jam

Brainstorming Skill

Generate diverse perspectives through structured focus group sessions.

When to Use

• User wants to explore an idea
• Decision needs multiple viewpoints
• Problem needs creative approaches
• User says "brainstorm", "perspectives", "focus group", "what do you think about"

Session Types

Full Brainstorm (/brainstorm)

• 4-6 personas
• 2-3 discussion rounds
• Full synthesis with insights
• Storage offered

Best for: Important decisions, complex problems, strategic thinking

Quick Jam (/jam)

• 4 personas
• 1 round (forced — never extends)
• Brief synthesis (2-3 insights max)
• No storage prompt

Best for: Quick decisions, gut checks, rapid exploration

Perspectives Only (/perspectives)

• 4-6 personas
• 1 round
• No synthesis (user decides)
• Raw viewpoints

Best for: Gathering input, self-directed thinking, discussion prep

Persona Design

Archetypes

Type	Focus	Examples
Technical	How it works	Architect, Debugger, Security
User	Who uses it	Power User, Newcomer, Support
Business	Why it matters	PM, Skeptic, Evangelist
Process	How it ships	Maintainer, Tester, Ops

Selection Principles

1. Relevance: Choose personas that care about the topic
2. Diversity: Cover multiple angles
3. Genuine: Each has legitimate concerns (no strawmen)
4. Buildable: Personas can respond to each other

Convergence Modes

Mode	Flag	Behavior
Normal	(default)	Run all planned rounds, then synthesize
Fast	--converge fast	After each round, check if signal is sufficient; skip remaining rounds if yes

Fast convergence checks three criteria after each round: (1) at least 2 actionable insights, (2) directional agreement, (3) tensions are well-characterized. If all pass, synthesis begins immediately.

Quick jam (/jam) always uses forced fast convergence (1 round, then synthesize).

Discussion Dynamics

Round 1: Initial Perspectives

Each persona shares their view:

• Position on the topic
• Key concern
• Suggestion or consideration

Convergence check (fast mode: synthesize if criteria met)

Round 2: Building

Personas respond to each other:

• Build on good ideas
• Challenge assumptions
• Find connections

Convergence check (fast mode: synthesize if criteria met)

Round 3: Convergence (optional)

Find common ground:

• Agreements
• Remaining tensions
• Key tradeoffs

Synthesis Quality

Good synthesis:

• 3-5 insights (not 10)
• Confidence levels (HIGH/MEDIUM/LOW)
• Action items (prioritized)
• Open questions (honest about unknowns)

Bad synthesis:

• Just summarizing what personas said
• Too many insights (information overload)
• No confidence assessment
• No actionable next steps

Integration

With wicked-garden:mem

At session start:

prior = mem_recall(topic)  # via /wicked-garden:mem:recall
inject_context(prior)

At session end:

if user_approves:
    mem_store(insights, type="decision", tags=[topic])  # via /wicked-garden:mem:store

With native tasks

Sessions are tracked as native tasks for process visibility:

# On session start
TaskCreate(
  subject="Jam: {topic}",
  metadata={
    "event_type": "task",
    "chain_id": "jam-{topic-slug}.root",
    "source_agent": "jam-facilitator",
    "initiative": "{topic-slug}"
  }
)

# After each persona round — append to description
TaskUpdate(taskId, description="{previous}\n\n{persona_name}: {key_insight}")

# After synthesis
TaskUpdate(taskId, description="{previous}\n\nSynthesis: {summary}")

# On decision
TaskUpdate(taskId, description="{previous}\n\nDecision: {decision_record}")

Wrap all task calls in graceful degradation — if TaskCreate/TaskUpdate fail, skip silently. Native tasks track process; wicked-garden:mem stores outcomes. Both can coexist.

With wicked-crew

Called during clarify, design, and build phases. See "Crew Engagement" section below.

Crew Engagement

Phase	Mode	When
Clarify	Quick jam (4 personas, 1 round)	Approach options, framing decisions
Design	Full brainstorm (4-6 personas, 2-3 rounds)	Architecture decisions, complex tradeoffs
Build checkpoint	Quick jam	Course corrections when unexpected complexity arises

Jam auto-engages when: ambiguity detected, complexity >= 4, architecture signals present, or content/documentation-heavy scope.

References

• refs/facilitation-patterns.md — Match persona archetypes to problem types; session length guidance; anti-patterns to avoid
• refs/synthesis-patterns.md — Synthesis structure, quality checklist, techniques (non-obvious connections, surprising agreements, productive tensions), decision record format, examples

Output Structure

Synthesis-first: ## Brainstorm: {Topic} → ### Key Insights (read first) → ### Action Items → ### Open Questions → ### Session Details (personas, rounds, summary).