Lark (Feishu) MCP Server

Overview of the MCP Server for Lark (Feishu)

The MCP server for Lark (also known as Feishu) bridges AI assistants with the rich suite of collaborative tools that Lark offers. By exposing a Model Context Protocol interface, it allows an AI assistant to read from and write to Lark spreadsheets, documents, and messages without needing direct API calls or manual authentication steps. This solves a common pain point for developers who want to embed AI into enterprise workflows: the friction of handling OAuth, permissions, and platform‑specific data formats.

At its core, the server provides a single powerful tool: write_excel. When invoked, this tool receives structured data and writes it to a specified sheet within Lark’s spreadsheet application. After the write operation, the server returns a direct link to the updated sheet, making it immediately accessible to users. The tool also supports granting access to collaborators by accepting an email address, ensuring that the right stakeholders can view or edit the data. Because it handles authentication internally (using Lark app credentials), developers can focus on business logic rather than token management.

Key capabilities include:

• Secure integration: The server requires only the Lark app ID and secret, which are used to obtain short‑lived access tokens via Lark’s OAuth flow.
• Read‑only permissions: The server is configured with , ensuring that it can read existing sheets while the write tool explicitly requests write access when needed.
• Link generation: After writing data, a shareable URL is returned, simplifying downstream steps such as notifying users or embedding the sheet in other applications.

Typical use cases span a range of enterprise scenarios. For example, an AI assistant could generate monthly sales reports and automatically populate them into a shared Lark spreadsheet, then send the link to the sales team. In project management, the assistant could pull task updates from a Lark document, analyze progress, and write summaries back to a dedicated sheet. Because the server operates via MCP, any AI platform that understands the protocol—Claude, Claude+, or custom models—can leverage these capabilities without bespoke SDKs.

What sets this server apart is its minimal footprint and tight coupling with Lark’s native data formats. Developers benefit from a single, well‑documented tool that handles all authentication nuances, while users enjoy seamless access to updated sheets directly within their familiar Lark environment. This tight integration reduces development time, lowers the barrier to entry for AI‑powered automation, and ensures that data stays within the secure confines of the enterprise’s existing collaboration platform.