feishu-dsh-plugin

Automating Feishu (Lark) workflows from an AI agent—sending messages, editing documents, manipulating bitable records, and managing drive permissions—used to mean stitching together API scripts or relying on third-party bridges. feishu-dsh-plugin plugs 25 native Feishu tools directly into DeepSeek Harness so the agent can drive the full Feishu surface with natural-language calls. It is pure Node with zero external dependencies and no Python requirement.

The plugin spans five core areas: chat and messaging (list groups, post text or rich messages, read history, recall), cloud documents (create, read/write blocks, insert tables, images, and locally rendered charts, resolve wiki nodes), bitable (create tables, manage fields, read and write records), classic spreadsheets (range read/write via the v2 API), and drive permissions (grant access, set link sharing). Documents can be created inside a designated knowledge-base space, and every artifact is auto-shared with a specified user so the operator retains ownership. Charts are rendered locally with a pure-Node PNG encoder, keeping all data on-host.

It is built for teams and individual developers who already run an enterprise self-built Feishu app and want their DeepSeek Harness agent to handle day-to-day Feishu tasks—auto-replies, generating weekly report documents, bulk-updating bitable records, or pulling spreadsheet data—without leaving the chat interface. Configure the App ID and Secret on the settings page, hit the test-connection button to verify credentials, and the agent is ready to work.