# Shift-Change Log to Structured Handover Prompt

> A reusable prompt that turns a messy pasted shift log into a grouped handover draft with safety items, pending work, conflicts and open questions.

## Install

Copy the content below into your project:

# Shift-Change Log to Structured Handover Prompt

A reusable prompt that turns a messy pasted shift log into a grouped handover draft with safety items, pending work, conflicts and open questions.

## Start here

Paste your raw shift log where the prompt says `[paste raw shift log here]`, then optionally add shift, site, next shift start time and who you hand over to. Paste the whole prompt into any ordinary AI chat that accepts text. No terminal or account setup is needed.

Check the draft yourself: every log line should appear as an item or an Open Question, each item should carry a short quoted fragment from your log, and any mention of injury, damage, near-miss, rule breach or hazard should appear under SAFETY / HOLD. If your log has no such mention, the draft should say "No safety items identified".

## What this prompt does

It reads only the text you paste and produces a draft handover, not a sent message. Items are sorted into SAFETY / HOLD, PENDING, DONE and FYI, with conflicts shown side by side and gaps listed as Open Questions. It never invents owners, times, locations or causes.

## Inputs you need

- The raw log text, pasted as-is. Shorthand, typos and out-of-order notes are fine.
- Optional context: shift and site, when the next shift starts, and names or roles receiving the handover.
- If a field is unknown, say so; the prompt asks the assistant to mark it rather than guess.

## Permissions and limits

The assistant cannot reach your systems, records or messaging tools and sends nothing. It works on pasted text only, so it cannot verify counts, close tickets or confirm that a safety issue was escalated. A human must review, correct and actually send the handover. Output quality depends on how much of the log you paste.

## Checking the output

| Check | What to look for |
| --- | --- |
| Coverage | Every log line maps to an item or an Open Question |
| Traceability | Each item has a short verbatim fragment or an [inferred] tag |
| Safety | Hazards are in SAFETY / HOLD, never softened |
| Conflicts | Shown side by side with no silent resolution |

The worked example inside the prompt is explicitly fictional and is only a shape reference, not a test result.

## FAQ

**Can it retrieve my actual shift records?** No. It only uses what you paste; nothing is fetched, and it cannot send the handover.
**What if my log is very short or missing details?** Items stay as written and missing owner, time or location appear as Open Questions instead of guesses.

## Verification note

source reviewed; runtime not tested.

## Source and thanks

Original TokRepo prompt, licensed CC BY 4.0. Reference context: [ChatGPT release notes](<https://help.openai.com/en/articles/6825453-chatgpt-release-notes>), reviewed 2026-10-04.

## Complete reusable prompt

You are helping me turn a messy shift-change log into a clean, reviewable handover for the next person or team. You do not have access to my systems, records or messaging tools. You only work from what I paste below, and you prepare a draft for a human to review and send.

INPUT I WILL PROVIDE
I will paste a raw shift log. It may be shorthand, out of order, contain typos, or mix facts with opinions. If a field is missing, say so rather than guessing.
Log:
[paste raw shift log here]

Also tell me, if I know them:
- Shift and site: [e.g. Night shift, Warehouse B]
- Next shift starts: [time or unknown]
- People to hand over to: [names or roles, or unknown]

TASK
1. Read the log and extract every distinct item as a separate entry. Do not merge or drop items, even minor ones.
2. Classify each item into exactly one group:
   - SAFETY / HOLD: anything that could injure someone, damage equipment or product, breach a rule, or that a next-shift person must not ignore.
   - PENDING: work started but not finished, or promised but not yet done.
   - DONE: completed, needing no handover action beyond awareness.
   - FYI: context with no action required.
3. For each item, quote the original log words (short verbatim fragment) so the reader can trace it back. If the item is inferred rather than stated, mark it [inferred] and keep the inference minimal.
4. Flag conflicts: if two entries disagree (times, counts, status), list both side by side and mark [conflict]. Do not silently pick one.
5. Flag gaps: if the next shift would clearly need information the log never provides (owner, location, time, next step), list it under Open Questions.

OUTPUT FORMAT
Handover draft - [shift] - [site] - [date if given]

1. SAFETY / HOLD - do not proceed without checking
   - Item: ... | Source: "..." | Status: ... | Owner (if stated): ...

2. PENDING - carry over
   - Item: ... | Source: "..." | What remains: ... | Owner (if stated): ... | Blocker: ...

3. DONE - awareness only
   - Item: ... | Source: "..."

4. FYI
   - Item: ... | Source: "..."

5. CONFLICTS
   - A: ... / B: ... | Which is which: ... | Needs human decision: yes

6. OPEN QUESTIONS for the next shift or the log author
   - ...

7. ONE-LINE SUMMARY
   - The single most important thing the next shift must know: ...

RULES
- Never invent an owner, time, location, cause or outcome that is not in the log.
- Never soften a safety item. If the log hints at risk, put it in SAFETY / HOLD and let a human downgrade it.
- Keep each item to one or two lines. No advice, no blame, no speculation about people.
- Do not tell me to contact anyone and do not claim you sent anything. You produce a draft only.

CHECK BEFORE YOU FINISH
- Did every log line map to at least one item or Open Question?
- Does every item have a short verbatim source fragment, or an [inferred] tag?
- Is at least one item in SAFETY / HOLD if the log mentions any injury, damage, near-miss, rule breach or hazard? If none, say 'No safety items identified' explicitly.
- Are all conflicts shown side by side with no silent resolution?

WORKED EXAMPLE (fictional)
Input log:
'2300 forklift 3 steering loose told Sam. aisle 7 pallet leaning. 0130 finished cycle count zone A 148/150 - two missing, think miscounted. 0200 break. told night supe about forklift, he said note it. restocked gloves. drowsy around 0300.'

Expected output shape:
1. SAFETY / HOLD
- Forklift 3 steering loose | Source: 'forklift 3 steering loose' | Owner: reported to Sam, night supe asked to note it | Status: unresolved, do not operate until checked.
- Leaning pallet in aisle 7 | Source: 'aisle 7 pallet leaning' | Owner: not stated | Status: unresolved.
2. PENDING
- Cycle count zone A | Source: '148/150 - two missing, think miscounted' | What remains: confirm two missing items | Blocker: possible miscount.
3. DONE
- Gloves restocked. 4. FYI: fatigue noted around 0300.
5. CONFLICTS: none stated, but 'think miscounted' is unverified - flag as [inferred].
6. OPEN QUESTIONS: Was the foreskinned loose steering escalated to a maintenance ticket? Who verifies the two missing items?
7. ONE-LINE SUMMARY: Do not operate forklift 3 until steering is checked.

Now wait for my log and produce the handover draft.

## 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 -->

# 把交班日志整理成结构化交接单的提示词

一段可复用的提示词，把粘贴进来的杂乱交班日志整理成分组交接草稿，突出安全事项、待办、冲突与待确认问题。

## 开始使用

把原始交班日志粘贴到提示词中 `[paste raw shift log here]` 的位置，也可以补充班次、地点、下一班开始时间以及交接对象。把整段提示词粘贴进任何接受文字的普通 AI 对话即可，无需终端或账号配置。

自己核对草稿：每条日志都应变成一条事项或一个待确认问题；每条事项应带一小段日志原文摘录；只要日志提到受伤、损坏、未遂事件、违规或隐患，就应出现在 SAFETY / HOLD 中。若日志没有这类内容，草稿应明确写出“未发现安全事项”。

## 这个提示词做什么

它只读取你粘贴的文字，产出一份交接草稿，而不是已发送的消息。事项被归入 SAFETY / HOLD、PENDING、DONE 和 FYI，冲突并列展示，信息缺口列为待确认问题。它不会编造责任人、时间、地点或原因。

## 需要准备的输入

- 原始日志文本，按原样粘贴；缩写、错字、顺序混乱都可以。
- 可选背景：班次与地点、下一班开始时间、接手人的姓名或岗位。
- 不知道的字段直接说明，提示词要求标记而不是猜测。

## 权限与限制

助手无法访问你的系统、记录或通讯工具，也不会发送任何内容，只处理粘贴的文字，因此无法核实数量、关闭工单或确认安全隐患是否已上报。必须由人工审核、修改并真正发出交接信息。输出效果取决于你粘贴了多少日志内容。

## 检查输出

| 检查项 | 关注点 |
| --- | --- |
| 覆盖 | 每条日志都对应一条事项或一个待确认问题 |
| 可追溯 | 每条事项都有短原文摘录或 [inferred] 标记 |
| 安全 | 隐患放在 SAFETY / HOLD，不被淡化 |
| 冲突 | 并列展示，不做无声取舍 |

提示词中的示例已注明是虚构的，仅用于示意格式，不是测试结果。

## 常见问题

**它能调取我真实的交班记录吗？** 不能。它只使用你粘贴的内容，不抓取任何数据，也无法发送交接信息。
**日志很短或缺少细节怎么办？** 事项保持原样，缺失的责任人、时间或地点会列入待确认问题，而不是被猜测。

## 验证说明

源码已审阅；未做运行时测试。

## 来源与致谢

TokRepo 原创提示词，采用 CC BY 4.0 许可。参考背景：[ChatGPT release notes](<https://help.openai.com/en/articles/6825453-chatgpt-release-notes>)，查阅于 2026-10-04。

## 完整可复制提示词

你正在帮我把一份混乱的交班日志整理成一份干净、可复核的交接材料，供下一位或下一班人员使用。你无法访问我的系统、记录或通讯工具。你只能依据我在下面粘贴的内容工作，并准备一份草稿供人工审核并发送。

我将提供的输入
我会粘贴一份原始交班日志。它可能是缩写、顺序混乱、包含错字，或把事实和意见混在一起。如果某字段缺失，请说明缺失，而不是猜测。
日志：
[paste raw shift log here]

如果我已知，也请告诉我：
- 班次与地点：[例如 Night shift, Warehouse B]
- 下一班开始时间：[时间或未知]
- 交接对象：[姓名或岗位，或未知]

任务
1. 阅读日志，把每一条独立事项提取为单独条目。不要合并或遗漏事项，即使是小事也不要。
2. 将每条事项归入且仅归入一组：
   - SAFETY / HOLD：任何可能使人受伤、损坏设备或产品、违反规则，或下一班人员绝不能忽略的事项。
   - PENDING：已开始但未完成，或已承诺但尚未完成的工作。
   - DONE：已完成，除知晓外无需交接动作。
   - FYI：无需动作的背景信息。
3. 对每条事项，引用原始日志文字（短的逐字片段），以便读者回溯。如果该事项是推断而非明示，请标记 [inferred]，并保持推断最小化。
4. 标出冲突：如果两条条目不一致（时间、数量、状态），并列列出两者并标记 [conflict]。不要无声地选择其中一个。
5. 标出缺口：如果下一班明显需要日志从未提供的信息（责任人、地点、时间、下一步），将其列在 Open Questions 下。

输出格式
Handover draft - [shift] - [site] - [date if given]

1. SAFETY / HOLD - do not proceed without checking
   - Item: ... | Source: "..." | Status: ... | Owner (if stated): ...

2. PENDING - carry over
   - Item: ... | Source: "..." | What remains: ... | Owner (if stated): ... | Blocker: ...

3. DONE - awareness only
   - Item: ... | Source: "..."

4. FYI
   - Item: ... | Source: "..."

5. CONFLICTS
   - A: ... / B: ... | Which is which: ... | Needs human decision: yes

6. OPEN QUESTIONS for the next shift or the log author
   - ...

7. ONE-LINE SUMMARY
   - The single most important thing the next shift must know: ...

规则
- 绝不要编造日志中没有的责任人、时间、地点、原因或结果。
- 绝不要淡化安全事项。如果日志暗示风险，就把它放入 SAFETY / HOLD，让人类来决定是否降级。
- 每条事项保持一到两行。不要给建议、不要归责、不要对人进行推测。
- 不要告诉我联系任何人，也不要声称你发送了任何内容。你只产出一份草稿。

完成前检查
- 每条日志是否至少对应一条事项或一个待确认问题？
- 每条事项是否都有短的逐字来源片段，或 [inferred] 标记？
- 如果日志提到任何受伤、损坏、未遂事件、违规或隐患，是否至少有一条事项在 SAFETY / HOLD 中？如果没有，明确写出 'No safety items identified'。
- 所有冲突是否都并列展示，且没有无声取舍？

示例（虚构）
输入日志：
'2300 forklift 3 steering loose told Sam. aisle 7 pallet leaning. 0130 finished cycle count zone A 148/150 - two missing, think miscounted. 0200 break. told night supe about forklift, he said note it. restocked gloves. drowsy around 0300.'

预期输出形态：
1. SAFETY / HOLD
- Forklift 3 steering loose | Source: 'forklift 3 steering loose' | Owner: reported to Sam, night supe asked to note it | Status: unresolved, do not operate until checked.
- Leaning pallet in aisle 7 | Source: 'aisle 7 pallet leaning' | Owner: not stated | Status: unresolved.
2. PENDING
- Cycle count zone A | Source: '148/150 - two missing, think miscounted' | What remains: confirm two missing items | Blocker: possible miscount.
3. DONE
- Gloves restocked. 4. FYI: fatigue noted around 0300.
5. CONFLICTS: none stated, but 'think miscounted' is unverified - flag as [inferred].
6. OPEN QUESTIONS: Was the foreskinned loose steering escalated to a maintenance ticket? Who verifies the two missing items?
7. ONE-LINE SUMMARY: Do not operate forklift 3 until steering is checked.

现在等待我的日志，并生成交接草稿。

## 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/shift-change-log-structured-handover-prompt-272a0564
Author: Prompt Lab