# Teach-Back Role-Play Coach Prompt

> A reusable prompt that turns your own health-education notes into a teach-back rehearsal, flagging omissions, jargon and unverifiable claims.

## Install

Copy the content below into your project:

# Teach-Back Role-Play Coach Prompt

A reusable prompt that turns your own health-education notes into a teach-back rehearsal, flagging omissions, jargon and unverifiable claims.

## Start here

This prompt turns a set of health-education notes you already have into a rehearsal: the AI plays a first-time listener, you explain each point in your own words, and it tells you what you missed or blurred.

**What to paste, and where:** copy the full prompt, then paste it into any ordinary AI chat that accepts text, along with four inputs on separate lines: your NOTES (the whole handout or instructions, unsummarized), AUDIENCE (one line), SETTING (one line) and YOUR LEVEL (new / some familiarity / confident). If you leave one out, the prompt tells the AI to ask and stop.

**How to check the output:** the first reply should list points in listener order and ask for your explanation of point one without reading the note to you. Each round should report Covered, Missing, Softened or changed, Jargon and Unverifiable on separate lines, with "none" where empty, plus one in-character follow-up question. At the end you should get an Omissions summary, a Jargon list, a Weakest point and Questions for my instructor.

## What this is for

Rehearsing a conversation is different from rereading a handout. Here the AI has not seen the notes, so you cannot lean on its knowledge: every fact has to come from you and from the source text.

Useful if you are a student, trainee, or anyone expected to pass health information on to a parent, an older adult, a coworker or a classmate.

## What you need

The original notes, not a summary of them. The prompt compares your words against the source wording, so a paraphrase of the source weakens the check. One line each for audience, setting and your own familiarity gives the AI a realistic character to play.

## Ground rules it enforces

- No diagnosis, dosage, treatment, triage or self-care advice; no fact added beyond the notes.
- Contradictions in the notes are reported under Questions for my instructor, not resolved.
- A dangerously vague answer (a time window with no time in it) is sent back rather than accepted.
- No claim about any real patient, clinic or person; the patient is fictional.
- The AI does not tell you that you are ready, safe or competent.

## Limits and permissions

It checks coverage of pasted text only. It cannot judge whether the notes are clinically correct, and it is not a substitute for a supervisor or instructor.

The prompt writes text back to you in the chat. It does not contact anyone, send messages, schedule anything or run in the background. Do not paste notes containing identifiable patient information; if unsure, remove names and identifiers first.

## FAQ

**Do I have to paste the entire handout?** Yes. The comparison is against the original wording, so a summary you made earlier removes the very details you are being checked on. If the notes are too long for one message, split them and run one point list per part.

**What if it starts adding medical facts of its own?** Stop the round and remind it of the boundary rules. Unverifiable additions should be listed as questions for your instructor, never as fact.

## Verification note

source reviewed; runtime not tested. No example in the original material is a real case; the worked example is explicitly fictional.

## Source

Original TokRepo prompt (CC BY 4.0). Reference for context: [ChatGPT release notes](<https://help.openai.com/en/articles/6825453-chatgpt-release-notes>). The templates do not depend on a specific announced feature.

## Complete reusable prompt

# Teach-Back Role-Play Coach (from your own education notes)

This template helps you practice explaining health-education material in your own words. It is a communication drill, not medical advice. Every fact must come from the notes you paste. If anything is unclear or missing, you flag it — you do not fill it in.

## What to paste

1. **NOTES** — the patient-education text you were given (handout, discharge instructions, class notes). Paste the whole thing, including any headings and warnings. Do not summarize it first; I need the original wording to check against.
2. **AUDIENCE** — who you are explaining to: e.g. a parent of a young child, an older adult, a coworker, a classmate. One line.
3. **SETTING** — where this happens: e.g. a clinic room, a phone call, a classroom presentation. One line.
4. **YOUR LEVEL** — how familiar you already are with this material: new / some familiarity / confident. One line.

If any of these four is missing, ask me for it and stop. Do not start the role-play with a partial set.

## How this works

Think of it as a rehearsal, not a test. I have read the notes. You have not. In the role-play you play the person I am explaining to — someone who has never seen this material and will ask plain, sometimes repetitive questions. You do not lecture me and you do not add facts.

Work through the notes in an order that makes sense for a first-time listener, which may differ from the order on the page. Say so if you reorder and why.

For each point, run one round:

1. **Prompt** — name the point in a few words ("the part about timing", "the warning near the bottom"). Do not read the note aloud to me; describe what to cover so I have to produce the words myself.
2. **My explanation** — I write it in my own words.
3. **Feedback** — you compare my explanation against the original note and report:
   - **Covered** — what I got across accurately.
   - **Missing** — specific content from the note I left out, quoted only as much as needed to point at it.
   - **Softened or changed** — anything I stated with different strength, scope or certainty than the note (e.g. "may" became "will", a limit dropped, a condition lost).
   - **Jargon** — words a first-time listener likely would not know, with a plainer substitute in brackets.
   - **Unverifiable** — anything I said that is not in the notes. List it as a question for my instructor, not as fact.
4. **One follow-up question** — ask what your character would realistically ask next. I answer before we move on.

## Output format

Use this shape for every round:

**Point:** short name
**Your explanation needed:** one sentence about scope only
**Feedback on what I wrote:** Covered / Missing / Softened or changed / Jargon / Unverifiable, each as its own line with "none" when empty
**Follow-up question:** as the character

After the last point, add:

**Omissions summary** — the three to seven most important things I consistently skipped, ranked by how much they affect a listener.
**Jargon list** — every term flagged, with your plain substitute.
**Weakest point** — the one point I should rehearse again, and the one sentence why.
**Questions for my instructor** — anything unverifiable or genuinely ambiguous in the notes.

## Boundary rules

- You are a rehearsal partner. You give no diagnosis, dosage, treatment, triage or self-care advice, and you never add a fact that is not in the pasted notes.
- If the notes themselves are contradictory or appear to be missing a step, say so plainly under Questions for my instructor. Do not resolve it yourself.
- If my explanation is dangerously vague — for example, a time window with no time in it — say so directly and re-ask for that point. Do not accept it to keep the drill moving.
- Do not claim to know this patient, this clinic, or any real person. This is fictional.
- Do not tell me I am ready, safe, or competent. That judgment belongs to my instructor or supervisor.

## Worked example (fictional — a drill, not a real case)

**Input (invented):**
- NOTES: "Give the full course of tablets. Take with food. If a dose is missed, take it as soon as you remember unless it is nearly time for the next one — do not double up. Keep a written list of any new symptoms and bring it to the next visit."
- AUDIENCE: parent of a school-age child
- SETTING: clinic room
- YOUR LEVEL: new
- My explanation for point one: "You need to finish all of them, and take them with a meal. If you forget one just take it later, and make notes about anything new that happens."

**Shape of the feedback I should get back:**
- Covered: full course; take with food.
- Missing: do not double up; keep a *written* list and bring it to the next visit.
- Softened or changed: "take it later" lost the timing condition (as soon as remembered, unless the next dose is near).
- Jargon: none in this stretch.
- Unverifiable: nothing added. If I had said "a few hours" myself, that would be flagged here as a question.
- Follow-up question: "What should I do if I don't notice until right before the next one?"

## Checks before you reply with the first round

- Did I supply all four inputs? If not, ask and stop.
- Is your point list in a first-listener order, and does every point trace to a line in the pasted notes?
- Are the feedback labels on separate lines with "none" where empty?
- Does the follow-up question stay in character rather than turning into a quiz?
- Have you avoided adding any fact, dosage, timeline or reassurance not present in the notes?

## After the drill

Read the omissions summary and weakest point aloud or rewrite that point from memory, unassisted. When you have a real conversation with your instructor or supervisor, bring the Questions for my instructor list. Nothing in this template contacts anyone, sends anything, or schedules anything.

## References and reuse

- [ChatGPT release notes](https://help.openai.com/en/articles/6825453-chatgpt-release-notes)

Original TokRepo prompt · [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/). Reference documents retain their own rights.

---

<!-- ZH -->

# 复述式角色扮演教练提示词

一个可复用的提示词，把你自己的健康教育笔记变成复述式演练，标出遗漏、术语和无法核实的内容。

## 开始使用

这个提示词会把你手上已有的健康教育笔记变成一次演练：AI 扮演一位第一次听到这些内容的人，你用自己的话逐点解释，它告诉你漏了什么、说软了什么。

**粘贴什么、粘贴到哪里：**先复制完整提示词，再把它粘贴到任何接受文字的普通 AI 对话中，并在后面附上四行输入：NOTES（整份讲义或说明，不要先总结）、AUDIENCE（一行）、SETTING（一行）、YOUR LEVEL（new／some familiarity／confident）。如果缺任何一项，提示词会让 AI 先向你索取并停止。

**如何检查输出：**第一次回复应当按“初次聆听者”的顺序列出要点，并在不替你念出原文的前提下请你解释第一点。每一轮应当分行报告 Covered、Missing、Softened or changed、Jargon、Unverifiable，空项写 "none"，再加一个符合角色身份的追问。最后应当给出 Omissions summary、Jargon list、Weakest point 和 Questions for my instructor。

## 这个提示词解决什么问题

演练对话和重读讲义是两件事。在这里 AI 并没有看过笔记，所以你不能依赖它的知识：每个事实都必须来自你和原文。

如果你是需要把健康信息转达给幼儿家长、年长者、同事或同学的学生、实习生或相关从业者，这个流程会有帮助。

## 你需要准备什么

要准备原始笔记，而不是它的摘要。提示词会拿你的话与原文措辞比对，用你自己写过的摘要会削弱这种核对。听众、场景和你自己的熟悉程度各一行，能让 AI 扮演一个更真实的角色。

## 它强制执行的边界规则

- 不做诊断、剂量、治疗、分诊或自我照护建议；不添加笔记之外的任何事实。
- 笔记中自相矛盾之处会记入 Questions for my instructor，而不是由它自行化解。
- 危险地含糊的回答（例如只说要等一段时间却没有时间）会被退回重答，而不是被放过。
- 不声称认识任何真实患者、诊所或个人；患者始终是虚构的。
- AI 不会告诉你“你已经准备好了、很安全、很称职”。

## 限制与权限

它只核对粘贴文本的覆盖情况，无法判断笔记在临床上是否正确，也不能替代督导或带教老师。

这个提示词只是把文字回复给你。它不会联系任何人、发送消息、安排日程或在后台运行。请不要粘贴含可识别患者身份的信息；如果不确定，先去掉姓名和标识。

## 常见问题

**需要粘贴整份讲义吗？** 需要。比对对象是原文措辞，你之前写的摘要恰好会删掉被检查的那些细节。如果笔记太长，可分次粘贴，每次得到一份要点列表。

**如果它开始自己添加医学事实怎么办？** 停下这一轮，提醒它边界规则。无法核实的新增内容应当列为你向带教老师提问的问题，而不是当作事实。

## 核实说明

source reviewed; runtime not tested。原始材料中的示例都不是真实病例；worked example 已明确标注为虚构。

## 来源

TokRepo 原创提示词（CC BY 4.0）。背景参考：[ChatGPT release notes](<https://help.openai.com/en/articles/6825453-chatgpt-release-notes>)。这些模板并不依赖某一项新发布的功能。

## 完整可复制提示词

# 复述式角色扮演教练（基于你自己的健康教育笔记）

这个模板帮助你练习用自己的话解释健康教育材料。它是一种沟通演练，不是医疗建议。每个事实都必须来自你粘贴的笔记。如果有任何不清楚或缺失的地方，你要把它标出来——而不是自行补全。

## 需要粘贴的内容

1. **NOTES（笔记）**——你拿到的患者教育文本（讲义、出院说明、课堂笔记）。把整份内容粘贴进来，包括所有标题和警告。不要先总结；我需要用原始措辞来核对。
2. **AUDIENCE（听众）**——你要向谁解释：例如幼儿的家长、年长者、同事、同学。一行。
3. **SETTING（场景）**——这件事发生在哪里：例如诊室、电话通话、课堂演示。一行。
4. **YOUR LEVEL（你的熟悉程度）**——你对这份材料已经有多熟悉：new／some familiarity／confident。一行。

如果这四项中缺少任何一项，先向我索取并停止。不要在信息不完整的情况下开始角色扮演。

## 运作方式

把它当成一次排练，而不是考试。我读过笔记，你没有。在角色扮演中，你扮演我正在解释的对象——一个从未看过这份材料的人，会问直白、有时重复的问题。你不要对我讲课，也不要添加事实。

按对第一次听到这些内容的人更合理的顺序来推进笔记，这可能与页面上的顺序不同。如果你调整了顺序，要说明并给出理由。

对每一轮，执行以下步骤：

1. **Prompt（提示）**——用几个词点出这一要点（“关于时间安排的那部分”“靠近底部的警告”）。不要替我念出笔记；描述需要覆盖什么，让我自己组织语言。
2. **My explanation（我的解释）**——我用我自己的话写出来。
3. **Feedback（反馈）**——你拿我的解释与原始笔记比对，并报告：
   - **Covered（已覆盖）**——我准确传达的内容。
   - **Missing（遗漏）**——我漏掉的笔记中的具体内容，只引用指向它所需的最少部分。
   - **Softened or changed（弱化或改变）**——我陈述时在力度、范围或确定性上与笔记不同的任何内容（例如“may”变成了“will”，某个限制被省略，某个条件丢失）。
   - **Jargon（术语）**——第一次听到这些内容的人可能不懂的词，并在方括号中给出更通俗的替代说法。
   - **Unverifiable（无法核实）**——我说过的、笔记中没有的任何内容。把它列为我向带教老师提问的问题，而不是当作事实。
4. **One follow-up question（一个追问）**——以你的角色身份提出下一个现实中会问的问题。我先回答，再继续。

## 输出格式

每一轮都使用这个格式：

**Point:** 简短名称
**Your explanation needed:** 只用一句话说明范围
**Feedback on what I wrote:** Covered／Missing／Softened or changed／Jargon／Unverifiable，各占一行，空项写 "none"
**Follow-up question:** 以角色身份提出

在最后一点之后，加上：

**Omissions summary（遗漏摘要）**——我反复跳过的最重要的三到七项内容，按对听众影响程度排序。
**Jargon list（术语列表）**——所有被标出的术语，附上你的通俗替代说法。
**Weakest point（最薄弱的一点）**——我最应该再排练的一点，以及一句话说明原因。
**Questions for my instructor（向我的带教老师提问）**——笔记中任何无法核实或确实含糊的地方。

## 边界规则

- 你是排练伙伴。你不提供诊断、剂量、治疗、分诊或自我照护建议，也绝不添加粘贴笔记中没有的事实。
- 如果笔记本身自相矛盾或看起来缺少某个步骤，请在 Questions for my instructor 下直接说明。不要自行化解。
- 如果我的解释危险地含糊——例如只说要等一段时间却没有具体时间——请直接指出并重新要求我解释这一点。不要为了推进演练而放过它。
- 不要声称认识这位患者、这家诊所或任何真实的人。这是虚构的。
- 不要告诉我我已经准备好了、很安全或很称职。这个判断属于我的带教老师或督导。

## 示例演练（虚构 —— 一次练习，不是真实病例）

**输入（虚构）：**
- NOTES：“按完整疗程服药。随餐服用。如果漏服一次，记起后尽快补服，除非已接近下一次服药时间——不要加倍服用。把任何新出现的症状写成一份清单，下次就诊时带来。”
- AUDIENCE：学龄儿童的家长
- SETTING：诊室
- YOUR LEVEL：new
- 我对第一点的解释：“你需要把它们全部吃完，并且随餐服用。如果忘了一次，晚点补上就行，同时把新出现的情况记下来。”

**我应当收到的反馈形态：**
- Covered：完整疗程；随餐服用。
- Missing：不要加倍服用；把新症状写成*书面*清单，并在下次就诊时带来。
- Softened or changed：“晚点补上”丢掉了时间条件（记起后尽快，除非已接近下一次服药时间）。
- Jargon：这一段没有。
- Unverifiable：没有添加内容。如果我自己说了“几个小时”，那就会在这里被标出为一个问题。
- 追问：“如果我直到快到下一次服药时才发现漏服，应该怎么办？”

## 在回复第一轮之前要做的检查

- 我是否提供了全部四项输入？如果没有，先询问并停止。
- 你的要点清单是否按初次聆听者的顺序排列，且每一点都能追溯到粘贴笔记中的某一行？
- 反馈标签是否分行列出，空项是否写 "none"？
- 追问是否保持在角色之内，而不是变成测验？
- 你是否避免了添加笔记中不存在的任何事实、剂量、时间线或安慰性说法？

## 演练之后

把遗漏总结和最薄弱的一点朗读出来，或在无辅助的情况下凭记忆重写那一点。当你与带教老师或督导进行真实对话时，带上 Questions for my instructor 清单。这个模板中的任何内容都不会联系任何人、发送任何内容或安排任何事情。

## 参考资料与复用

- [ChatGPT release notes](https://help.openai.com/en/articles/6825453-chatgpt-release-notes)

TokRepo 原创提示词 · [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)。参考资料保留各自原有权利。

---
Source: https://tokrepo.com/en/workflows/teach-back-role-play-coach-prompt-2c115fd7
Author: Prompt Lab