md for lens details.

Six Thinking Hats: Generate 1-2 insights per hat (6 hats = 6-12 insights). Each insight represents one perspective on the problem.

Pros/Cons/Risks: Generate 3-5 items per column (Pros, Cons, Risks) for each option the user is evaluating.

Format each idea as:

**Idea N: [One-line title]**
[2-3 sentence explanation of the idea, how it applies the framework lens, and
why it could work given the stated constraints]

Present all ideas at once after generation. Do NOT present them one at a time.

---

Step 4: Review Ideas (Gate 2)

Present the complete idea set and ask for confirmation before converging:

AskUserQuestion: "Here are [N] ideas using [Framework]. Ready to evaluate, or want me to adjust?"
Options: ["Evaluate these", "Generate 3 more", "Drop idea [N] and replace", "I want to modify the direction"]

This is Gate 2 -- do NOT proceed to convergent evaluation without explicit confirmation.

If user asks for more ideas or modifications:

• Generate additional ideas following the same framework structure
• Re-present the full updated set
• Ask Gate 2 again

---

Step 5: Convergent Evaluation

Transition explicitly to the convergent phase. Make the shift visible to the user.

5a: Establish evaluation criteria

Suggest 3 default criteria based on problem type:

Problem Type	Suggested Criteria
Product/process improvement	Feasibility, Impact, Effort
Multi-perspective analysis	Alignment, Risk, Novelty
Option evaluation	Cost, Benefit, Risk

AskUserQuestion: "I suggest evaluating ideas on [Criterion 1], [Criterion 2], and [Criterion 3]. Want different criteria?"
Options: ["Use these criteria", "I'll suggest my own", "Add a 4th criterion"]

5b: Rate each idea

Rate every idea against each criterion using High / Medium / Low.

Present as a table:

| Idea | [Criterion 1] | [Criterion 2] | [Criterion 3] | Overall |
|------|----------------|----------------|----------------|---------|
| 1. [Title] | High | Medium | Low | Medium |
| 2. [Title] | High | High | Medium | High |
...

5c: Present ranked results

Sort by overall rating. Present the top 3:

"Here are the top 3 based on [criteria]. Which resonates most?"

AskUserQuestion: "Which ideas do you want to keep as your top picks?"
Options: ["Top 3 as ranked", "Just #1", "I'll pick my own set", "Re-evaluate with different criteria"]

---

Step 6: Final Output (Gate 3)

Compile the full brainstorm summary using ${CLAUDE_SKILL_DIR}/template.md.

Present to user for final confirmation:

AskUserQuestion: "Here's your brainstorm summary with [N] top picks. Save this?"
Options: ["Save as-is", "Edit before saving", "Add more detail to top picks"]

This is Gate 3 -- confirm before writing any file.

If user wants edits, apply them and re-present Gate 3.

After Confirmation

1. Save brainstorm file:

Write to ~/Tandem/creative/brainstorms/brainstorm-YYYY-MM-DD-slug.md using the template format. The slug is: lowercase problem title, spaces to hyphens, strip special characters, truncate to 40 characters.

Create the directory if it does not exist.

2. Log to stats.json:

Read ~/Tandem/stats.json. If it does not exist, create it as []. Parse the JSON array, append a new entry, and write it back:

{
  "type": "brainstorm",
  "action": "completed",
  "count": 1,
  "timeSavedMinutes": 15,
  "description": "Brainstorm: [slug] via [framework]",
  "timestamp": "<current ISO 8601 UTC>"
}

3. Run /sync:

After appending to stats.json, follow the /sync workflow from tandem-skills/core/sync/SKILL.md to rebuild ~/Tandem/dashboard.html with updated statistics. If /sync fails, continue -- the brainstorm file is the primary deliverable.

4. Confirm to user:

Report the saved file path: "Brainstorm saved to ~/Tandem/creative/brainstorms/brainstorm-YYYY-MM-DD-slug.md"

Cross-Skill Offers

After confirming the save, offer one-time cross-skill handoffs if the brainstorm output lends itself to visualization or prompt creation:

AskUserQuestion: "Your brainstorm is saved. Would you like to take it further?"
Options: ["Visualize as mind map", "Create a prompt from the top idea", "Done"]

Rules:

• Offer ONCE only — do not repeat if declined
• Only offer "Visualize as mind map" if the brainstorm produced 3+ ideas
• Only offer "Create a prompt from the top idea" if the brainstorm topic relates to AI/content/creative work where a prompt would be useful
• If user selects an option, briefly describe what the next skill would do, then let the user invoke it naturally (do NOT auto-invoke another skill)

---

Anti-Patterns

• Do NOT skip any gate -- even if the problem seems simple
• Do NOT generate ideas before understanding the problem (Step 1 must complete)
• Do NOT proceed from divergent to convergent without user confirmation (Gate 2)
• Do NOT implement or take action on ideas -- brainstorm output is ideas only
• Do NOT use the Bash tool -- all operations via allowed-tools only
• Do NOT batch multiple questions in one AskUserQuestion -- one question at a time
• Do NOT show empty sections in output -- omit sections with no content

---

Error Handling

• User gives minimal input in Step 1: Work with what is provided. Generate ideas based on available context. Note assumptions in the output file.
• User wants a framework not in the list: Use the closest match and explain the mapping. Mention that additional frameworks may be added in future updates.
• Stats.json write fails: Log the error but do not block. The brainstorm file is already saved and that is the primary deliverable.
• Sync fails: Continue. Dashboard update is secondary.
• No constraints provided: Proceed with unconstrained ideation. Note "No constraints specified" in the output file.

---

Memory

Memory is user-triggered only. Offer to remember:

• Preferred brainstorming framework
• Evaluation criteria preferences
• Domain/industry context for future sessions