lark-docs

Lark Docs Access (飞书云文档权限)

以用户身份访问和搜索飞书云文档，包括文档、表格、多维表格和知识库等。

Access and search Lark cloud documents as the authenticated user, including docs, sheets, bitables, and wiki spaces.

Quick Start

Search for Documents (搜索文档)

搜索包含'OKR'的文档
Search for documents containing 'product roadmap'

Access Document Content (访问文档内容)

获取这个文档的内容：https://example.feishu.cn/docx/abc123
Get the content of this document: https://example.feishu.cn/wiki/xyz789

Search Wiki Spaces (搜索知识库)

在知识库中搜索关于'架构设计'的内容
Search wiki for 'system architecture' documentation

Key Features

1. Document Search (文档搜索)

Search across all document types that the user has access to:

• Documents (文档): docx format documents
• Sheets (表格): spreadsheet documents
• Bitables (多维表格): database/table documents
• Wiki (知识库): wiki pages and spaces

Available MCP Tools:

• mcp__lark__docx_builtin_search - Search cloud documents
• mcp__lark__wiki_v1_node_search - Search wiki content

Search Parameters:

• search_key (required): Keyword to search for
• docs_types: Filter by document type (doc, sheet, bitable, mindnote, file)
• owner_ids: Filter by document owner's open_id
• chat_ids: Filter by group where document is located
• count: Number of results (max 50, default 10)
• offset: Pagination offset

2. Document Content Access (文档内容访问)

Retrieve the actual content of documents:

Available MCP Tools:

• mcp__lark__docx_v1_document_rawContent - Get plain text content
• mcp__lark__wiki_v2_space_getNode - Get wiki node information

Parameters:

• document_id: The unique document identifier
• lang: Language for @mentions (0=Chinese, 1=English)

3. Permission Management (权限管理)

Check and manage document permissions:

Available MCP Tools:

• mcp__lark__drive_v1_permissionMember_create - Add permissions

Permission Roles:

• view: Read-only access
• edit: Can edit the document
• full_access: Full management access

Common Workflows

Workflow 1: Search and Read Documents

1. User: "搜索包含'产品路线图'的文档"
2. Assistant uses: mcp__lark__docx_builtin_search
   - search_key: "产品路线图"
   - count: 10
3. Display results with titles and URLs
4. User: "打开第一个文档"
5. Assistant uses: mcp__lark__docx_v1_document_rawContent
   - document_id: extracted from URL
6. Display document content

Workflow 2: Wiki Navigation

1. User: "Show me the engineering wiki"
2. Assistant uses: mcp__lark__wiki_v1_node_search
   - query: "engineering"
3. Display wiki pages and their hierarchy
4. User: "Get the content of the API documentation page"
5. Assistant uses: mcp__lark__wiki_v2_space_getNode
   - token: wiki_page_token
6. Display wiki content

Workflow 3: Document Discovery by Owner

1. User: "张三创建的所有文档"
2. First get user ID: mcp__lark__contact_v3_user_batchGetId
   - emails: ["zhangsan@company.com"]
3. Then search: mcp__lark__docx_builtin_search
   - search_key: ""
   - owner_ids: [user_open_id]
4. Display documents created by that user

Examples (示例)

Example 1: Basic Search (基础搜索)

User (Chinese):

帮我搜索包含'季度OKR'的文档

Assistant Response:

I'll search for documents containing '季度OKR'.

<uses mcp__lark__docx_builtin_search>
- search_key: "季度OKR"
- count: 10

Found 3 documents:

1. **2024 Q4 OKR Review** (docx)
   - Owner: 张三
   - Updated: 2024-01-10
   - Link: https://example.feishu.cn/docx/abc123

2. **Team OKR Template** (docx)
   - Owner: 李四
   - Updated: 2024-01-08
   - Link: https://example.feishu.cn/docx/def456

3. **Q1 2025 OKR Planning** (sheet)
   - Owner: 王五
   - Updated: 2024-01-12
   - Link: https://example.feishu.cn/sheets/ghi789

Would you like me to open any of these documents?

Example 2: Get Document Content (获取文档内容)

User (English):

Get the content of this document: https://example.feishu.cn/docx/abc123

Assistant Response:

I'll retrieve the content of that document.

<uses mcp__lark__docx_v1_document_rawContent>
- document_id: abc123

Here's the document content:

# 2024 Q4 OKR Review

## Engineering Team

### Objective 1: Improve System Reliability
- KR1: Achieve 99.9% uptime ✓ (Completed: 99.95%)
- KR2: Reduce MTTR to under 30 minutes ✓ (Completed: 25 min)
- KR3: Complete disaster recovery testing ⚠ (In Progress: 80%)

### Objective 2: Enhance Developer Productivity
- KR1: Reduce build time by 40% ✓ (Completed: 45% reduction)
- KR2: Implement automated code review ✓ (Completed)
- KR3: Launch internal developer portal ✗ (Delayed to Q1 2025)

[... more content ...]

Would you like me to summarize the key points or search for related documents?

Example 3: Wiki Search (知识库搜索)

User (Chinese):

在工程知识库中搜索关于'微服务架构'的文档

Assistant Response:

我会在工程知识库中搜索'微服务架构'相关内容。

<uses mcp__lark__wiki_v1_node_search>
- query: "微服务架构"

找到 5 个相关页面:

1. **微服务架构设计指南**
   - 路径: 工程文档 > 架构设计 > 微服务
   - 更新时间: 2024-01-10
   - [查看页面](https://example.feishu.cn/wiki/page1)

2. **服务拆分最佳实践**
   - 路径: 工程文档 > 架构设计 > 最佳实践
   - 更新时间: 2024-01-08
   - [查看页面](https://example.feishu.cn/wiki/page2)

3. **API Gateway 配置说明**
   - 路径: 工程文档 > 基础设施 > 网关
   - 更新时间: 2024-01-12
   - [查看页面](https://example.feishu.cn/wiki/page3)

需要我打开其中某个页面查看详细内容吗?

Important Notes

Permission Boundaries (权限边界)

• All operations respect the user's actual Lark permissions
• If a document is not accessible, you'll receive a permission error
• Cannot access private documents or groups the user is not a member of

Document ID Extraction (文档ID提取)

Document URLs have this format:

• Docx: https://*.feishu.cn/docx/{document_id}
• Wiki: https://*.feishu.cn/wiki/{wiki_token}
• Sheets: https://*.feishu.cn/sheets/{spreadsheet_id}
• Bitable: https://*.feishu.cn/base/{app_token}

Extract the ID/token from the URL when accessing content.

Search Limitations (搜索限制)

• Maximum 50 results per search (count parameter)
• Use pagination (offset) for more results
• Search is case-insensitive
• Results are sorted by relevance

Content Format (内容格式)

• docx_v1_document_rawContent returns plain text (no formatting)
• @mentions are converted to display names based on lang parameter
• For rich formatted content, use the API to get structured data

Error Handling

Common errors and solutions:

1. "Permission denied" (权限被拒绝)

   • The user doesn't have access to this document
   • Ask the document owner to grant permission

2. "Document not found" (文档未找到)

   • Document ID is invalid or document was deleted
   • Verify the document URL is correct

3. "Invalid token" (无效令牌)

   • MCP server authentication failed
   • Check LARK_APP_ID and LARK_APP_SECRET environment variables
   • Ensure OAuth token is valid

Tips for Effective Use

1. Start with search: Use keyword search before asking for specific documents
2. Be specific: Include relevant keywords to narrow down results
3. Use filters: Filter by document type or owner to find documents faster
4. Check permissions: Verify access before trying to read document content
5. Bilingual support: Works with both Chinese and English queries

Related Skills

• lark-messages: Access Lark messages and groups
• Future: lark-approval, lark-calendar, lark-drive