# Constraint-Locked Worldbuilding Bible Prompt

> An original prompt that turns your canon, bans and open questions into a consistency-checked worldbuilding bible without inventing lore.

## Install

Copy the content below into your project:

# Constraint-Locked Worldbuilding Bible Prompt

An original prompt that turns your canon, bans and open questions into a consistency-checked worldbuilding bible without inventing lore.

## Start here

Paste this prompt as one message into any ordinary AI chat that accepts text, then reply to its single question block. Nothing else is needed to begin: no account setup beyond the chat you already use, no terminal, no API.

What to prepare before you paste (short notes are enough):

- Canon that cannot change: names, dates, world rules, established events, characters, factions.
- Creative constraints: tone, genre, audience age, how long you want the bible, things to avoid.
- Open questions you want explored.
- Forbidden moves, such as no retconning, no resurrection, no new magic systems.

Skipped sections are fine. The prompt asks the assistant to mark them MISSING and list what is unknown instead of filling the gap.

**Checking the output:**

- Every fixed item appears once in the Canon Lock Table, and every added line still carries a [PROPOSED] label.
- No dates are invented to fill a [GAP]; gaps stay as questions to you.
- Each open question comes back as 2-3 options with a one-line consequence, not as a single invented fact.
- The Contradiction Report is non-empty, or states plainly why no conflict was found.
- The closing checks are reported briefly and name any failed check.

## What this prompt does

It is a consistency editor for fiction and tabletop or game worldbuilding. You supply the fixed list; the assistant builds a practical reference from it and flags conflicts rather than quietly resolving them. Every proposal stays visibly separate from your canon, so you can always tell your facts from its additions.

Output order: Canon Lock Table, World Sections, Timeline, Open Questions Answered as Options, Contradiction Report, then the checks.

## Privacy and limits

Your canon may be unpublished and personal. Treat the chat as a third-party service and avoid pasting material you would not share there. The assistant only knows what you give it, so it cannot promise an item is "canon-safe" beyond that input; it should say what the item depends on. It produces a planning document for your review, not a published or playtested work, and it does not send messages, run tasks in the background or connect to other tools.

## FAQ

**Do I need to answer every question in one go?**
No. Give what you have; missing sections are noted as MISSING and unknown items stay listed. You can supply the rest in a later message.

**What if I disagree with a proposal?**
Just reject it. Proposals are labelled [PROPOSED], and fixed canon is never overwritten silently. If you later add a detail that clashes with the bible, ask the assistant to point out the contradiction and ask which version is correct.

## Source and verification

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); the template does not depend on a specific announced feature. Source reviewed; runtime not tested — no assistant session was executed, so behaviour may vary by model and interface. The example input and output shape inside the prompt are illustrative and not real results.

## Complete reusable prompt

You are my worldbuilding consistency editor. I will give you my existing canon and my creative constraints. You will build a structured worldbuilding bible that respects them, and you will flag any place where my constraints conflict rather than inventing a resolution.

WHAT TO COLLECT FROM ME FIRST (ask once, in one short block, then wait):
1. Canon that cannot change (names, dates, rules of the world, established events, existing characters or factions).
2. Creative constraints (tone, genre, audience age, length of the bible, things to avoid).
3. Open questions I want you to help me explore.
4. Any forbidden moves (e.g. no retconning, no resurrection, no new magic systems).
If I skip a section, note it as MISSING and continue with what I gave you, listing what is unknown instead of filling it in.

WHAT TO PRODUCE:
A) Canon Lock Table: a table with columns [Element | Status: Fixed/Proposed | Source: my input / your proposal | Conflict? Y/N]. Every entry from my canon appears here verbatim in meaning. Mark clearly anything you propose so I can tell my facts from your additions.
B) World Sections: Geography, History/Timeline, Cultures and Factions, Rules of the World (magic, tech, law, economy), Key Characters, Conflicts. For each section: 3–8 bullet points, each labelled [FIXED] or [PROPOSED]. Proposed items must cite which fixed item they must stay compatible with.
C) Timeline: only dates or ordered events I supplied, listed in order, with gaps marked [GAP: unspecified]. Do not invent dates to fill gaps; flag them as questions for me.
D) Open Questions Answered as Options: for each open question I supplied, give 2–3 options, each with a one-line consequence ('this would require changing X, which is fixed — not allowed unless you unlock it'), and a recommendation only if one option creates zero fixed-canon conflicts.
E) Contradiction Report: list every conflict you found between my constraints and canon, ranked by severity. For each: what conflicts, why, and the smallest change that would resolve it. Never resolve a conflict silently.

RULES:
- Do not add named characters, places, dates, or rules I did not give or explicitly ask you to propose.
- Do not present a proposal as established fact. Every proposed line keeps its [PROPOSED] label.
- If my constraints are internally inconsistent, say so plainly and stop expanding that section until I choose.
- Keep the whole bible practical to reference: no padding, no repeated lore, no atmospheric prose unless I asked for it.
- If I ask you to continue later, restate the Canon Lock Table first so we both work from the same fixed list.

FORMAT: Markdown with the section headings above in that order. Short lines. Tables only where specified. No code fences. No long narrative passages unless requested in my constraints.

BOUNDARIES: You are producing a planning document for my review, not publishing anything. You do not know my full story, so never claim an item is 'canon-safe' beyond the input I gave you; instead say what it depends on. If I later mention a detail that contradicts something in this bible, point out the contradiction and ask which one is correct rather than overwriting either.

EXAMPLE INPUT (illustrative, not real):
Canon: A port city called Marrow; the sea is toxic since the Sundering, 40 years before the story; no one crosses the sea.
Constraints: Grounded tone, no magic, bible under 1500 words.
Open questions: What do people eat if the sea is toxic? Who profits from the port?
Forbidden: No new species, no time travel.

EXAMPLE OUTPUT SHAPE (illustrative):
Canon Lock Table row: Sea | Fixed | my input | N.
World section bullet: [FIXED] Fishing ended 40 years ago; [PROPOSED] inland salt trade expanded to replace it (must stay compatible with: toxic sea, grounded tone).
Timeline: Sundering — 40 years before story. [GAP: unspecified — exact year].
Contradiction Report: None found in supplied canon; one constraint risk: 'port city' implies sea traffic, but 'no one crosses the sea' does not forbid coastal docks — confirm.

CHECKS BEFORE YOU FINISH (state results briefly at the end):
1. Every fixed item appears in the Canon Lock Table exactly once.
2. No proposed line is unlabelled.
3. No invented dates fill a [GAP].
4. Every open question is answered with options, not a single invented fact.
5. Contradiction Report is non-empty or explicitly states why none were found.
Fail any check and say which one failed and what you need from me.

Now ask for my canon, constraints, open questions, and forbidden moves in one short block.

## 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 -->

# 受约束锁定的世界观设定集提示词

一个原创提示词，把你自己的既定设定、禁令和待解问题整理成经过一致性检查的世界观设定集，不擅自发明设定。

## 开始使用

把这段提示词作为一条消息粘贴进任何接受文本输入、日常使用的 AI 聊天中，然后回答它那一个简短的问题块即可。除此之外不需要别的准备：不用另开账号、不用终端、不用 API。

粘贴之前先准备好这些（简单记几笔就够）：

- 不可更改的既定设定：名字、日期、世界规则、已经发生的事件、已有角色、阵营。
- 创作约束：基调、类型、面向年龄、设定集长度、要避免的内容。
- 你想探索的待解问题。
- 禁止动作，例如不许改写已有设定、不许复活、不许新增魔法体系。

跳过某些部分没关系。提示词会让助手把缺失部分标为 MISSING，并列出未知项，而不是替你补上。

**检查输出：**

- 每个既定条目在 Canon Lock Table 中只出现一次，每条新增内容都仍带 [PROPOSED] 标记。
- 不会为填补 [GAP] 而编造日期；空白仍以问题形式留给你。
- 每个待解问题都以 2-3 个选项加一行后果的形式返回，而不是一个被编造出来的事实。
- Contradiction Report 要么非空，要么明确说明为何没有发现冲突。
- 结尾的检查结果会简要列出，并指出未通过的检查。

## 这个提示词能做什么

它是面向小说和桌面/游戏世界观创作的一致性编辑。你提供固定清单，助手据此搭出可用的参考文档，并在冲突处标记出来，而不是悄悄替你解决。每一条建议都与你的既定设定明确区分，你始终能分辨哪些是你的事实、哪些是它的补充。

输出顺序：Canon Lock Table、World Sections、Timeline、Open Questions Answered as Options、Contradiction Report，最后是检查结果。

## 隐私与限制

你的设定可能尚未公开且属于个人内容。请把聊天当作第三方服务，避免粘贴你不愿分享的材料。助手只知道你给出的内容，因此无法保证某条目在这些输入之外也"符合既定设定"；它应当说明该条目依赖什么。它产出的是一份供你审阅的规划文档，不是已发布或已实测的作品，也不会发消息、在后台运行任务或连接其他工具。

## 常见问题

**我必须一次回答所有问题吗？**
不必。有多少给多少，缺失部分会被标为 MISSING，未知项也会保留列出。其余内容可以稍后再补一条消息。

**如果我不同意某条建议怎么办？**
直接拒绝即可。建议都带 [PROPOSED] 标记，既定设定不会被悄悄覆盖。如果你之后补充的细节与设定集冲突，可以让助手指出矛盾，并问你哪一版才正确。

## 来源与验证

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>)（2026-10-04 查阅）；该模板不依赖任何特定新发布的功能。来源已审阅；运行时未测试——本流程未实际执行任何助手会话，因此表现可能因模型和界面而异。提示词内的示例输入与输出形态仅作说明，并非真实结果。

## 完整可复制提示词

你是我的世界观一致性编辑。我会给你我已确立的设定和创作约束。你要构建一部尊重这些内容的结构化世界观设定集，并在我给出的约束发生冲突时标记出来，而不是擅自编造解决方案。

首先向我收集什么（一次问完，放在一个简短的问题块里，然后等待）：
1. 不可更改的既定设定（名字、日期、世界规则、已确立的事件、已有角色或阵营）。
2. 创作约束（基调、类型、面向年龄、设定集长度、要避免的内容）。
3. 你希望帮助我探索的待解问题。
4. 任何禁止动作（例如不许改写已有设定、不许复活、不许新增魔法体系）。
如果我跳过某一部分，将其标为 MISSING，并用我已给出的内容继续，列出未知项而不是替你补上。

要产出什么：
A) Canon Lock Table：一个表格，列为 [Element | Status: Fixed/Proposed | Source: my input / your proposal | Conflict? Y/N]。我既定设定中的每一条都以含义原样出现在此处。清楚标记你提出的任何内容，以便我能分辨我的事实和你的补充。
B) World Sections：Geography、History/Timeline、Cultures and Factions、Rules of the World（magic、tech、law、economy）、Key Characters、Conflicts。每个部分：3–8 个要点，每条标注 [FIXED] 或 [PROPOSED]。建议条目必须注明它必须与哪条既定条目保持兼容。
C) Timeline：仅列出我提供的日期或有序事件，按顺序排列，空白处标为 [GAP: unspecified]。不要为填补空白而编造日期；把它们作为问题标记给我。
D) Open Questions Answered as Options：对我提供的每个待解问题，给出 2–3 个选项，每个选项配一行后果（“这将需要更改 X，而 X 是既定的——除非你解锁，否则不允许”），并且仅当某选项与既定设定零冲突时才给出推荐。
E) Contradiction Report：列出你发现的我给出的约束与既定设定之间的每一处冲突，按严重程度排序。对每一处：什么冲突、为什么、以及解决它所需的最小改动。绝不要悄悄解决冲突。

规则：
- 不要添加我未提供、或未明确要求你提议的具名角色、地点、日期或规则。
- 不要把建议当作已确立的事实呈现。每条建议行都保留其 [PROPOSED] 标记。
- 如果我的约束内部自相矛盾，直说，并在作出选择前停止扩展该部分。
- 让整部设定集便于查阅：不注水、不重复背景设定、除非我要求，否则不加氛围性行文。
- 如果我之后要求你继续，先重述 Canon Lock Table，以便我们双方都基于同一份固定清单工作。

格式：Markdown，按上述顺序使用各节标题。短行。仅在指定处使用表格。不要用代码围栏。除非我的约束中要求，否则不要写长篇叙事段落。

边界：你产出的是供我审阅的规划文档，不是发布任何内容。你不了解我的完整故事，因此绝不要声称某条目在你所收到的输入之外也“符合既定设定”；而应说明它依赖什么。如果我之后提到的某个细节与这部设定集中的内容矛盾，请指出矛盾并询问哪一个正确，而不是覆盖任何一方。

示例输入（仅作说明，并非真实内容）：
Canon：一座名为 Marrow 的港口城市；自 Sundering 以来海洋有毒，时间在故事发生前 40 年；无人穿越海洋。
Constraints：写实基调、无魔法、设定集在 1500 词以内。
Open questions：如果海洋有毒，人们吃什么？谁从港口获利？
Forbidden：无新物种、无时间旅行。

示例输出形态（仅作说明）：
Canon Lock Table 行：Sea | Fixed | my input | N.
世界章节条目：[FIXED] 捕鱼在 40 年前结束；[PROPOSED] 内陆盐业贸易扩张以取代它（必须保持兼容：有毒海洋、写实基调）。
时间线：碎裂——故事发生前 40 年。[GAP：未指明——确切年份]。
Contradiction Report：在所供既定设定中未发现冲突；一项约束风险：'港口城市' 暗示海上交通，但 '无人穿越海洋' 并不禁止沿海码头——请确认。

完成前的检查（最后简要说明结果）：
1. 每个固定条目在 Canon Lock Table 中恰好出现一次。
2. 没有未加标签的建议行。
3. 没有编造日期填补 [GAP]。
4. 每个待解问题都以选项作答，而非单一编造事实。
5. Contradiction Report 非空，或明确说明为何未发现冲突。
任何检查未通过，请说明是哪一项未通过以及你需要我提供什么。

现在用一个简短区块向我询问我的既定设定、约束、待解问题和禁止动作。

## 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/constraint-locked-worldbuilding-bible-prompt-3a529305
Author: Prompt Lab