# Overview
The Feishu Open Platform has introduced MCP (Model Context Protocol) capabilities. Building on the existing OpenAPI, this feature encapsulates and optimizes interfaces for AI scenarios, upgrading traditional API calls into "tool-based capabilities" tailored for Large Language Models (LLMs). This results in more accurate, faster, and effective model interactions.
With MCP, AI can more naturally understand business semantics and perform data reading, writing, and business operations with lower integration costs.
Currently, the **Docs** scenario is supported. View the [supported MCP tools](https://open.larkoffice.com/document/mcp_open_tools/supported-tools). More scenarios (such as Base and Calendar) will be released later. You are welcome to [provide feedback](https://bytedance.larkoffice.com/share/base/form/shrcnGy0YNaviYYLjAX8zMqIW3f) on your usage scenarios 🥳
## Scenarios
User Type | Core Goal | Invocation Mode | How to Use
---|---|---|---
Agent User | Use natural language to easily access Feishu's powerful capabilities within your preferred AI tools (e.g., TRAE, Cursor), enabling automation for document creation, information retrieval, content management, and more. | **Remote Invocation Mode (Recommended)**: Grant AI permissions to access and manage your Feishu account data via personal authorization.
**Notice**：Currently supports "**Docs**" scenarios; more tools will be available soon. | [End user use Feishu remote MCP service](https://open.larkoffice.com/document/mcp_open_tools/end-user-call-remote-mcp-server)
Developer | Deeply integrate Feishu's open capabilities into your custom AI Agents via coding, intended for distribution to other users. | **Remote Invocation Mode (Recommended)**: Integrate and invoke Feishu MCP capabilities within your applications or services using either an app or user identity.
**Notice**：Currently supports "**Docs**" scenarios; more tools will be available soon. | [Developer call remote MCP service](https://open.feishu.cn/document/mcp_open_tools/developers-call-remote-mcp-server)
Developer | Deeply integrate Feishu's open capabilities into your custom AI Agents via coding, intended for distribution to other users. | **Local Invocation Mode (Not Recommended)**: Integrate and invoke Feishu MCP capabilities within your services using either an app or user identity.
**Notice**：- Supports all Feishu server-side OpenAPI capabilities; refer to the [Feishu MCP Tool List](https://github.com/larksuite/lark-openapi-mcp/blob/main/docs/reference/tool-presets/tools-zh.md).
- Developers must deploy the Feishu MCP service themselves. | [Overview of local OpenAPI MCP integration](https://open.larkoffice.com/document/mcp_open_tools/mcp-overview)
## Docs examples
The Feishu Docs scenario empowers AI with comprehensive document capabilities, bridging the "search, read, write" loop. It upgrades AI from a simple conversationalist to an efficient document collaboration expert, significantly enhancing knowledge accumulation and management efficiency.
Example Scenario | Effect | Run Record
---|---|---
Research today's major AI events and generate a Feishu Doc, attaching the original links for the news. |  | 
Based on my template [Weekly Work Summary Template](https://example.com), and combined with this week's MCP documentation content, generate a Feishu Doc for this week's work and review. |  | 
Generate a complete technical proposal based on the current code project, adding a sequence diagram and a flowchart. |  | 