Skip to main content

conduct-a-case-study-interview

Gather attributable customer evidence through a prepared and consent-based interview. Use when a case study needs verified context, process, contribution, results, limitations, and approval rather than a promotional testimonial.

ソース情報

リポジトリ
protosphinx/sphinxstack
ソースの最終更新活動
2026年7月29日 13:55
検出された SKILL.md の言語
英語
スター
3
フォーク
0

インストール方法

デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。

ソースファイルを確認

インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。

ファイルエクスプローラー
2 ファイル

SKILL.md を表示中

SKILL.md
ソースの指示 · 読み取り専用プレビュー
name
conduct-a-case-study-interview
category
write
description
Gather attributable customer evidence through a prepared and consent-based interview. Use when a case study needs verified context, process, contribution, results, limitations, and approval rather than a promotional testimonial.
# conduct-a-case-study-interview Collect the customer's account without forcing it into a predetermined success story. ## Procedure 1. Confirm interview purpose, participant authority, incentives, recording, attribution, review, use, and withdrawal terms. 2. Research the organization, situation, implementation, available metrics, known problems, and sensitive boundaries. 3. Write open questions covering before state, decision, alternatives, implementation, effort, contribution, result, limitation, and next step. 4. Share scope and consent terms in advance without scripting answers. 5. At the start, reconfirm recording and attribution and explain how quotes and facts will be reviewed. 6. Ask for specific examples, time periods, definitions, sources, and counterexamples. 7. Distinguish participant observation, organizational records, inference, and opinion in the notes. 8. Probe negative outcomes, workarounds, dependencies, and what the product did not cause. 9. Summarize key claims back to the participant and list evidence or approvals still required. 10. Preserve the recording securely, prepare a source log, and route quotes and claims through agreed review. ## Guardrails - Never invent, polish beyond meaning, or pressure a participant to endorse a claim. - Protect customers, employees, end users, and confidential business information. - A gift or incentive must not be conditioned on positive statements. - Do not publish a metric without its definition, source, period, and approval. ## Done - Interview notes or recording preserve consent, context, attribution, and exact claims - A source log separates verified facts, participant statements, and open evidence - The participant receives the agreed factual and quotation review - Sensitive material and use limits are recorded
GitHubで見る