Skip to main content
Glama
Yuque MCP Server
by yuque

Let AI assistants read and write your Yuque (语雀) knowledge basethrough the Model Context Protocol.

CI npm version npm downloads License

Quick Start · Tools · Troubleshooting · Docs · 中文文档

Once connected, ask your assistant things like:

"Search my Yuque for everything about canary releases and give me a one-page summary."

"Turn today's meeting notes into a doc in my Tech Research book."

"Add a flowchart of this deployment pipeline to the design doc."

Quick Start

1. Get a token — create one at Yuque Developer Settings. If you use a team token bound to a Yuque space, also note the space host (e.g. https://your-space.yuque.com) — you will pass it as --host.

2. Install — one command locates the right config file for your OS and merges a yuque entry into it, without touching other servers:

npx yuque-mcp install --token=YOUR_TOKEN --client=cursor

Supported clients: claude-desktop · vscode · cursor · windsurf · cline · trae · qoder · opencode. Prefer an interactive flow? Run npx yuque-mcp setup.

Register the server directly:

claude mcp add yuque -- npx -y yuque-mcp --token=YOUR_TOKEN

Any client that supports stdio transport works — see docs/clients.md for per-client config paths.

{
  "mcpServers": {
    "yuque": {
      "command": "npx",
      "args": ["-y", "yuque-mcp"],
      "env": { "YUQUE_TOKEN": "YOUR_TOKEN" }
    }
  }
}

3. Restart your client and start asking.

Related MCP server: Yuque MCP Server

Configuration

Setting

Env var / CLI flag

Description

Token (required)

YUQUE_TOKEN / --token

Personal or team Yuque API token

Host (optional)

YUQUE_HOST / --host

Site or space host, e.g. https://your-space.yuque.com — required for space-bound team tokens and private deployments

Site roots are normalized to /api/v2; when unset, the host defaults to https://www.yuque.com/api/v2.

# Team token / private deployment
npx yuque-mcp install --token=YOUR_TOKEN --client=cursor --host=https://your-space.yuque.com

YUQUE_PERSONAL_TOKEN, YUQUE_BASE_URL, and --base-url still work as legacy fallbacks. Precedence: YUQUE_TOKEN > YUQUE_PERSONAL_TOKEN > --token, and YUQUE_HOST > --host > YUQUE_BASE_URL > --base-url. New configs should use YUQUE_TOKEN and YUQUE_HOST.

Tools (19)

Each tool maps to exactly one Yuque API route.

Category

Tool

Description

User

yuque_get_user

Get the authenticated user for the current token

Search

yuque_search

Search docs or repos, with paging

Books

yuque_list_books

List books (知识库) of a user

yuque_get_book

Get a book by ID or namespace

yuque_create_book

Create a book

yuque_update_book

Update name, slug, description, or visibility

Docs

yuque_list_docs

List docs in a book, with paging

yuque_get_doc

Get full content — markdown, lake, or html

yuque_create_doc

Create a doc in a book

yuque_update_doc

Update a doc's body or metadata

TOC

yuque_get_toc

Get a book's table of contents

yuque_update_toc

Append or move a single TOC node

Notes

yuque_list_notes

List notes (小记), with paging and status filter

yuque_get_note

Get a note with full content

yuque_create_note

Create a note

yuque_update_note

Update a note

Boards

yuque_get_resource

Read a board (mindmap / flowchart / diagram) from a doc

yuque_create_resource

Create a board in a doc

yuque_update_resource

Update a board in a doc

In particular, yuque_update_doc cannot combine a markdown body with title / slug / public changes in a single call — update metadata separately. The full contract, including format routing between the YMD markdown API and the legacy document API, is documented in docs/capability-scope.md.

Not covered (yet): comments, attachment upload and file management, permission and member management, section-level doc edits, and structured resources other than boards.

Write access

The create/update tools modify real content in your knowledge base, and the server can do whatever your token can do. Keep the token secret, and prefer a space-scoped team token (with YUQUE_HOST) when you only work within one space. To report a vulnerability, see SECURITY.md.

Troubleshooting

Error

Solution

YUQUE_TOKEN ... is required

Set YUQUE_TOKEN=YOUR_TOKEN or pass --token=YOUR_TOKEN

401 Unauthorized

Token invalid or expired — regenerate it

429 Rate Limited

Too many requests — wait a moment and retry

410 Gone

Target permanently deleted or endpoint deprecated — check the doc/book exists

Tool not found

Update to the latest version: npx -y yuque-mcp@latest

npx command not found

Install Node.js v18 or later

Development

git clone https://github.com/yuque/yuque-mcp-server.git
cd yuque-mcp-server
npm install
npm test              # run tests
npm run build         # compile TypeScript
npm run dev           # dev mode with hot reload

Architecture, tech stack, and the full tool contract live in docs/. Contributions are welcome — see CONTRIBUTING.md.

License

MIT

Available Tools

19 tools
yuque_create_bookB

Create a new book (知识库) for the current user

Inputs: login (required), name (required), slug (required), description, +2 more.

yuque_create_docA

Create a new document in a repo/book

Inputs: repo_id (required), title (required), slug, body, +2 more.

yuque_create_noteB

Create a new note (小记)

Inputs: body (required).

yuque_create_resourceB

Create a structured resource in a Yuque document. Currently resource_type only supports board: mindmap, flowchart, or architecturediagram.

Inputs: resource_type (required), type (required), dsl (required), doc_id, +2 more.

yuque_get_bookB

Get a specific book (知识库) by ID or namespace

Inputs: repo_id (required).

yuque_get_docB

Get a specific document with full content

Inputs: repo_id (required), doc_id (required), format, include_lake.

yuque_get_noteB

Get a specific note with full content

Inputs: note_id (required).

yuque_get_resourceA

Read a structured resource view from a Yuque document. Currently resource_type only supports board; pass the raw resource_id, not board://<resource_id>.

Inputs: resource_type (required), resource_id (required), doc_id, url.

yuque_get_tocB

Get the table of contents (TOC) for a repo/book

Inputs: repo_id (required).

yuque_get_userA

Get current authenticated user information

Inputs: no named parameters.

yuque_list_booksB

List all books (知识库) for the current user

Inputs: login (required).

yuque_list_docsA

List all documents in a repo/book

Inputs: repo_id (required).

yuque_list_notesB

List all notes (小记) for the current user with pagination

Inputs: status, page, limit.

yuque_update_bookC

Update a book (知识库)

Inputs: repo_id (required), name, slug, description, +1 more.

yuque_update_docC

Update an existing document

Inputs: repo_id (required), doc_id (required), title, slug, +3 more.

yuque_update_noteC

Update an existing note

Inputs: note_id (required), body (required).

yuque_update_resourceA

Update a structured resource in a Yuque document. Currently resource_type only supports board; provide exactly one of text or dsl.

Inputs: resource_type (required), resource_id (required), doc_id, url, +2 more.

yuque_update_tocA

Update the table of contents (TOC) for a repo/book. The toc_data must be a single-operation JSON object (not an array). Required fields: "action" (e.g. "appendNode"), "action_mode" ("child" or "sibling"), "target_uuid" (empty string for root level). For new nodes: include "type" ("TITLE" or "DOC") and "title". To move existing nodes: use "node_uuid" instead. Example: {"action":"appendNode","action_mode":"child","target_uuid":"","type":"TITLE","title":"New Section"}

Inputs: repo_id (required), toc_data (required).

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    B3.2/5.0

    Scored across 19 tools

    Disambiguation5/5

    Each tool targets a specific resource type and action (e.g., create, get, list, update) with clear naming prefixes like yuque_create_book, yuque_get_doc, etc. No two tools have overlapping purposes, making selection unambiguous.

    Naming Consistency5/5

    All tools follow a consistent 'yuque_verb_noun' pattern using lowercase snake_case. The verbs are uniform (create, get, list, update) and the nouns correspond to resource types (book, doc, note, resource, toc), ensuring predictability.

    Tool Count4/5

    With 19 tools covering books, docs, notes, resources, TOC, search, and user info, the count is on the higher side but still reasonable for a platform with multiple resource types. Each tool serves a distinct purpose without redundancy.

    Completeness2/5

    The toolset lacks delete operations for any resource type (books, docs, notes, resources, TOC). There is also no listing for resources or TOC items. These gaps prevent full lifecycle management and may cause agent failures when cleanup is needed.

    Maintenance

    ActivityInactive
    ResponsivenessMostly unanswered

    Related MCP Connectors

    Related MCP Servers