Co Scientist Research Planning

Research Planning

Research question formulation, scope definition, and methodology selection.

Use This Skill When

• Starting a new research project from scratch.
• Defining or refining a research question.
• Choosing between qualitative, quantitative, or mixed methods.
• Scoping a feasibility study or pilot.

Workflow

1. Structure the user's request into 6 components:

   • PURPOSE: What decision or knowledge gap to address
   • TARGET: Specific phenomenon, population, or system
   • SCOPE: Breadth and depth boundaries
   • CONSTRAINTS: Time, budget, data access, ethical limits
   • METHODOLOGY: Candidate approaches ranked by fit
   • DELIVERABLES: Expected outputs and success criteria

2. Present the structured plan and wait for user approval ⏸️

3. After approval, generate:

   • results/research-plan.md with the full plan
   • results/methodology-rationale.md with methodology justification

4. Append to logs/process-log.jsonl

Deliverables

• report.md: structured plan summary in user's language.
• results/research-plan.md: 6-component plan.
• results/methodology-rationale.md: methodology selection rationale.

Quality Gates

• All 6 components are addressed.
• Methodology choice is justified with rationale.
• Constraints and limitations are explicitly stated.
• Plan is approved by user before proceeding.

If any gate fails: identify the specific failing check, fix the issue, and re-validate before proceeding.

Gotchas

• 研究テーマが曖昧な場合、1回の質問で明確化しようとせず、1問ずつ段階的にヒアリングする
• 方法論の選択では、データ入手可能性を最初に確認すること。理想的な手法でもデータがなければ実行不可能
• スコープを広げすぎると実行不能になる。最小限の検証可能なスコープから始めることを推奨する

Validation Loop

1. 6 components plan を生成
2. チェック:
   • PURPOSE が意思決定に繋がる形で記述されているか
   • SCOPE が検証可能な範囲に絞られているか
   • CONSTRAINTS に「データ入手可能性」が含まれるか
3. 不合格なら修正して再チェック
4. ユーザー承認後のみ完了