/wbIdea — Expert

What /wbIdea architecturally is

A scored idea pipeline between /wbVision (dreaming) and /wbPlan (committing). It provides the missing lifecycle phase: speculative ideas tracked with advisory scores, validated through exploration, and promoted to plan tasks only when justified.

wbVision ──► wbIdea ──► wbPlan ──► wbWork ──► wbValid
 (dream) (capture) (commit) (execute) (verify)

Ideas are not tasks. No worker/validator assignments, no cost estimates, no DAG dependencies. They have scores, verdicts, and a promotion protocol. The idea file is a staging area; the plan file is the execution contract.

Three design decisions

1. Advisory Score (1–10) vs Priority (P0–P3)

Score = "should this become a task?" Priority = "how urgently should we evaluate it?" A P1/Score-3 idea means "evaluate urgently, but the idea is marginal." Scoring heuristic: impact × 0.4 + feasibility × 0.3 + urgency × 0.3.

2. Promotion Protocol (ideas → plan)

When /wbValid gives 🎯 Promoted, the idea's → Task column links to the plan. When /wbPlan runs, it scans for promoted ideas and auto-ingests them. Unidirectional: ideas flow into plans, never reversed.

3. Dual-pipeline with /wbVision

/wbVision always registers its ideas in idea_*.md. Two outputs: free-form vision file (human reading) + tracked idea rows (pipeline processing).

Verdict scale

Verdict	Score	Meaning
🎯 Promote	8–10	Deserves a plan row. Auto-promotes.
✅ Recommend	5–7	Good idea. Revisit when capacity allows.
⏸️ Defer	3–4	Wrong timing. Context may change.
🚫 Reject	1–2	Not worth pursuing. Reasoning documented.

Producer/Consumer integration

Producers: /wbIdea (native), /wbVision (always), /wbAudit --ideas, /wbWork "idea:...", /wbNext --ideas.

Consumers: /wbWork idea_*.md (explores), /wbExplain (details), /wbValid (verdict + promotion), /wbPlan (ingestion).

Where it leaks

1. Score inflation — models tend toward 6–8. Counter: "re-score assuming you implement it this week."
2. Idea drift — 30-day-old rows never validated. Run --resume monthly.
3. Vision spam — cross-package vision adds 6–10 rows. By design; validation filters.
4. Promotion without exploration — --promote skips feasibility analysis.

What wb-flow-docs's playbook gets wrong about /wbIdea: framing it as a brainstorming tool, when the actual design is score-gated — ideas are evaluated and scored before promotion. The promotion protocol (idea to plan task) is the key innovation, not the idea capture itself.

One-paragraph verdict

Fills the gap between brainstorming and planning. Score, verdict scale, and promotion protocol create a lightweight lifecycle for speculative improvements. Most valuable when ideas exceed capacity. Least valuable when ideas are few and obvious — skip to /wbPlan.

---