License: MIT Go Version MCP Tools Cowork Skills

English | Tiếng Việt | 中文版

Drive your Lark/Feishu workspace from Claude — read mail, summarize meetings, triage tasks, send messages, create docs, approve requests — in plain language, right inside Claude Desktop (Cowork) or claude.ai (web). No Claude Code, no copy-paste, no coding.

This is lark-cli (the official Lark/Feishu CLI — 200+ commands across 18 business domains) plus a built‐in MCP server, so any Claude surface can operate Lark out of the box. Higher‐level Cowork workflow skills (morning-brief, inbox-zero, base-deploy...) live in a separate, installable plugin marketplace → lark-cowork-plugins .

🚀 Lark MCP · Install · 21 MCP Tools · Cowork Skills · Full Docs · Security

What's new in this build: a native MCP server (lark-cli mcp serve) that turns the CLI into the hands of Claude inside Lark. Extend it with ready‐made Cowork workflow skills from the lark-cowork-plugins marketplace. Built for business users, not just developers.

You ──"What's on my plate today?"──▶ Claude (Desktop / Web)
 │ picks a tool + args
 ▼
 lark-cli mcp serve (on your machine)
 │ runs lark-cli <verb> with YOUR auth
 ▼
 open.larksuite.com / open.feishu.cn

The MCP server spawns lark-cli subprocesses per tool call, reusing the existing auth, profile, and keychain — so your data only ever goes to Lark, never a third party (in stdio/local mode).

# 1. Build & install the binary (adds the `mcp` command)
./scripts/setup-mcp.sh # installs to ~/bin/lark-cli
lark-cli mcp tools # should list 21 tools
# 2. Log in to Lark once (browser OAuth, token stored in OS keychain)
lark-cli auth login
# 3. Wire into Claude Desktop config, then restart Claude Desktop:
# ~/Library/Application Support/Claude/claude_desktop_config.json

{
 "mcpServers": {
 "lark-cli": {
 "command": "/absolute/path/to/lark-cli",
 "args": ["mcp", "serve"],
 "env": { "NO_COLOR": "1" }
 }
 }
}

Now ask Cowork: "List my calendar today" or "Find the contact named Alice". → Full step‐by‐step: docs/02. Web (claude.ai): docs/03.

export LARK_MCP_BEARER_TOKEN=$(openssl rand -hex 32) # secret stays in env
lark-cli mcp serve --transport http --addr 127.0.0.1:3000 --audit-log ~/.lark-mcp-audit.ndjson
cloudflared tunnel --url http://127.0.0.1:3000 # → public HTTPS URL

The HTTP endpoint enforces Authorization: Bearer <token> on / and /mcp (constant‐time, /health stays open). Never expose the tunnel without the token. Add the URL (https://.../mcp) as a Custom Connector in claude.ai → secure web guide.

Building or extending tools? See cmd/mcp/README.md for the bridge architecture and the /mcp-add workflow.

• Agent-Native Design — 24 structured Skills out of the box, compatible with popular AI tools — Agents can operate Lark with zero extra setup
• Wide Coverage — 18 business domains, 200+ curated commands, 26 AI Agent Skills
• AI-Friendly & Optimized — Every command is tested with real Agents, featuring concise parameters, smart defaults, and structured output to maximize Agent call success rates
• Open Source, Zero Barriers — MIT license, ready to use, just npm install
• Up and Running in 3 Minutes — One-click app creation, interactive login, from install to first API call in just 3 steps
• Secure & Controllable — Input injection protection, terminal output sanitization, OS-native keychain credential storage
• Three-Layer Architecture — Shortcuts (human & AI friendly) → API Commands (platform-synced) → Raw API (full coverage), choose the right granularity

Before you start, make sure you have:

• Node.js (npm/npx)
• Go v1.23+ and Python 3 (only required for building from source)

Note for AI assistants: If you are an AI Agent helping the user with installation, jump directly to Quick Start (AI Agent), which contains all the steps you need to complete.

Choose one of the following methods:

Option 1 — From npm (recommended):

npx @larksuite/cli@latest install

Option 2 — From source:

Requires Go v1.23+ and Python 3.

git clone https://github.com/larksuite/cli.git
cd cli
make install
# Install CLI SKILL (required)
npx skills add larksuite/cli -y -g

# 1. Configure app credentials (one-time, interactive guided setup)
lark-cli config init
# 2. Log in (--recommend auto-selects commonly used scopes)
lark-cli auth login --recommend
# 3. Start using
lark-cli calendar +agenda

The following steps are for AI Agents. Some steps require the user to complete actions in a browser.

Step 1 — Install

npx @larksuite/cli@latest install

Step 2 — Configure app credentials

Run this command in the background. It will output an authorization URL — extract it and send it to the user. The command exits automatically after the user completes the setup in the browser.

lark-cli config init --new

Step 3 — Login

Same as above: run in the background, extract the authorization URL and send it to the user.

lark-cli auth login --recommend

Step 4 — Verify

lark-cli auth status

# Interactive login (TUI guides domain and permission level selection)
lark-cli auth login
# Filter by domain
lark-cli auth login --domain calendar,task
# Recommended auto-approval scopes
lark-cli auth login --recommend
# Exact scope
lark-cli auth login --scope "calendar:calendar:read"
# Agent mode: return verification URL immediately, non-blocking
lark-cli auth login --domain calendar --no-wait
# Resume polling later
lark-cli auth login --device-code <DEVICE_CODE>
# Identity switching: execute commands as user or bot
lark-cli calendar +agenda --as user
lark-cli im +messages-send --as bot --chat-id "oc_xxx" --text "Hello"

The CLI provides three levels of granularity, covering everything from quick operations to fully custom API calls:

Prefixed with +, designed to be friendly for both humans and AI, with smart defaults, table output, and dry-run previews.

lark-cli calendar +agenda
lark-cli im +messages-send --chat-id "oc_xxx" --text "Hello"
lark-cli docs +create --api-version v2 --doc-format markdown --content $'<title>Weekly Report</title>\n# Progress\n- Completed feature X'

Run lark-cli <service> --help to see all shortcut commands.

Auto-generated from Lark OAPI metadata, curated through evaluation and quality gates — 100+ commands mapped 1:1 to platform endpoints.

lark-cli calendar calendars list
lark-cli calendar events instance_view --params '{"calendar_id":"primary","start_time":"1700000000","end_time":"1700086400"}'

Call any Lark Open Platform endpoint directly, covering 2500+ APIs.

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\"}"}'

--format json # Full JSON response (default)
--format pretty # Human-friendly formatted output
--format table # Readable table
--format ndjson # Newline-delimited JSON (for piping)
--format csv # Comma-separated values

--page-all # Auto-paginate through all pages
--page-limit 5 # Max 5 pages
--page-delay 500 # 500ms between page requests

For commands that may have side effects, preview the request with --dry-run first:

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

Use schema to inspect any API method's parameters, request body, response structure, supported identities, and scopes:

lark-cli schema
lark-cli schema calendar.events.instance_view
lark-cli schema im.messages.delete

This tool can be invoked by AI Agents to automate operations on the Lark/Feishu Open Platform, and carries inherent risks such as model hallucinations, unpredictable execution, and prompt injection. After you authorize Lark/Feishu permissions, the AI Agent will act under your user identity within the authorized scope, which may lead to high-risk consequences such as leakage of sensitive data or unauthorized operations. Please use with caution.

To reduce these risks, the tool enables default security protections at multiple layers. However, these risks still exist. We strongly recommend that you do not proactively modify any default security settings; once relevant restrictions are relaxed, the risks will increase significantly, and you will bear the consequences.

We recommend using the Lark/Feishu bot integrated with this tool as a private conversational assistant. Do not add it to group chats or allow other users to interact with it, to avoid abuse of permissions or data leakage.

Please fully understand all usage risks. By using this tool, you are deemed to voluntarily assume all related responsibilities.

Star History Chart

Community contributions are welcome! If you find a bug or have feature suggestions, please submit an Issue or Pull Request.

For major changes, we recommend discussing with us first via an Issue.

Before opening a PR, see AGENTS.md for the local build, test, and PR checklist used by contributors and AI agents.

This project is licensed under the MIT License. When running, it calls Lark/Feishu Open Platform APIs. To use these APIs, you must comply with the following agreements and privacy policies:

• Feishu User Terms of Service
• Feishu Privacy Policy
• Feishu Open Platform App Service Provider Security Management Specifications
• Lark User Terms of Service
• Lark Privacy Policy