DOMAIN
  │
  ▼
BRAINSTORM (diverge) ──→ pick top 1-2 gems
  │
  ▼
RESEARCH (ground in reality)
  │
  ▼
THINK (converge) ──→ intelligence brief
  │
  ▼
RED-TEAM (stress-test)
  │
  ├──→ SURVIVES ──→ FINAL STAKE (done)
  │
  ├──→ WOUNDED ──→ iterate (research gaps → re-think → re-red-team)
  │
  └──→ DEAD ──→ PIVOT DECISION:
                  │
                  ├──→ Adjacent brainstorm (narrow: what's NEAR the dead idea?)
                  │     │
                  │     ▼
                  │   Pick new vein → RESEARCH → THINK → RED-TEAM → ...
                  │
                  └──→ Return to original brainstorm gems (try next best)
                        │
                        ▼
                      RESEARCH → THINK → RED-TEAM → ...

undefined

Wait for user input via AskUserQuestion. If the user specified an idea in the
original invocation AND it appears in the brainstorm results, highlight it.
If their idea did NOT appear, note what the brainstorm found instead.

---

ADJACENT OPPORTUNITY DETECTION:
As you analyze this idea, watch for signs that the REAL opportunity might be
adjacent to or different from the stated idea. If your framework suggests
a better version, a different market, a different approach, or a pivoted thesis
that's stronger than the original — flag it explicitly:

"ADJACENT SIGNAL: [description of the adjacent opportunity and why it might
be stronger than the current idea]"

This is NOT the same as the normal CROSS-FRAMEWORK NOTE. Adjacent signals
suggest the idea itself should change, not just that another framework should
know about a finding.

undefined

Wait for user input via AskUserQuestion.

**Step 4.3: Execute the Pivot**

If **Adjacent Brainstorm**: Run a FOCUSED brainstorm. This is NOT the full 6-agent
brainstorm from Phase 1. It's a targeted 3-agent brainstorm with:
- The Analogist (what solved the adjacent problem elsewhere?)
- The Combinator (what if we combine the dead idea's strengths with a different approach?)
- The Contrarian (what assumption from the dead idea was wrong, and what's true instead?)

Each agent receives: the original brainstorm, the research, the think brief, AND
the red-team kill sheet. They brainstorm WITHIN the context of what was learned.

Write to `thoughts/prospect/YYYY-MM-DD-<domain-slug>-brainstorm-pivot-[N].md`

Then present the new candidates and loop back to Phase 2.

If **Next Gem**: Loop back to Phase 2 with the next idea from the original brainstorm.

/* Prospect-specific */
--prospect-gradient: linear-gradient(135deg, #7c6ff7, #22d3ee, #4ade80);

/* Trail visualization */
.trail-node { /* circles connected by lines */ }
.trail-node.dead { border-color: var(--red); }
.trail-node.wounded { border-color: var(--amber); }
.trail-node.survives { border-color: var(--green); }
.trail-pivot { /* dashed line showing the pivot */ }

undefined

---

You asked about [specific idea]. I brainstormed the broader [domain] space and
found [N] ideas. Your idea appears as Gem #[X]. But the brainstorm also surfaced:

- **[Gem #1]** which scores higher because [reason]
- **[Gem #2]** which is adjacent and avoids [risk your idea has]

Which should I drill into? Your original idea, one of these, or multiple?

undefined

undefined

I hear you — but /prospect's value comes from exploring BEFORE committing.
If you already know the idea and just want to analyze it, /deepthink or /think
are better tools. /prospect is for when the best idea hasn't been found yet.

That said, I can:
A. Run a quick 3-agent brainstorm (5 min) to see if there's something better nearby
B. Skip brainstorming and go straight to research + think + red-team (this is /deepthink)
C. Run the full brainstorm (10 min) as designed