description: "Research phase orchestrator: AI Share of Voice & Brand GEO Prompts Discovery" version: "1.0"

/seo-research-brand-prompts

Purpose

Executes Phase C2: Brand Positioning & AI Share of Voice (SoV) Prompts Discovery. It uses the brand context, USP guidelines, and competitors list to dynamically generate a scale-adaptive list of conversational search queries (between 15 and 100 queries) to monitor brand mentions and sentiment on AI search systems.

The output is .seo/data/brand-prompts.toon, which will be used in Phase 5 (/seo-monitor) to calculate the brand's live Share of Voice (SoV).

Skills Activated (3)

Prerequisites

• Required: Project configuration project.json must exist.
• Recommended: Onboarding completed (/seo-onboard).

Sub-Agent Coordination

• Primary Coordinator: @[rd/algo-forecaster] (Orchestrates prompt discovery and runs evaluations)
• Generator Group (Sub-Agents):
  • Top-Funnel Specialist: Focuses on organic recommendation, USP-led informational prompts.
  • Mid-Funnel Specialist: Focuses on symmetrical competitor cross-comparisons.
  • Bottom-Funnel Specialist: Focuses on direct brand factuality and health queries.
• Socratic Critic: @[content/search-intent-profiler] (with @[content/topical-authority-planner] to debate and validate queries for plausibility and niche-alignment).

Execution Flow

Step 1: Init Context

• Chạy lệnh Python sau để trích xuất các biến ngữ cảnh từ project.json và lưu thành tệp context:
  • Lệnh: python .agents/scripts/run_plugin_script.py seo-geo-monitor brand_prompts_generator.py {domain} --action prepare
  • Output: .seo/data/content_plan_batches/brand_prompts_context.json

Step 2: Context Check (Mini-Interview)

• If essential information (like brand name, products, or competitors) is missing or marked as unknown in project.json:
  • Pause and run a mini-interview directly in the chat to collect:
    1. Tên thương hiệu hiển thị (Brand display name).
    2. Danh sách 3-5 sản phẩm/dịch vụ cốt lõi.
    3. Danh sách 3 đối thủ cạnh tranh chính.
  • Automatically save this information back to .seo/project.json to enrich the project context and re-run Step 1.

Step 3: Multi-Agent Parallel Prompt Generation (Funnel Batching)

• Coordinator (rd/algo-forecaster) spawns 3 specialized Generator Sub-Agents using invoke_subagent to write to separate batch files:
  1. Top-Funnel Specialist loads brand_prompt_gen_top.md.
     • Output: .seo/data/content_plan_batches/brand_prompts_draft_top.toon
  2. Mid-Funnel Specialist loads brand_prompt_gen_mid.md.
     • Output: .seo/data/content_plan_batches/brand_prompts_draft_mid.toon
  3. Bottom-Funnel Specialist loads brand_prompt_gen_bottom.md.
     • Output: .seo/data/content_plan_batches/brand_prompts_draft_bottom.toon

Step 4: Socratic Debate & Critic Refinement

• Critic Agent (content/search-intent-profiler) loads brand_prompt_critic.md.
• Scans all draft TOON files and reviews each generated query:
  • Top Funnel: Checks for absolute brand/competitor exclusion, and natural solution-seeking intent.
  • Mid Funnel: Checks for balanced comparison symmetry (same products/features) and user plausibility.
  • Bottom Funnel: Checks for factual utility (pricing, reputation, contact details).
• For any REJECTED queries, the corresponding Generator Sub-Agent is invoked to rewrite the query according to suggested_rewrite.
• Once all queries are APPROVED, the final drafts are ready.

Step 5: Merge & Finalize Outputs

• Chạy lệnh Python sau để tự động tìm kiếm, gộp toàn bộ tệp nháp đã qua kiểm duyệt (brand_prompts_draft_*.toon), chạy các bộ quy tắc validation cứng (diacritics-aware contains_phrase checks), xuất báo cáo và ghi nhận checkpoint:
  • Lệnh: python .agents/scripts/run_plugin_script.py seo-geo-monitor brand_prompts_generator.py {domain} --action import
  • Output:
    • Cơ sở dữ liệu câu hỏi validated: .seo/data/brand-prompts.toon
    • Báo cáo tóm tắt câu hỏi: .seo/reports/brand-prompts-{date}.md
• Cập nhật checkpoint và ghi nhận lịch sử chạy (runs) vào project.json tự động qua script.

Pause Gate (Duyệt Brand Prompts)

Before proceeding to traditional keyword harvesting, the workflow halts:

1. Please open and review the generated tracking queries at .seo/reports/brand-prompts-{date}.md.
2. Confirm if the queries reflect your positioning correctly. You can edit the query list directly by editing .seo/data/brand-prompts.toon using standard TOON format.
3. Once approved, run /seo-research again to get the next step suggestion.

Final Dashboard

# 🎯 Báo cáo Thiết lập Câu hỏi Theo dõi SoV — `{domain}`

**Trạng thái**: Hoàn tất sinh bộ câu hỏi Share of Voice (AI SoV)

## 📊 Tổng quan Bộ câu hỏi

Từ thông tin định vị thương hiệu và đối thủ cạnh tranh, hệ thống đã sinh bộ câu hỏi:

- **🔑 Tổng số câu hỏi theo dõi**: **{total}** câu hỏi.
- **Top Funnel (Đầu phễu)**: {top_count} câu hỏi thông tin ngách chung.
- **Mid Funnel (Giữa phễu)**: {mid_count} câu hỏi so sánh giải pháp.
- **Bottom Funnel (Cuối phễu)**: {bottom_count} câu hỏi trực diện thương hiệu.

## 📁 Dữ liệu & Lưu trữ

- **Cơ sở dữ liệu Prompts (TOON)**: `.seo/data/brand-prompts.toon`
- **Báo cáo tóm tắt câu hỏi**: `.seo/reports/brand-prompts-{date}.md`
- **Dấu thời gian**: {timestamp}

---

💡 **Bước tiếp theo (Next Step):**
Duyệt qua danh sách câu hỏi trong báo cáo, sau đó tiếp tục bước Nghiên cứu Từ khóa cốt lõi:
👉 `/seo-research-keywords` (Keyword Harvesting)