lark-mcp

The lark-mcp skill provides a standardized integration layer for interacting with the official Feishu (Lark) ecosystem through MCP tools. It is designed to simplify and unify operations across messaging, collaboration, document management, knowledge bases, and structured data systems like Bitable. The primary problem it solves is the complexity of working with multiple Lark APIs and inconsistent parameter conventions by providing a consistent tool-naming and request-structure framework.

This skill enables core capabilities such as sending and listing messages in chats, creating and managing group chats, and performing operations on Feishu Bitable (multi-dimensional tables) including creating, searching, and updating records. It also supports document-related workflows such as importing documents, retrieving raw document content, and searching both documents and knowledge base nodes. A key requirement is proper configuration for OAuth when accessing certain search functionalities, specifically document and wiki search tools, to avoid permission errors. The system also enforces strict conventions for tool naming (using hyphens in MCP tool identifiers), structured request fields (path, params, data), and identity modes through the useUAT flag to distinguish user-level and tenant-level access.

This skill is intended for developers and automation engineers building internal enterprise workflows on top of Lark. Typical use cases include automated notification systems, team collaboration bots, knowledge base search assistants, and structured data entry pipelines into Bitable. It is especially useful in enterprise environments where consistent access control, multi-tenant behavior, and structured API handling are critical. By enforcing strict parameter formats and highlighting common pitfalls such as JSON string formatting, correct ID types, and field naming differences, the skill reduces integration errors and improves reliability across Lark-based automation scenarios.