Lark CLI — 19 AI Agent Skills for Lark/Feishu

What it is

Lark CLI is a command-line tool for the Lark/Feishu Open Platform, designed for both humans and AI agents. It provides 200+ commands covering 11 core business domains and packages them into 19 AI agent skills that coding assistants like Claude Code, Cursor, and OpenAI Codex can use directly. Lark (international) and Feishu (China) are ByteDance's workplace collaboration platform, used by millions of knowledge workers.

The toolkit is MIT licensed and available as an npm package. It covers messaging, documents, spreadsheets, calendar, tasks, email, wiki, drive, meetings, contacts, and more. Each domain has a dedicated skill that exposes curated commands for common operations, plus an OpenAPI Explorer for accessing any raw Lark API endpoint.

How it saves time or tokens

Before Lark CLI, automating Lark operations required writing custom code against the Open Platform REST APIs: handling OAuth, constructing request bodies, parsing responses, and managing pagination. Each integration took hours of development time. With Lark CLI, an AI agent installs the skills and immediately has structured access to all Lark operations through simple CLI commands.

The 19 pre-built skills also reduce token consumption. Instead of an AI agent reasoning through raw API documentation, it uses the skill's curated command interface. The skill handles API details internally, so the agent's prompt only needs the command name and parameters, not the full API specification.

How to use

1. Install Lark CLI and all 19 skills:

```bash

npm install -g @larksuite/cli

npx skills add larksuite/cli -y -g

```

1. Configure and authenticate:

```bash

lark-cli config init

lark-cli auth login --recommend

```

1. Start using any skill:

```bash

lark-cli calendar +agenda

lark-cli im +messages-send --chat-id 'oc_xxx' --text 'Hello'

lark-cli task +create --summary 'Review PR'

```

Example

The 19 skills organized by domain:

A typical multi-skill workflow:

# Morning routine automated by an AI agent:
# 1. Check today's agenda
lark-cli calendar +agenda

# 2. Get pending tasks
lark-cli task +list --status open

# 3. Generate and post standup report
lark-cli standup +run --chat-id 'oc_teamchat'

# 4. Check new messages
lark-cli im +messages-list --chat-id 'oc_teamchat' --page-size 5

Related on TokRepo

• AI Tools for Automation — Browse automation tools and compare with Lark CLI's approach.
• AI Tools for Task Management — Explore task management tools that complement Lark's built-in system.

Common pitfalls

• Trying to use skills before completing authentication. All 19 skills depend on the Shared skill's auth configuration. Run lark-cli config init and lark-cli auth login --recommend first. Every other command will fail without valid credentials.
• Not configuring the correct region. Lark (international) and Feishu (China) use different API endpoints. If your workspace is on Feishu but the CLI targets Lark, all requests fail. Set the correct endpoint during config init.
• Installing the CLI without installing the skills. The npm package provides the CLI binary, but the 19 skills are installed separately via npx skills add larksuite/cli -y -g. Without this step, only basic CLI commands are available.