Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@novel-writer-mcp List my novel projects."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
v0.2.0 — 网文结构引擎 MCP：项目管理、场景 checklist、伏笔台账、draw.io 导入、推进建议、正文关联与 Task Master PRD 导出。
MCP 是「结构秘书」，不替写正文。程序与小说数据分离，支持双工作区布局。
git clone <your-repo-url> novel-writer-mcp
cd novel-writer-mcp
npm install
npm run build复制 .cursor/mcp.json.example 为 .cursor/mcp.json（或写入全局 Cursor MCP 配置）
将 <MCP_ROOT> 换成本仓库路径，<WRITING_ROOT> 换成你的小说数据目录
在 Cursor 中重载 MCP
验收：在对话中说「调用 list_profiles」或「get_project，project_id 为 _example」。
Related MCP server: Storywright MCP
文档 | 说明 |
安装、配置、首次 tool 调用 | |
引擎 vs 数据、环境变量 | |
MCP 仓 vs 写作仓 | |
mcp.json 详解 | |
从脑洞到连载 | |
副本章、constraints | |
记忆点、代际伏笔 | |
防发散、对话模板 | |
20 tools + 参数 | |
无限流 / 线性史诗 | |
标签与嵌套容器 | |
登记→回收 | |
PRD 导出 | |
JSON schema | |
兼容性 | |
GitHub / npm |
内置示例项目 novels/_example/（无限流，ch1 kind: dungeon）。
npm run build
npm test详见 CONTRIBUTING.md。
标记场景完成，写入 change，可选关联伏笔
Inputs: scene_id (required), project_id (required), change, foreshadow_ids.
从 profile 模板初始化小说项目目录
Inputs: id (required), title (required), profile (required), extends, +1 more.
导出章节 PRD 到 .taskmaster/docs/ 供 Task Master parse_prd 使用
Inputs: chapter_id (required), project_id (required).
整章 checklist + pacing 校验 + 伏笔到期 issues
Inputs: chapter_id (required), project_id (required).
返回 planned_payoff_chapter <= 当前章号且未 closed 的伏笔
Inputs: project_id (required), current_chapter.
查询开放中的伏笔（含 chapters_since_hint）
Inputs: project_id (required).
读取项目配置与 pacing
Inputs: project_id (required).
返回场景 checklist 及章级聚合校验（缺爽点/缺章末钩子）
Inputs: scene_id (required), project_id (required).
读取正文前 N 字 + 轻量校验 goal/change 关键词是否出现
Inputs: scene_id (required), project_id (required), max_chars.
再次暗示伏笔，关联 scene 并更新 last_hinted_chapter
Inputs: scene_id (required), project_id (required), foreshadow_id (required), chapter_number.
解析 .drawio XML，按节点标签生成 arc/chapter/scene 骨架
Inputs: project_id (required), xml, file_path.
设置 scene.prose_path，可选自动统计 word_count
Inputs: scene_id (required), project_id (required), prose_path (required), auto_word_count.
列出 profiles/*.json 可用模板
Inputs: no named parameters.
列出所有小说项目
Inputs: no named parameters.
标记伏笔已回收（status: closed）
Inputs: project_id (required), foreshadow_id (required), payoff_chapter, payoff_scene_id.
登记伏笔
Inputs: title (required), project_id (required), id, status, +4 more.
将某章拆为 N 个 scene，自动分配 checklist（按 chapter.kind 过滤）
Inputs: n (required), chapter_id (required), project_id (required).
基于 advance_operators 与上一场 change 输出候选场景，支持 constraints 过滤
Inputs: project_id (required), count, constraints.
更新章节 title/kind；kind 变更默认不重写已有 checklist，可选 rebuild_checklists
Inputs: chapter_id (required), project_id (required), kind, title, +1 more.
更新项目节奏参数
Inputs: project_id (required), hook_on, scenes_per_chapter, chapter_word_target, +1 more.
更新场景的 goal/conflict/change/checklist 勾选
Inputs: scene_id (required), project_id (required), pov, goal, +5 more.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 21 tools
Every tool targets a distinct resource/action: project CRUD+pacing, chapter/scene editing, foreshadow lifecycle, import/export, and manuscript linking. No two tools appear to do the same thing; even similar getters differ by scope (scene vs chapter, open vs due foreshadows).
All tool names use snake_case with a consistent verb_noun pattern (create_project, get_scene_checklist, pay_off_foreshadow). Minor variation like 'hint_foreshadow_again' still follows the same verb-first convention and remains predictable.
21 tools is on the heavier side but justified by the multi-faceted domain (project setup, chapter/scene management, foreshadowing, import/export). Each tool has a specific role, so the count does not feel bloated.
Covers the full novel-writing workflow: project initialization, chapter/scene breakdown, checklist validation, foreshadow tracking, outline import, and PRD export. Minor gaps like delete/archive operations are absent but not critical for the core creative process.
MCP Server for Slima - AI Writing IDE for Novel Authors with AI Beta Reader.
MCP server for generating rough-draft project plans from natural-language prompts.
MCP book-writing server: 81 classics, genre recipes, drafting pipelines.
ADHD-friendly task management MCP server for ChaTodo: organize tasks, lists, and priorities.
