What is Lark CLI
Lark CLI is an open-source command-line tool from Feishu, designed specifically for AI Agents, enabling Claude Code、Codex and other AI assistants to operate Feishu directly. The tool covers 11 major business domains including calendar, messaging, docs, and Base, providing 200+ commands and 19 out-of-the-box Agent Skills. Users only need natural language instructions, and AI can automatically check schedules, send messages, write docs, and create spreadsheets. The launch of Lark CLI marks Feishu's comprehensive transformation toward the Agent era.
Main features of Lark CLI
- Calendar: Supports viewing schedules, creating meeting events, inviting attendees, querying free/busy status, and obtaining time suggestions.
- Instant Messaging: Supports sending and replying to messages, creating and managing group chats, viewing chat history and threads, searching historical messages, and downloading media files.
- Docs: Supports creating, reading, updating, and searching document content, as well as reading and writing materials and whiteboards, enabling seamless format conversion between Markdown and Feishu Docs.
- Cloud Space: Supports uploading and downloading various files, searching documents and knowledge base content, and managing file comments.
- Base: Supports creating and managing Base, managing fields and records, configuring views and dashboards, and performing data aggregation analysis.
- Sheets: Supports creating, reading, writing, appending, data lookup, and export operations for spreadsheets.
- Tasks: Supports creating, querying, updating, and completing tasks, managing task lists and subtasks, and setting comments and reminders.
- Wiki: Supports creating and managing knowledge spaces, and maintaining knowledge nodes and document structures.
- Contacts: Supports searching for users by name, email, or phone number, and obtaining detailed user information.
- Mail: Supports browsing, searching, and reading emails, sending, replying to, and forwarding emails, managing drafts, and listening for new email notifications.
- Video Meetings: Supports searching meeting records, querying meeting minutes, and accessing meeting recordings.
Key information and usage requirements of Lark CLI
- Product Positioning: A command-line tool for the Feishu/Lark Open Platform, enabling both humans and AI Agents to operate Feishu in the terminal.
- Coverage: Covers 11 major business domains, providing 200+ selected commands and 19 AI Agent Skills.
- Design Philosophy: Agent-native design, optimized specifically for AI, with every command validated through actual Agent testing.
- Architecture Design: Three-layer invocation architecture: shortcut commands → API commands → raw API calls.
- Security Features: Includes protection mechanisms such as input anti-injection, terminal output sanitization, and OS-native keychain credential storage.
- OpenClaw Integration: The "Feishu Official OpenClaw Plugin" with all built-in CLI capabilities is coming soon.
- Environment Requirements: Node.js (npm / npx required); Go v1.23+ and Python 3 (only required when compiling from source).
Core advantages of Lark CLI
- Agent-native design: 19 Skills are provided out of the box, compatible with mainstream AI tools such as Claude Code, Codex, and Cursor, so Agents can operate Lark without additional adaptation.
- AI-friendly tuning: Every command has been verified through real Agent testing, providing friendlier parameters, smart defaults, and structured output, greatly improving the Agent call success rate.
- Comprehensive coverage: Covers 11 major business domains and 200+ selected commands, covering core office scenarios such as calendar, messages, docs, Base, email, tasks, and meetings.
- Three-tier flexible architecture: Supports shortcut commands (human-friendly) -> API commands (platform-synced) -> generic calls (full API coverage), with granularity chosen as needed.
How to use Lark CLI
- Install the CLI: Run
npm install -g @larksuite/cliin the terminal to globally install the Lark command-line tool. - Install Skills: Run
npx skills add larksuite/cli -y -gto install the 19 required AI Agent Skills. - Configure app credentials: Run
lark-cli config init --newto start the interactive guide, create a Lark app, and obtain the App ID and Secret. - Log in and authorize: Run
lark-cli auth login --recommendto complete OAuth authorization, supporting browser QR code scanning or link confirmation. - Restart the AI tool: After installation, be sure to restart AI Agent tools such as Claude Code and Codex to ensure the Skills are fully loaded.
- Start using: Issue instructions to the AI in natural language, such as "check my schedule today" or "create a weekly report document", and the AI will automatically call the CLI to perform the operation.
Lark CLI project address
- GitHub repository：https://github.com/larksuite/cli
Comparison of Lark CLI with similar competitors
| Comparison dimension | Lark CLI | GitHub Copilot CLI | Microsoft Graph CLI |
|---|---|---|---|
| Developer | ByteDance (Lark) | GitHub / Microsoft | Microsoft |
| Product positioning | Native CLI for an office platform, designed specifically for AI Agents, enabling AI to directly operate the Lark office suite | Native CLI for a code platform, enabling AI to operate GitHub repositories, PRs, and Issues | General-purpose CLI for an enterprise platform, enabling developers/AI to operate the entire Microsoft 365 ecosystem |
| Core goal | Make AI Agents become Lark "digital employees" and automate office workflows | Let AI automatically handle development processes such as code review, issue triage, and PR merging | Let AI/scripts automatically manage enterprise users, email, calendars, SharePoint, etc. |
| AI Agent native design | ✅ Yes, 19 out-of-the-box Skills, with parameters and outputs optimized specifically for AI | ✅ Yes, GitHub Copilot natively supports CLI mode | ❌ No, traditional developer tools, AI requires additional adaptation |
| Coverage | Full Lark ecosystem: Calendar, Docs, Base, Messages, Meetings, Tasks, Mail, Wiki | Full GitHub ecosystem: code repositories, PRs, Issues, Actions, code review | Full Microsoft 365 ecosystem: Teams, Outlook, SharePoint, Azure AD, OneDrive |
| Number of commands | 200+ curated commands | Dozens of core commands (gh + copilot extension) | Hundreds of commands (covering the full Microsoft Graph API) |
| Installation method | npm install -g @larksuite/cli |
gh copilot(GitHub CLI extension) |
winget install Microsoft.GraphCLI |
| Open source license | MIT License, fully open source | Partially open source (GitHub CLI is open source, Copilot features are closed source) | Open source (MIT), but depends on the Microsoft commercial ecosystem |
| AI tool compatibility | Native support Claude Code、Codex、Cursor、OpenClaw | Native support GitHub Copilot, can connect to other AI via MCP | Requires indirect support through PowerShell or community toolchains |
| Barrier to use | Extremely low: 3-minute installation, no review required, can be operated with natural language | Low: requires a GitHub account and Copilot subscription | High: requires Azure AD permission configuration, enterprise-level complexity |
Application scenarios of Lark CLI
- Meeting automation: After a meeting ends, the Agent automatically extracts to-dos from the meeting minutes, creates tasks, and sends relevant documents to attendees, with no manual organizing required.
- Intelligent document collaboration : Users use natural language instructions to have AI create a draft document in Lark, provide revision suggestions through comments, and the AI reads the comments and directly revises the content.
- Batch message notifications: Through a single natural language instruction, the Agent automatically finds a specific group of people in the contacts (such as department members), sends personalized messages, and greatly improves internal communication efficiency.
- Seamless Markdown conversion: Technical documents or weekly reports written in AI tools such as Claude Code can be converted with one click into fully formatted Lark documents, preserving tables, code blocks, multi-level headings, and other formatting.
- Intelligent schedule management: Let the Agent query today's schedule, check busy/free status, automatically create meetings, and invite relevant people, enabling a full set of schedule operations through voice or text commands.