Show a worked continuity trail

README.md

@@ -91,6 +91,8 @@ A session enters, reads the relevant route, contributes what it needs to contrib
 
 A typical trail is **packet → response → review → decision**.
 
+[See a fictional source-to-decision trail, including a corrected AI claim](docs/WORKED_CONTINUITY_TRAIL.md).
+
 Once LabNote is set up, you can say things like:
 
 ```text

docs/README.md

@@ -22,6 +22,8 @@ one.
   normal piece of work in the right route with the minimum useful record.
 - [Understand review and decisions](review_workflow.md) — see how a response
   becomes a reviewed outcome.
+- [See a fictional continuity trail](WORKED_CONTINUITY_TRAIL.md) — follow a
+  source through response, correction, decision and next action.
 - [Understand visitor and session identity](visitor_lobby_model.md) — learn
   what session records mean in a live workspace.
 - [Understand message routing](message_routing_model.md) — follow the model

docs/WORKED_CONTINUITY_TRAIL.md

@@ -0,0 +1,101 @@
+# A worked continuity trail
+
+This fictional example shows the smallest useful shape of a LabNote trail. It
+does not represent a live project or a deposit in this public source
+repository.
+
+## The question
+
+A small team is choosing a durable, visible colour for a new pedestrian bridge.
+They want the next person or AI to understand the evidence, what was proposed,
+what was challenged and what still needs doing.
+
+They do **not** need every chat message preserved.
+
+## 1. Keep the useful source
+
+The operator selects two useful items: a supplier's finish guide and a local
+design brief. They are placed in the appropriate project route with short
+descriptions and references.
+
+The record says what these sources are for. It does not turn a pile of browser
+tabs or a full conversation transcript into “memory.”
+
+## 2. Frame a packet
+
+A packet asks an AI to produce a shortlist using those sources:
+
+~~~
+Task: propose three colour options for the pedestrian bridge.
+
+Use: the recorded finish guide and design brief.
+
+Include: the source behind each option, any uncertainty and a recommended next
+check.
+
+Do not decide the final colour.
+~~~
+
+The packet gives the next session a bounded question rather than asking it to
+guess the whole project.
+
+## 3. Keep the AI response beside the task
+
+The AI returns a shortlist and recommends a dark blue finish. It correctly
+links the supplier guide, but it also claims that the design brief requires
+dark blue.
+
+That claim is not supported by the brief.
+
+The response remains useful, but it is not silently treated as accepted work.
+
+## 4. Review the claim
+
+A human reviewer leaves a short review note:
+
+~~~
+The design brief requires good contrast and low glare; it does not require
+dark blue. Keep dark blue as an option, remove the unsupported claim, and ask
+for a contrast check against the planned surroundings.
+~~~
+
+The review is attached to the response it concerns, rather than becoming an
+untraceable correction in a later chat.
+
+## 5. Record the decision and next action
+
+The operator decides:
+
+~~~
+Decision: no final colour selected.
+
+Accepted: dark blue, white and yellow remain the shortlist.
+
+Rejected: the unsupported statement that dark blue is required.
+
+Next action: obtain a contrast assessment before choosing.
+~~~
+
+The decision is deliberately small. It says what changed, what was accepted,
+what was rejected and what should happen next.
+
+## 6. Let the next session continue
+
+A later AI or person starts at the shared entry route and finds the retained
+sources, packet, response, review and decision. It does not need the original
+chat to know:
+
+- what the question was;
+- why one claim was corrected;
+- what has not yet been decided; or
+- what to do next.
+
+## What this demonstrates
+
+LabNote does not make the AI's first answer correct, and it does not capture
+everything automatically. It gives the project a visible route for preserving
+the selected work, examining a mistake and carrying the corrected state
+forward.
+
+For the practical file-and-record steps, see the [first-use
+walkthrough](quickstart.md) and [review workflow](review_workflow.md).

locales/zh-CN/README.md

@@ -91,6 +91,8 @@ registry/       ← 记录发生了什么、何时发生以及由谁完成
 
 一条典型轨迹是 **packet → response → review → decision**；对应地，它是工作包（packet）→ 回复（response）→ 审阅（review）→ 决定（decision）。
 
+[查看一条从来源到决定的虚构轨迹，其中包括对 AI 说法的修正](docs/WORKED_CONTINUITY_TRAIL.md)。
+
 设置好 LabNote 后，你可以这样对 AI 说：
 
 ```text

locales/zh-CN/docs/README.md

@@ -14,6 +14,7 @@
 
 - [进行常规投递](../lobby/ROUTINE_DEPOSIT_QUICKSTART.md) —— 以最小但有用的记录，把一项普通工作放到正确路径中。
 - [理解审阅与决定](review_workflow.md) —— 查看回复如何成为经过审阅的结果。
+- [查看虚构的连续性轨迹](WORKED_CONTINUITY_TRAIL.md) —— 跟随来源材料经历回复、修正、决定和下一步行动。
 - [理解访客会话与会话身份](visitor_lobby_model.md) —— 了解实际工作区中的会话记录代表什么。
 - [理解消息路由](message_routing_model.md) —— 遵循会话或角色之间的消息模型。
 - [理解登记记录](REGISTRY_RECORDS.md) —— 查看持久登记记录的用途。

locales/zh-CN/docs/WORKED_CONTINUITY_TRAIL.md

@@ -0,0 +1,80 @@
+# 一个连续性轨迹示例
+
+这个完全虚构的示例展示了一条 LabNote 轨迹最小而有用的形态。它不代表真实项目，也不是这个公开源仓库中的实际投递记录。
+
+## 问题
+
+一个小团队正在为一座新的步行桥选择耐用、醒目的颜色。他们希望下一位人或 AI 能理解证据、提出过什么、哪些内容受到质疑，以及还有什么工作要做。
+
+他们**不**需要保存每一条聊天消息。
+
+## 1. 保留有用的来源
+
+操作者选择两份有用的材料：供应商的饰面指南和当地设计简报。它们被放入项目中的适当路径，并附上简短说明和引用。
+
+记录说明这些来源是做什么用的。它不会把一堆浏览器标签页或完整对话记录变成“记忆”。
+
+## 2. 形成一个工作包
+
+一个工作包要求 AI 根据这些来源提出一份候选清单：
+
+~~~
+任务：为步行桥提出三个颜色选项。
+
+使用：已记录的饰面指南和设计简报。
+
+包括：每个选项的来源、任何不确定之处，以及建议进行的下一项核查。
+
+不要决定最终颜色。
+~~~
+
+工作包为下一次会话提供一个有边界的问题，而不是要求它猜测整个项目。
+
+## 3. 将 AI 回复保留在任务旁
+
+AI 返回一份候选清单，并推荐深蓝色饰面。它正确链接了供应商指南，但它也声称设计简报要求使用深蓝色。
+
+该说法没有得到设计简报的支持。
+
+这份回复仍然有用，但不会被悄悄当作已接受的工作。
+
+## 4. 审阅该说法
+
+一位人类审阅者留下简短的审阅记录：
+
+~~~
+设计简报要求良好的对比度和低眩光；它并不要求深蓝色。保留深蓝色作为一个选项，删除没有依据的说法，并要求根据计划中的周边环境进行对比度核查。
+~~~
+
+审阅附在它所涉及的回复上，而不会变成后来某个聊天里无法追溯的修正。
+
+## 5. 记录决定和下一步行动
+
+操作者作出决定：
+
+~~~
+决定：尚未选定最终颜色。
+
+接受：深蓝色、白色和黄色仍是候选清单。
+
+拒绝：没有依据的“必须使用深蓝色”这一说法。
+
+下一步行动：在选择之前取得对比度评估。
+~~~
+
+这个决定刻意保持精简。它说明发生了什么变化、哪些内容被接受、哪些被拒绝，以及下一步应做什么。
+
+## 6. 让下一次会话继续
+
+后来的 AI 或人从共享入口路径开始，可以找到保留的来源、工作包、回复、审阅和决定。它不需要原始聊天，也能知道：
+
+- 原来的问题是什么；
+- 为什么某个说法被修正；
+- 什么尚未决定；以及
+- 下一步该做什么。
+
+## 这个示例说明什么
+
+LabNote 不会让 AI 的第一个答案自动正确，也不会自动捕获一切。它为项目提供一条可见的路径，用来保留经过选择的工作、检查错误，并把修正后的状态带到下一步。
+
+如需了解实际的文件和登记步骤，请参阅[首次使用流程](quickstart.md)和[审阅工作流](review_workflow.md)。