# Peer-Review Prompt Sheet Builder

> Turn an assignment brief into a reviewer-ready prompt sheet with criteria, questions and sentence starters traceable to the brief.

## Install

Copy the content below into your project:

# Peer-Review Prompt Sheet Builder

Turn an assignment brief into a reviewer-ready prompt sheet with criteria, questions and sentence starters traceable to the brief.

## Start here
Paste your text into an ordinary AI chat that accepts text. Copy the full original prompt (appended below), then add three parts in this order:

1. **ASSIGNMENT BRIEF** — the complete text of the brief: task, learning goals, constraints, rubric, format, word or time limits.
2. **REVIEWER CONTEXT** (optional) — who the reviewers are, how long they have, whether their comments count toward a grade.
3. **REVIEW FORMAT** — written comments, a live session, a template, or your choice.

If the brief is missing or contradictory, the prompt stops and asks for what is needed. Run it, then check the output with the Review checks below.

## What this prompt is for
Peer feedback often turns into vague praise. This original TokRepo prompt builds a compact sheet: brief snapshot, criteria table, reviewer questions, fill-in sentence starters, a session plan and a gaps list. Every criterion points back to a brief section or is labelled [inferred - verify].

## Checking the output
- **Traceability** — each criterion row names a brief section or carries the inferred label.
- **No invented standards** — no points, weights, deadlines or policies beyond the brief.
- **Blanks** — every sentence starter has something to fill in.
- **Length** — 10–20 questions; sheet fits two pages.
- **Boundary** — the sheet never claims to submit, share or change a file.

## FAQ
**What if my brief has no rubric?** The prompt adds generic criteria like clarity or structure, labels each [inferred - verify], and lists the undefined points as neutral questions for the author or instructor.

**Is my brief kept private?** The prompt only prepares comments in the chat. Treat drafts as you would any course material and review anything sensitive before pasting.

## Limits and attribution
Source reviewed; runtime not tested. Original TokRepo prompt, CC BY 4.0; the linked reference retains its own rights. The fictional 900-word example shows the expected output shape only — it is not a test result.

Source: [ChatGPT release notes](<https://help.openai.com/en/articles/6825453-chatgpt-release-notes>) (reviewed 2026-10-04).

## Complete reusable prompt

# Peer-Review Prompt Sheet Builder

You are helping me turn an assignment brief into a peer-review prompt sheet that real reviewers can use to give specific, useful feedback. I will paste the assignment brief. You will produce criteria, questions and sentence starters that are traceable to that brief. Do not invent requirements, point values, deadline rules or grading standards that the brief does not state.

## My input
Paste, in this order:
1. ASSIGNMENT BRIEF: the full text of the brief (task description, learning goals, constraints, rubric, submission format, word or time limits).
2. REVIEWER CONTEXT (optional): who the reviewers are (classmates, colleagues, beginners), how long they have, and whether the final grade uses their comments.
3. REVIEW FORMAT: written comments, a live session, a template to fill in, or my choice.

If the brief is missing, contradictory or too vague to derive criteria, stop and list exactly what I must supply or clarify before you continue.

## What to produce

### 1. Brief snapshot
In 3-6 bullet points, restate only facts the brief states: deliverable, purpose, audience, limits and any stated evaluation criteria. Mark each bullet [from brief] or [inferred - verify]. Put nothing in this section that the brief does not support.

### 2. Criteria table
One row per review criterion. Columns:
- Criterion (plain-language name)
- What it means in this assignment (one sentence, tied to the brief)
- Look-for evidence (what a reviewer can point to in the work)
- Common miss (the gap reviewers usually overlook)
- Source (brief section or [inferred - verify])

Derive 4-8 criteria. Prefer criteria the brief actually names. If the brief names fewer than four, you may add generic ones such as clarity, structure or evidence, but label each as [inferred - verify] and place them after the brief-based rows. Never assign weights or point values unless the brief states them.

### 3. Reviewer question set
For each criterion, write 2-3 open questions a reviewer can ask. Questions must invite description of what is on the page, not taste judgments. Avoid yes/no questions. Keep the full set to roughly 10-20 questions so it stays usable in one review session.

### 4. Sentence starters
For each criterion, give 2 fill-in-the-blank starters, one for a strength and one for a gap. Cover these four moves across the set:
- Point to a specific place: "In the paragraph where ..., I noticed ..."
- Name the effect on me: "As a reader, this made me ..."
- Offer a next step: "One way to strengthen this would be ..."
- Ask a real question: "What were you trying to show when ...?"
Every starter must include a blank or bracket the reviewer fills from the actual work. No starter may praise or criticize without pointing to a location.

### 5. Review session plan
A short sequence for one review: read the work once without comments, note reactions, then work criterion by criterion; aim for at least two strengths and two gaps; end with one prioritized next step chosen by the author, not the reviewer. State a suggested time split and a line for reviewers to record how long drafting and reviewing actually took, if they want to.

### 6. Feedback hygiene
A short do/don't list: comment on the work not the person; quote or paraphrase the exact passage; separate what the brief requires from personal preference; do not rewrite the whole piece; flag uncertainty rather than guessing intent. Note clearly that reviewers should not change the author's file, and that this sheet only prepares comments - it does not submit, share or send anything.

### 7. Gaps and uncertainties
List anything the brief leaves undefined (for example, whether citations are required, how sources count, whether the format is flexible). For each gap, give one neutral question the reviewer can ask the author or instructor instead of assuming an answer.

## Output style
Use Markdown with headings and tables. Keep the whole sheet compact enough to print on two pages. No flattery, no filler, no bullet that merely repeats another. Do not write the author's feedback for them; produce the tool reviewers use.

## Review checks before I use the sheet
Run these checks yourself and report pass or fail for each, with the fix if it fails:
1. Traceability: every criterion row names a brief section or is labelled [inferred - verify].
2. No invented standards: no points, weights, deadlines or policies appear that the brief did not state.
3. Blank requirement: every sentence starter contains a fill-in blank or bracket.
4. Length: the question set is 10-20 questions; the whole sheet fits the two-page target.
5. Action clarity: the session plan tells a reviewer what to do first, second and last.
6. Boundary: the sheet never claims to submit, share, schedule or modify anything.

## Worked example
FICTIONAL INPUT: Assignment brief for a 900-word community proposal: "Describe a local problem, propose one solution, use at least two sources, address one objection." Reviewer context: first-year students, 20 minutes each.
Expected output shape: Brief snapshot bullets (deliverable, length, source minimum, objection requirement, all [from brief]); a criteria table row for source use with look-for evidence "author names and applies each source," common miss "source quoted but not connected to the proposal," source "brief: use at least two sources"; questions such as "Which sentence shows how your first source shaped the proposal?"; starters such as "In the section where you describe the problem, I noticed [specific detail], which made me wonder ..."; a 20-minute plan (5 read, 10 comment, 5 discuss); a gap row noting the brief does not say what citation style is expected, with a neutral question to the instructor.
PASS/FAIL CHECKS ON THE EXAMPLE: Traceability passes only if the source-use row cites the brief line and the citation-style gap is listed as undefined rather than guessed. The blank check fails if any starter is a finished compliment like "great use of sources" with nothing to fill in. The no-invented-standards check fails if the sheet adds a point value for sources that the brief never gave. The length check fails if the questions expand past 20 or the sheet sprawls beyond two pages.

Start by asking for the assignment brief if I have not pasted it, then produce the sheet in the order above.

## References and reuse

- [ChatGPT release notes](https://help.openai.com/en/articles/6825453-chatgpt-release-notes) · Reviewed 2026-10-04

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

---

<!-- ZH -->

# 同行评审提示单生成器

把作业说明转成评审者可直接使用的提示单，包含可追溯到说明的评审标准、提问与句式开头。

## 开始使用
把文本粘贴到任何能接收文本的普通 AI 对话中。复制下方完整原始提示词，然后按此顺序补充三部分：

1. **作业说明（ASSIGNMENT BRIEF）** — 说明的完整文本：任务、学习目标、限制条件、评分表、提交格式、字数或时长限制。
2. **评审者背景（REVIEWER CONTEXT，可选）** — 评审者是谁、有多少时间、其意见是否计入成绩。
3. **评审形式（REVIEW FORMAT）** — 书面评论、现场交流、填写模板，或由你决定。

如果说明缺失或自相矛盾，提示词会停下来，列出需要你补充或澄清的内容。运行后，用下方“输出检查”核对结果。

## 这个提示词做什么
同伴反馈常沦为笼统的夸奖。这份 TokRepo 原创提示词生成一张紧凑的提示单：说明速览、标准表、评审提问、填空式句式开头、评审流程与未定义事项清单。每条标准都指回说明中的某一部分，或标注 [inferred - verify]。

## 输出检查
- **可追溯** — 每条标准注明说明中的出处，或带有推断标签。
- **不臆造标准** — 不出现说明未提及的分数、权重、截止日期或政策。
- **留空** — 每个句式开头都有需要填写的内容。
- **篇幅** — 提问 10–20 个；提示单不超过两页。
- **边界** — 提示单不会声称提交、分享或改动任何文件。

## 常见问题
**我的作业说明没有评分表怎么办？** 提示词会补充清晰度、结构等通用标准，逐条标注 [inferred - verify]，并把未定义之处整理成向作者或教师提出的中性问题。

**我的说明会被保密吗？** 该提示词只在对话中准备评论。请按对待课业材料的方式对待草稿，粘贴前检查敏感内容。

## 限制与来源
来源已审阅；未做运行测试。原创 TokRepo 提示词，CC BY 4.0；所引参考链接保留其自身权利。虚构的 900 字示例仅展示预期输出形态，并非测试结果。

来源：[ChatGPT 发布说明](<https://help.openai.com/en/articles/6825453-chatgpt-release-notes>)（审阅于 2026-10-04）。

## 完整可复制提示词

# 同伴评审提示单生成器

你正在帮助我把一份作业说明转化为一张同伴评审提示单，让真实的评审者能据此给出具体、有用的反馈。我会粘贴作业说明。你要生成可追溯到该说明的标准、提问和句式开头。不要臆造说明中未声明的要求、分值、截止日期规则或评分标准。

## 我的输入
按此顺序粘贴：
1. 作业说明（ASSIGNMENT BRIEF）：说明的完整文本（任务描述、学习目标、限制条件、评分表、提交格式、字数或时长限制）。
2. 评审者背景（REVIEWER CONTEXT，可选）：评审者是谁（同学、同事、初学者）、有多少时间，以及最终成绩是否采用其评论。
3. 评审形式（REVIEW FORMAT）：书面评论、现场交流、填写模板，或由我决定。

如果说明缺失、自相矛盾或过于笼统以致无法推导标准，请停下来，准确列出在你继续之前我必须补充或澄清的内容。

## 需要生成的内容

### 1. 说明速览
用 3-6 个要点，仅复述说明中声明的事实：交付物、目的、受众、限制条件及任何已声明的评价标准。为每个要点标注 [from brief] 或 [inferred - verify]。本节不得写入说明未支持的任何内容。

### 2. 标准表
每条评审标准一行。列为：
- 标准（通俗名称）
- 在本作业中的含义（一句话，与说明挂钩）
- 可寻找的证据（评审者能在作品中指出的内容）
- 常见遗漏（评审者通常忽略的缺口）
- 来源（说明中的某一部分，或 [inferred - verify]）

推导出 4-8 条标准。优先采用说明实际提到的标准。如果说明提到的少于四条，你可以补充清晰度、结构或证据等通用标准，但每条都要标注 [inferred - verify]，并放在基于说明的行之后。除非说明已声明，否则绝不要赋予权重或分值。

### 3. 评审者提问集
针对每条标准，写 2-3 个评审者可提的开放式问题。问题必须邀请对页面上实际内容的描述，而非品味判断。避免是非题。整套问题保持在约 10-20 个，以便在一次评审中就能用得上。

### 4. 句式开头
针对每条标准，给出 2 个填空式开头，一个用于优点，一个用于缺口。整套内容要覆盖以下四种动作：
- 指向具体位置：“在……的那一段中，我注意到……”
- 说出对我的影响：“作为读者，这让我……”
- 提供下一步：“加强这一点的一种方式是……”
- 提出真实问题：“当……时，你试图展示什么？”
每个开头都必须包含一个空或方括号，由评审者根据实际作品填写。任何开头都不得在未指向位置的情况下进行夸奖或批评。

### 5. 评审流程
一次评审的简短步骤：先不带评论地读一遍作品，记下反应，然后逐条标准进行；力求至少两条优点和两条缺口；最后以作者（而非评审者）选定的一项优先下一步作结。给出建议的时间分配，以及一行供评审者在愿意时记录起草和评审实际花费的时间。

### 6. 反馈规范
一份简短的该做/不该做清单：评论作品而非人；引用或转述确切段落；把说明要求与个人偏好区分开；不要重写整篇作品；标记不确定之处而非猜测意图。明确指出评审者不应改动作者的文件，并说明本提示单仅用于准备评论——它不会提交、分享或发送任何东西。

### 7. 未定义事项与不确定之处
列出说明中所有未定义的内容（例如，是否要求引用、来源如何计数、格式是否灵活）。对于每一处未定义事项，给出一条中性问题，供评审者向作者或教师提出，而不是自行假定答案。

## 输出风格
使用 Markdown，包含标题和表格。整张提示单要足够紧凑，可打印在两页内。不要奉承，不要填充内容，不要有仅重复上一条的要点。不要替作者写反馈；产出评审者使用的工具。

## 使用提示单前由我执行的输出检查
请自行运行这些检查，逐项报告通过或未通过，未通过时给出修正：
1. 可追溯：每条标准行都注明说明中的出处，或标注 [inferred - verify]。
2. 不臆造标准：不出现说明未陈述的分数、权重、截止日期或政策。
3. 留空要求：每个句式开头都包含需要填写的空位或方括号。
4. 篇幅：提问集为 10–20 个问题；整张提示单符合两页目标。
5. 行动清晰度：评审流程告诉评审者先做什么、再做什么、最后做什么。
6. 边界：提示单不会声称提交、分享、安排或改动任何内容。

## 示例
虚构输入：一份 900 字社区提案的作业说明：“描述一个本地问题，提出一个解决方案，至少使用两个来源，回应一个反对意见。”评审者背景：一年级学生，每人 20 分钟。
预期输出形态：说明速览要点（交付物、字数、来源数量下限、反对意见要求，均标注 [from brief]）；标准表中一行关于来源使用，证据看点是“作者说出并应用了每个来源”，常见遗漏是“来源被引用但未与提案建立联系”，出处为“说明：至少使用两个来源”；问题例如“哪句话显示你的第一个来源如何影响了这项提案？”；句式开头例如“在你描述问题的那一部分，我注意到 [具体细节]，这让我想问……”；一份 20 分钟流程（5 分钟阅读，10 分钟评论，5 分钟讨论）；一行未定义事项，指出说明没有写预期使用哪种引用格式，并附一条向教师提出的中性问题。
对示例执行通过/未通过检查：只有当来源使用行引用说明原句，且引用格式这一未定义事项被列为未定义而非猜测时，可追溯才通过。如果有任何句式开头是“很好地使用了来源”这类没有可填写内容的成品式夸奖，留空检查就不通过。如果提示单为来源添加了说明从未给出的分值，不臆造标准检查就不通过。如果问题扩展到 20 个以上，或提示单超出两页，篇幅检查就不通过。

如果我没有粘贴作业说明，先向我索取，然后按上述顺序生成提示单。

## 参考资料与复用

- [ChatGPT release notes](https://help.openai.com/en/articles/6825453-chatgpt-release-notes) · Reviewed 2026-10-04

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/peer-review-prompt-sheet-builder-3e730f44
Author: Prompt Lab