roleplay-mode-implementation
规划、实现或重构角色扮演模式（默认上帝视角下的单角色接管）时使用。适用于一期文本决策、二期统一有限选择框架、三期依附 Conversation 动作的自由对话，以及对应的 runtime/API/frontend/test 收口。
Best for Game developers implementin…Works with GitHub
⌘source
- author
@4thfever- language
- Python
★1.6kstars
179forks
Apr 18, 2026
✦overview.md
Key Features
- ·Implements or refactors roleplay mode with single-character takeover
- ·Maintains default god-view perspective without information narrowing
- ·Pauses only at decision boundaries, not mid-action chains
- ·Clears runtime state (RoleplaySession, pendingrequest, conversationsession) on load/reinit/reset/start
- ·Uses unified limited-choice framework for phase two decisions
- ·Attaches phase three dialogue to Conversation actions, not free chat
Use Cases
- →Adding capabilities or fixing regressions for phase one text decisions
- →Connecting player takeover to phase two singlechoice or mutual action choice systems
- →Modifying phase three Conversation dialogue chains, summaries, or dock chat UI
- →Adjusting RoleplaySession, pendingrequest, or conversationsession components
Best for
- ✓Game developers implementing roleplay features
- ✓Teams maintaining multi-phase decision systems
- ✓Engineers working on runtime session management
Not ideal for
- !Projects without structured decision frameworks
- !One-off script modifications
- !Systems requiring independent personal modes
FAQs
name
roleplay-mode-implementation
description
规划、实现或重构角色扮演模式（默认上帝视角下的单角色接管）时使用。适用于一期文本决策、二期统一有限选择框架、三期依附 Conversation 动作的自由对话，以及对应的 runtime/API/frontend/test 收口。
适用场景
当任务涉及以下任一方向时使用本技能：
- 实现或重构角色扮演模式
- 调整
RoleplaySession、pending_request、conversation_session - 给一期文本决策补能力或修回归
- 给二期
single_choice/ mutual action choice 接玩家接管 - 给三期
Conversation对话链、summary、dock chat UI 做改动
先确认什么
- 这次改动属于一期、二期还是三期？
- 这次改动是在补主链，还是在做收尾/UI/测试？
- 是否会影响 runtime 暂停语义、存档语义、continuation 语义？
- 是否正在无意中把“上帝视角下的单角色接管”做回成独立个人模式？
必守心智
1. 默认始终是上帝视角
- 不把个人扮演做成与上帝模式并列的全局模式。
- 玩家扮演时仍可观察全世界，不做信息收窄。
2. 只在决策边界暂停
- 不能打断执行到一半的动作链。
- 只能在“当前动作已空 + planned actions 已空，需要新决策”时等待玩家。
3. 扮演态不进存档
RoleplaySessionpending_requestconversation_session- choice continuation
以上都属于 runtime，必须在 load / reinit / reset / start 后清空。
4. 二期优先走统一有限决策框架
- 不在每个业务点散落
if roleplay ... else ... - 优先让业务层只保留：
build_request()apply_decision(...)
- 决策来源交给统一 resolver / continuation
5. 三期对话必须依附 Conversation
- 不新增脱离动作语义的自由聊天入口
- 玩家只控制一侧发言，对方仍由 LLM 回复
- 原始聊天记录不进长期事件流，只以 summary 落地
推荐工作流
1. 先看锚点文档
优先查看：
docs/specs/avatar-roleplay-mode.mddocs/specs/single-choice-unified-framework.md.cursor/rules/roleplay-mode.mdc
如果改动涉及外接 API，再补看：
docs/specs/external-control-api.md
$install
1-click copynpx skills add 4thfever/cultivation-world-simulator --skill roleplay-mode-implementationSafety assessment
★
Clarity score
How clear and easy to understand the SKILL.md instructions are, rated from 1 to 5.
4/ 5
very goodClear and well structured, with only minor parts that might need a second read.
◎
Actionability score
How directly an agent can act on the SKILL.md instructions, rated from 1 to 5.
4/ 5
highMostly actionable with clear steps; only a few small gaps remain.