Lark CLI Complete Guide: Control Feishu with Terminal and AI Agents

The Problem

You get a request from your manager: export all meetings from next week’s calendar, send a notification to 50 group chats, or batch-organize documents into the wiki. You end up clicking through the UI one by one, copying and pasting until your eyes glaze over.

Now with Lark CLI, a single terminal command handles it. Even better, it integrates directly with AI Agents like Claude Code — you describe what you want in natural language, and the AI does the rest. This guide covers everything from installation to advanced usage.

What Is Lark CLI?

One-Line Definition

Lark CLI is the official command-line tool for the Feishu (Lark) Open Platform, developed and open-sourced by the larksuite team under the MIT license. It covers 11 business domains, provides 200+ curated commands and 19 AI Agent Skills. As of March 2026, it has 1.7k GitHub Stars (source).

In plain terms: it’s a remote control for Feishu — manage calendars, messages, documents, spreadsheets, and more without opening a browser or app.

Lark CLI vs Direct API Calls

Three-Layer Architecture

┌─────────────────────────────────────────────┐
│  Layer 1: Shortcuts (+ prefix)              │
│  Human & AI friendly, covers common tasks   │
│  e.g., lark-cli calendar +agenda            │
├─────────────────────────────────────────────┤
│  Layer 2: API Commands                      │
│  Maps 1:1 to Feishu API endpoints, 100+     │
│  e.g., lark-cli calendar events list        │
├─────────────────────────────────────────────┤
│  Layer 3: Raw API Calls                     │
│  Covers all 2,500+ Open Platform endpoints  │
│  e.g., lark-cli api GET /open-apis/...      │
└─────────────────────────────────────────────┘

Prerequisites

Make sure you have Node.js (v18+ recommended, includes npm and npx):

node -v   # Should output v18.x.x or higher
npm -v    # Should output 9.x.x or higher

macOS users without Node.js: brew install node

Installation

The AI-Native Way: Let AI Install It For You

If you’re already using Claude Code or Codex, the simplest approach is to paste the README link directly:

Install everything from: https://github.com/larksuite/cli/blob/main/README.zh.md

The AI reads the docs, runs the install commands, and follows the prompts automatically. You just click “Authorize” when the browser pops up. You don’t even need to read the install guide — the AI reads it and does it for you.

Manual Way: Just 4 Commands

The entire setup takes 4 commands. config init automatically guides you through creating a Feishu app and configuring credentials — no need to manually visit the developer console.

# 1. Install Lark CLI
sudo npm install -g @larksuite/cli

# 2. Install AI Agent Skills (19 skills for Claude Code, Cursor, etc.)
npx skills add larksuite/cli -y -g

# 3. Initialize config (follow the prompts — auto-creates Feishu app)
lark-cli config init

# 4. Authorize (auto-selects common permission scopes)
lark-cli auth login --recommend

That’s it. Verify with lark-cli auth status and lark-cli doctor.

On macOS/Linux, sudo is needed for global npm install. Windows users can omit it. You can also install specific Skills only: npx skills add larksuite/cli -s lark-calendar -y

5 Commands Every Beginner Should Try

View Today’s Calendar

lark-cli calendar +agenda

Search Contacts

lark-cli contact +search-user --query "John"

Find Group Chats

lark-cli im +chat-search --query "Engineering"

Explore API Schema

lark-cli schema
lark-cli schema calendar.events.instance_view

Dry Run (Preview Without Executing)

lark-cli im +messages-send --chat-id oc_xxx --text "Test" --dry-run

Advanced Usage

Claude Code Integration

With Skills installed, Claude Code understands Feishu operations natively:

You: Check my meetings tomorrow
Claude Code: (runs lark-cli calendar +agenda for tomorrow)

You: Send a reminder to the dev chat about the 3pm meeting
Claude Code: (runs lark-cli im +messages-send)

You: Summarize who I chatted with last week
Claude Code: (runs lark-cli im +chat-messages-list across chats,
              then analyzes contacts and conversation summaries)

The real value: you don’t need to remember any commands — the AI picks the right combination for you.

Output Format Control

lark-cli calendar +agenda --format table
lark-cli base records list --params '...' --format csv
lark-cli calendar events list --params '...' --format pretty

Pagination

lark-cli base records list --params '...' --page-all
lark-cli im +chat-messages-list --chat-id oc_xxx --page-all --page-limit 5 --page-delay 500

Identity Switching

lark-cli calendar +agenda --as user
lark-cli im +messages-send --chat-id oc_xxx --text "Notice" --as bot

Raw API Calls

lark-cli api GET /open-apis/calendar/v4/calendars
lark-cli api POST /open-apis/im/v1/messages \
  --params '{"receive_id_type":"chat_id"}' \
  --data '{"receive_id":"oc_xxx","msg_type":"text","content":"{\"text\":\"hello\"}"}'

All 19 AI Agent Skills

FAQ

Permission Error (EACCES) During Install?

Add sudo or configure npm global directory:

mkdir ~/.npm-global && npm config set prefix '~/.npm-global'
# Add to ~/.zshrc: export PATH=~/.npm-global/bin:$PATH

Browser Won’t Open During Auth Login?

Use --no-wait and copy the URL manually:

lark-cli auth login --recommend --no-wait

What Does config init Need?

Just follow the terminal prompts. It automatically guides you through creating a Feishu app and configuring credentials — no preparation needed.

Scope/Permission Errors?

lark-cli auth check      # Check current permissions
lark-cli auth scopes     # List available scopes
lark-cli auth login --domain calendar,im  # Login with specific domains

Security Notes

1. Credential storage: Encrypted in OS keychain (macOS Keychain / Linux Secret Service)
2. AI Agent risk: Model hallucination and uncontrolled execution are inherent risks — use the connected Feishu bot as a private assistant only
3. Injection protection: Built-in input sanitization, but validate parameters in automation pipelines
4. Least privilege: Only authorize scopes you actually need
5. Regular audit: Use lark-cli auth list to review and lark-cli auth logout to clean up

Summary

Lark CLI solves a real pain point: Feishu operations are fragmented, and automation has a high barrier to entry.

Its three-layer architecture (Shortcuts → API Commands → Raw API) covers everything from simple to complex. The 19 AI Agent Skills turn Feishu into a capability module for AI tools like Claude Code and Cursor — not just a CLI wrapper, but a bridge between AI and enterprise collaboration.

Links

• Lark CLI GitHub (1.7k Stars, as of March 2026)
• Feishu Open Platform
• Feishu API Documentation

Further Reading

• Claude Code Complete Guide - Get started with Claude Code
• Claude Code Skills Top 20 - More useful Skills
• Claude Code Skill Development Guide - Build your own Skills
• AI Development Workflow - Integrate Lark CLI into your workflow
• Terminal Tools Guide - Upgrade your terminal

If this guide helped you, share your experience in the comments!

Updated March 2026 | Written for lark-cli v1.0.0