Align context projection defaults with gm-first play model.

Richer synth fallbacks and context-order examples: gm gets narrative guide and history; narrator and auditor skip history; auditor reads settlement at turn end.

Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
2026-08-25 00:51:47 +08:00
co-authored by Cursor
parent 8360e885bb
commit ac474bcd1a
7 changed files with 66 additions and 44 deletions
@@ -36,7 +36,7 @@ feeds: design_only
```opening
按已验收产物的概括,给每个固定执行单元分配上下文顺序(只排序,不改设定、不演游玩)。
名册由程序钉死:旁观维护(表格)/ 主世界层 / 叙事转述 / 可选角色视角 / 按需机遇。
名册由程序钉死:主世界层 / 叙事转述 / 旁观维护(回合末)/ 可选角色视角 / 按需机遇(已并入主世界层工具)。
若前面产物够用,直接回复「按已有概括排序」。
```
@@ -55,7 +55,9 @@ feeds: design_only
1. 禁止发明白名单外 id(含画图等未入册槽)。
2. 禁止重写上游长文;只写顺序与投影。
3. 每轮槽:order 0=该槽固定人设;其后按「稳→对话.历史→易变」软排。
4. 主世界层与转述必须含 ref=`对话.历史`(默认 projection=summary);旁观维护默认不含历史。
4. 主世界层必须含 ref=`对话.历史`(默认 projection=summary);转述与旁观维护默认不含历史。
5. 主世界层应双挂「设计.叙事指南与故事推进」等 mount 到 gm 的块;转述只读裁决+文风/壳。
6. 旁观维护在回合末:须含「运行.本轮.裁决」与变量合同,不含历史。
5. 按需槽(机遇)enabled 时 inserts 可空。
6. 越稳越靠前;本轮输入/真值/裁决通常在历史之后。
7. 同一内容可挂多槽;chance 不必排进每轮 inserts。
@@ -93,20 +95,6 @@ feeds: design_only
"schema": "context-order.v1",
"brief": "一句话:本局有哪些执行单元、各自上下文怎么排",
"agents": [
{
"id": "auditor",
"label": "旁观维护",
"enabled": true,
"when": "every_turn",
"duty": "查表/补规则,不写正文、不写真相",
"inserts": [
{ "order": 0, "ref": "worker.persona", "projection": "fixed", "note": "人设" },
{ "order": 1, "ref": "设计.变量设计与更新规则", "projection": "summary" },
{ "order": 2, "ref": "设计.生成规则", "projection": "summary" },
{ "order": 3, "ref": "变量.当前", "projection": "fields" },
{ "order": 4, "ref": "用户.最新输入", "projection": "full" }
]
},
{
"id": "gm",
"label": "主世界层",
@@ -116,9 +104,13 @@ feeds: design_only
"inserts": [
{ "order": 0, "ref": "worker.persona", "projection": "fixed" },
{ "order": 1, "ref": "设计.舞台骨架", "projection": "summary" },
{ "order": 2, "ref": "对话.历史", "projection": "summary" },
{ "order": 3, "ref": "变量.当前", "projection": "fields" },
{ "order": 4, "ref": "用户.最新输入", "projection": "full" }
{ "order": 2, "ref": "设计.实现机制", "projection": "summary" },
{ "order": 3, "ref": "设计.变量设计与更新规则", "projection": "summary" },
{ "order": 4, "ref": "设计.生成规则", "projection": "summary" },
{ "order": 5, "ref": "设计.叙事指南与故事推进", "projection": "summary" },
{ "order": 6, "ref": "对话.历史", "projection": "summary" },
{ "order": 7, "ref": "变量.当前", "projection": "fields" },
{ "order": 8, "ref": "用户.最新输入", "projection": "full" }
]
},
{
@@ -130,10 +122,26 @@ feeds: design_only
"inserts": [
{ "order": 0, "ref": "worker.persona", "projection": "fixed" },
{ "order": 1, "ref": "设计.叙事指南与故事推进", "projection": "summary" },
{ "order": 2, "ref": "对话.历史", "projection": "summary" },
{ "order": 2, "ref": "设计.正文组成", "projection": "summary" },
{ "order": 3, "ref": "运行.本轮.裁决", "projection": "full" }
]
},
{
"id": "auditor",
"label": "旁观维护",
"enabled": true,
"when": "every_turn",
"duty": "回合末查表/补规则,不写正文",
"inserts": [
{ "order": 0, "ref": "worker.persona", "projection": "fixed" },
{ "order": 1, "ref": "设计.变量设计与更新规则", "projection": "summary" },
{ "order": 2, "ref": "设计.生成规则", "projection": "summary" },
{ "order": 3, "ref": "设计.变量控制上下文", "projection": "summary" },
{ "order": 4, "ref": "运行.本轮.裁决", "projection": "full" },
{ "order": 5, "ref": "变量.当前", "projection": "fields" },
{ "order": 6, "ref": "用户.最新输入", "projection": "full" }
]
},
{
"id": "perspective",
"label": "角色视角",
@@ -162,7 +170,8 @@ feeds: design_only
- [ ] 未重写上游长文,只有排序/投影
- [ ] 未把游玩选项示例当成当前提问
- [ ] 已启用的每轮槽有 order 0 人设位
- [ ] 主世界层与转述含「对话.历史」;旁观维护默认无历史
- [ ] 主世界层含「对话.历史」;转述与旁观维护默认无历史
- [ ] 旁观维护含「运行.本轮.裁决」;转述含裁决且不含历史
- [ ] 少变块整体靠前、本轮/真值靠后(软规则)
- [ ] 与游玩拓扑勾选一致
```
@@ -54,7 +54,7 @@ boundary: |
核心操作:让用户勾选固定槽,写成 `play_slots`(世界模拟)或 `writing_slots`(扩写)。**禁止**新建未在固定列表中的 ref。
固定 ref 白名单:
- 世界模拟每轮:`auditor`(旁观维护)、`world-simulator`(gm)、`narrator`、`role-decide`(perspective,默认关)
- 世界模拟每轮:`world-simulator`(gm)、`narrator`、`auditor`(回合末)、`role-decide`(perspective,默认关)
- 世界模拟按需:`chance`(机遇裁定,默认关;invocation=on_demand)
- 扩写:`outline`、`chapter-writer`
- 禁止:`variable-update`、自造 kebab、为世界观/性格再拆槽
@@ -62,7 +62,7 @@ boundary: |
执行顺序:
1. 读配方与体验契约,判断路径:世界模拟 vs 写手分段。
2. 默认世界模拟:`auditor: true, gm: true, narrator: true, perspective: false, chance: false`;
调度序 auditor → perspective? → gm → narrator;
调度序 gm → perspective? → narrator → auditor;
仅信息隔离才开 perspective;需要骰子/抽签/比点等真随机时开 chance。
3. 写手路径:`outline` + `chapter-writer` 默认都开;用户明确只要正文则可关 outline。
4. 可写简短 `mount_notes`(哪类上游产物挂哪槽),不粘贴长文。