Feishu Releases Command-Line Tool Enabling AI Agents to Operate Full Suite

Feishu has launched an open-source command-line interface (CLI) tool, lark-cli, designed to allow both human users and AI agents to interact with the entire Feishu suite directly from a terminal. The tool integrates 19 pre-built AI Agent Skills and supports over 200 commands across 11 business domains.

Developer's desk with laptop showing lark-cli, tablet with Feishu app, and subtle AI network overlay.

The lark-cli is engineered for native AI agent integration, offering capabilities that are directly callable by mainstream AI tools. Each command has undergone testing and verification by AI agents to ensure compatibility and performance. The developers state that the tool can be set up and used for its first API call within three minutes.

Supported Functionality and AI Agent Skills

The tool provides extensive coverage across core Feishu applications:

• Calendar: Manages schedules, invitations, availability, and time suggestions.

• Instant Messaging: Handles message sending, replies, group chat management, history, searches, and media downloads.

• Docs: Facilitates creation, reading, updating, and searching of documents, materials, and whiteboards.

• Cloud Storage: Supports file uploads, downloads, document searches, knowledge base management, and comment handling.

• Multi-dimensional Table: Enables creation and management of tables, fields, records, views, dashboards, and data analysis.

• Spreadsheet: Allows creation, reading, writing, appending, finding, and exporting spreadsheet data.

• Task: Supports creating, querying, updating, and completing tasks, including subtasks, comments, and reminders.

• Knowledge Base: Manages knowledge spaces, nodes, and documents.

• Contacts: Provides search functionality for users by name, email, or phone number.

• Email: Manages browsing, searching, reading, sending, replying, forwarding, and drafting emails, alongside new email monitoring.

• Video Conferencing: Facilitates searching meeting records and querying summary artifacts such as summaries, to-dos, and verbatim transcripts.

Holographic icons of Feishu apps (Calendar, Messaging, Docs) interconnected, symbolizing comprehensive functionality.

The 19 integrated AI Agent Skills include lark-calendar, lark-im, lark-doc, lark-drive, lark-sheets, lark-base, lark-task, lark-mail, lark-contact, lark-wiki, lark-event for real-time event subscriptions, lark-vc, lark-whiteboard for DSL rendering, lark-minutes for metadata and AI artifacts, lark-openapi-explorer for API exploration, and lark-skill-maker for custom skill creation. Workflow-specific skills like lark-workflow-meeting-summary and lark-workflow-standup-report are also included.

Command Invocation Architecture

The lark-cli features a three-layer command invocation architecture:

1. Shortcut Commands: These are prefixed with + and offer human and AI-friendly encapsulations with built-in intelligent default values, table output, and dry-run previews. Examples include lark-cli calendar +agenda and lark-cli im +messages-send.

2. API Commands: Automatically generated from Feishu Open API metadata, these commands correspond directly to platform endpoints. Over 100 selected commands are available, such as lark-cli calendar calendars list.

3. Generic API Calls: This layer allows direct invocation of any Feishu Open Platform endpoint, covering more than 2,500 APIs. An example is lark-cli api GET /open-apis/calendar/v4/calendars.

Programmer typing on a keyboard, screen showing lark-cli command invocation architecture with three layers.

Installation and Usage

The lark-cli requires Node.js (npm/npx) and optionally Go v1.23+ and Python 3 for source builds. Installation is recommended via npm using npm install -g @larksuite/cli. AI Agent Skills can be installed using npx skills add larksuite/cli.

Configuration involves initializing application credentials with lark-cli config init and then logging in via OAuth using lark-cli auth login. The tool supports various authentication commands to manage login status, authorized scopes, and user accounts.

Advanced features include flexible output formats (JSON, pretty, table, NDJSON, CSV), pagination controls for retrieving large datasets, a dry-run mode for command previews, and schema introspection for exploring API structures.

The project's source code is available on GitHub at https://github.com/larksuite/cli.