dingtalk-wiki-mcp
Enables AI agents to read, browse, and create DingTalk Wiki content, including workspaces, folders, documents, and mind maps. It complements the official DingTalk MCP by providing comprehensive document and workspace management capabilities.
README
dingtalk-wiki-mcp
DingTalk Wiki / Docs read-write MCP server that fills the gap left by DingTalk official MCP.
DingTalk's official MCP does not provide Wiki / Docs read-write capability.
This project is an open-source complement that makes AI agents and MCP clients actually able to read, browse, and create DingTalk Wiki / Docs content.
Repository Highlights
- Official MCP gap: Wiki / Docs read-write is not covered
- This project adds it: workspace browsing, node browsing, and document creation
- MCP-compatible: works with stdio-based MCP clients
- Agent-ready: includes
SKILL.mdfor OpenClaw-style skill workflows
Quick Start
1) Install
npm install
2) Configure environment
cp .env.example .env
Required:
DINGTALK_APP_KEY=your-app-key
DINGTALK_APP_SECRET=your-app-secret
index.js now auto-loads .env from the current working directory (or the repo directory) if those variables are not already present in the environment.
3) Prepare local config
cp config.example.json config.json
4) Run
npm start
Or:
node index.js
npx dingtalk-wiki-mcpis a future-friendly path after npm publishing.
This repository already includes the correct CLI entry (bin), but npm distribution is not part of the current release yet.
DingTalk official MCP vs this project
| Capability | DingTalk official MCP | dingtalk-wiki-mcp |
|---|---|---|
| Wiki read | Not covered | ✅ |
| Wiki write | Not covered | ✅ |
| Create docs | Not covered | ✅ |
| Create folders | Not covered | ✅ |
| Create mind maps | Not covered | ✅ |
| Browse workspaces | Not covered | ✅ |
| Browse nodes / folders | Not covered | ✅ |
Read Notable / .able records |
Not covered | ✅ |
| MCP client compatibility | Partial / official scope only | ✅ stdio MCP-compatible |
| OpenClaw skill packaging | No | ✅ includes SKILL.md |
Positioning principle: this project does not replace the official DingTalk MCP. It complements it by filling the Wiki / Docs gap.
Core capabilities
Wiki / Docs
- List Wiki workspaces
- Get workspace details
- List Wiki nodes (folders / docs)
- Create:
DOCWORKBOOKMINDFOLDER
- Search Wiki by linking to DingTalk search
- Read Notable /
.ablesheets and records via official API
Organization
- List departments
- List department users
- Get user info
Operator / Config
- Set current operator (
unionId) - Use a default operator from local config
- Inspect current local config
Skill included
This repo is not only an MCP server. It also includes:
SKILL.md
So it can be reused as a skill package in OpenClaw-style agent workflows.
Demo
1. List Wiki workspaces
2. Browse workspace nodes
3. Create a document
These demo images are illustrative documentation assets built from representative command/output flows, with all tenant-specific data removed.
Real use cases
1) AI automatically creates weekly report docs
Your AI agent can create a fresh DingTalk Wiki document every week for sales, product, or ops reporting.
2) Agent explores Wiki structure before writing
Before generating content, an agent can inspect workspaces and folders first, then choose the right target node.
3) Auto-initialize project knowledge-base structure
When a new project starts, automation can create a standard folder tree such as:
- Project Overview
- Weekly Reports
- Specs
- Release Notes
- Retrospectives
Client integration examples
Example usage
Registered server mode
If you have registered this server in your MCP client config under the name dingtalk-wiki:
mcporter call dingtalk-wiki.show_config
mcporter call dingtalk-wiki.list_wiki_workspaces
mcporter call dingtalk-wiki.list_wiki_nodes workspace_id="your_workspace_id"
mcporter call dingtalk-wiki.create_wiki_doc workspace_id="your_workspace_id" name="Weekly Summary" doc_type="DOC"
mcporter call dingtalk-wiki.get_user_info userid="your_user_id"
Direct stdio mode
If you want to run the server directly without pre-registering it, call the tool names directly:
mcporter call --stdio "node ./index.js" show_config
mcporter call --stdio "node ./index.js" list_wiki_workspaces
mcporter call --stdio "node ./index.js" list_wiki_nodes workspace_id="your_workspace_id"
mcporter call --stdio "node ./index.js" create_wiki_doc workspace_id="your_workspace_id" name="Weekly Summary" doc_type="DOC"
Available MCP tools
set_operatorshow_configlist_wiki_workspacesget_wiki_workspacelist_wiki_nodesget_wiki_nodecreate_wiki_docsearch_wikilist_departmentsget_department_usersget_user_infolist_notable_sheetslist_notable_records
Requirements
- Node.js 18+
- A DingTalk app with the required API permissions
- A stdio-compatible MCP client, such as:
- OpenClaw
- mcporter
- other MCP hosts / clients
Permissions
Depending on what you use, your DingTalk app may need permissions such as:
Document.WorkspaceDocument.Write- Wiki read permissions
- Department read permissions
- User read permissions
Please refer to DingTalk Open Platform documentation for the latest permission names and approval requirements.
Trust materials
Security notes
config.jsoncontains your local user and workspace metadata, so do not commit it- this repository already ignores
config.jsonand.env - inject AppKey / AppSecret via environment variables instead of hardcoding them
Limitations
- This is a community-maintained complement, not an official DingTalk project
- Some APIs require enterprise approval on the DingTalk side
search_wikiis currently more of a search-entry helper than a full-text search implementation- Reading normal DingTalk document body content via official public API is still not implemented in this project
- Notable /
.ablesupport currently covers sheets and records, not arbitrary document-body export
Related links
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。