Brainstorm Product Ideas (multi-perspective, evidence-seeded)

Multi-perspective ideation (PM / Designer / Engineer) for product discovery. The difference from a generic brainstorm: every idea is seeded by a real unmet need pulled from Evermuse and mapped back to the customer quote that justifies it. If an idea has no evidencing quote, it's flagged as a pure hypothesis.

Step 0 — Relevance & availability

Confirm this is product ideation and Evermuse is connected (see using-evermuse Step 0). If disconnected, produce framework-only ideas labeled ⚠ ungrounded — Evermuse not connected and tell the user to authorize the MCP — but say plainly the ideas are guesses, not evidence-seeded.

Two modes (ask which, or infer)

• Existing product — continuous discovery. Lean on nature=evidence: mine unmet needs, pains, and feature asks from real conversations. This is the default and the strong mode.
• New product / new concept — initial discovery. Evidence may be thin, so lean more on nature=context (market/competitor signals) and nature=guidance (the company's own strategy/vision), and be honest that ideas rest on hypotheses rather than a customer corpus.

Evermuse Grounding (required)

Follow ${CLAUDE_PLUGIN_ROOT}/skills/using-evermuse/SKILL.md.

Required — one parallel batch of searches. Verify the product (Rule 1), then fire in a single parallel batch:

• 3–4 evidence searches, each worded from a different angle (limit up to 50). Evidence comes back rich and varied — expect large, useful result sets.
• one guidance search and one context search (limit up to 50). These are usually sparse or empty; run them anyway and note when they're thin.

Read each response's digest — it reports how many more results exist. Use judgment on whether a query is worth pulling deeper (raise limit toward the 100 max and/or page with next_offset to avoid repeats), weighing payload size, remaining context, task complexity, and the value of the data. For deep pulls, consider spawning sub-agents — instruct them to return every citation with the same metadata the tools return (url, who_said_it, meeting_name, created_at) so you can still cite.

Optional — considered use. Once grounded, reach for the other tools only when they add value: a quote-angled search for verbatim voice, read_source for a single deep dive, inline citations (references/citations.md), and add_source to save the deliverable. Sources, citations, and saving are optional — not required.

Instructions

1. Understand the opportunity. Confirm product, objective, target segment, and desired outcome. Ask if ambiguous — don't ideate against the wrong goal.
2. Extract the unmet needs first. Before generating anything, distill the grounded evidence into a short list of customer unmet needs, each with its sharpest verbatim quote and how many distinct accounts voiced it. These are the seeds. (In new-concept mode where evidence is thin, seed instead from market gaps and strategy, and say so.)
3. Ideate from three perspectives — generate ~5 ideas each, each idea explicitly tied to one of the seed needs where possible:
   • Product Manager — business value, strategic alignment, customer impact.
   • Product Designer — user experience, usability, delight, onboarding.
   • Software Engineer — technical possibilities, data leverage, scalable/novel solutions. ("Best ideas often come from engineers." — Teresa Torres.)
4. Prioritize the top 5 across all perspectives. Weight by: strength of the evidencing signal (how many accounts, how sharp the pain), strategic alignment, impact on the desired outcome, feasibility, and differentiation. For a new concept, weight instead toward core value delivery, speed to validate, and differentiation.
5. For each of the top 5, give: a name + one-sentence description, why it was selected (naming the evidence), and the key assumption to validate next (hand off to /evermuse:assumptions).

Deliverable — the idea→evidence mapping table

Lead with the top 5, then the full idea→evidence map so nothing is unsourced:

### Top 5 ideas
1. **[Name]** — [one-sentence description]. *Selected because:* [reason naming the need].

### Idea → evidencing quote
| Idea | Persona lens | Seed need | Evidencing quote | Accounts |
|------|--------------|-----------|------------------|----------|
| [Idea] | PM/Design/Eng | [unmet need] | > "[verbatim]" — [Name], [Meeting], [Date] [`1`](LINK) | 3 |
| [Idea] | Eng | — | *hypothesis — no evidence yet* | 0 |

Be honest about thin evidence: an idea seeded by a single mention is a weaker bet than one seeded by six across distinct accounts — reflect that in the ranking, don't hide it.

Further reading

• Adapted from phuryn/pm-skills (MIT) — brainstorm-ideas (existing + new). Product Trio & Opportunity Solution Tree: Teresa Torres, Continuous Discovery Habits.