lark

name	lark
description	Work with Lark Docs, Sheets, Bitable (database), and Drive. Use this skill when the user requests operations on Lark documents, spreadsheets, or databases.
allowed-tools	Bash(lark_cli:*)

Lark Integration Skill

Overview

This skill enables you to interact with Lark's productivity suite:

• Docs - Read and create documents
• Sheets - Read, write, and append spreadsheet data
• Bitable - Query and manage database records (like Airtable)
• Drive - List and manage files/folders
• Sharing - Share files with users via email

Authentication

The CLI uses environment variables for authentication:

• LARK_APP_ID - Your Lark app ID
• LARK_APP_SECRET - Your Lark app secret

These are pre-configured in the environment.

Available Commands

Documents

# Get document metadata
lark_cli get-doc --document-id "<doc_token>"

# Read document content (returns blocks)
lark_cli read-doc --document-id "<doc_token>"

# Create a new document
lark_cli create-doc --title "Meeting Notes"
lark_cli create-doc --title "Report" --folder-token "<folder_token>"

Sheets

# Get spreadsheet metadata
lark_cli get-sheet --spreadsheet-id "<sheet_token>"

# Read a range of cells
lark_cli read-range --spreadsheet-id "<sheet_token>" --sheet-id "Sheet1" --range "A1:C10"

# Write values to a range (JSON 2D array)
lark_cli write-range --spreadsheet-id "<sheet_token>" --sheet-id "Sheet1" --range "A1:B2" --values '[["Name","Age"],["John",30]]'

# Append rows to a sheet
lark_cli append-rows --spreadsheet-id "<sheet_token>" --sheet-id "Sheet1" --values '[["Jane",25],["Bob",35]]'

Bitable (Database)

Bitable is Lark's database product (similar to Airtable/Notion databases).

# List tables in a bitable app
lark_cli list-tables --app-token "<app_token>"

# Get table metadata (field definitions)
lark_cli get-table --app-token "<app_token>" --table-id "<table_id>"

# Query records (with optional filter)
lark_cli query-records --app-token "<app_token>" --table-id "<table_id>"
lark_cli query-records --app-token "<app_token>" --table-id "<table_id>" --filter 'CurrentValue.[Status]="Done"' --page-size 50

# Create a record (fields as JSON object)
lark_cli create-record --app-token "<app_token>" --table-id "<table_id>" --fields '{"Name":"John","Status":"Active"}'

# Update a record
lark_cli update-record --app-token "<app_token>" --table-id "<table_id>" --record-id "<record_id>" --fields '{"Status":"Completed"}'

# Delete a record
lark_cli delete-record --app-token "<app_token>" --table-id "<table_id>" --record-id "<record_id>"

Field Types:

Type	Example Value
Text	"Hello"
Number	123 or 45.67
Checkbox	true / false
Date	1672531200000 (unix ms)
Single Select	"Option A"
Multi Select	["Option A", "Option B"]
Person	[{"id": "ou_xxx"}]

Drive

# List files (root folder)
lark_cli list-files

# List files in a specific folder
lark_cli list-files --folder-token "<folder_token>" --page-size 50

# Get file metadata
lark_cli get-file --file-token "<file_token>"

# Create a folder
lark_cli create-folder --name "Projects" --parent-token "<parent_folder_token>"

Sharing

Share documents, sheets, or bitable apps with users:

# Share a document with a user by open_id (default, edit access)
lark_cli share-file --token "<doc_token>" --file-type docx --member-id "ou_xxx"

# Share with view-only access
lark_cli share-file --token "<doc_token>" --file-type docx --member-id "ou_xxx" --perm view

# Share by email instead of open_id
lark_cli share-file --token "<doc_token>" --file-type docx --member-type email --member-id "[e-mail]"

# Share a spreadsheet
lark_cli share-file --token "<sheet_token>" --file-type sheet --member-id "ou_xxx"

# Share a bitable with full access
lark_cli share-file --token "<app_token>" --file-type bitable --member-id "ou_xxx" --perm full_access

Parameters:

• --token - The file/doc token from the create response
• --file-type - docx, sheet, bitable, file, folder
• --member-type - openid (default), email, or userid
• --member-id - The user's open_id (like "ou_xxx"), email, or user_id
• --perm - view, edit (default), or full_access

Important: When you create a document/sheet/bitable, it is owned by the bot app. You must share it with the user for them to access it.

• Use the user's open_id from "Lark Open IDs" in the User Context section
• If not available, the user needs to link their Lark account at dowhiz.com first

Getting Tokens from URLs

Lark URLs contain tokens you'll need:

URL Pattern	Token Location
https://xxx.larksuite.com/docx/ABC123	Document: ABC123
https://xxx.larksuite.com/sheets/ABC123	Spreadsheet: ABC123
https://xxx.larksuite.com/base/ABC123	Bitable App: ABC123
https://xxx.larksuite.com/drive/folder/ABC123	Folder: ABC123

Example Workflows

Read and Summarize a Document

# 1. Read the document
lark_cli read-doc --document-id "doxcn4RM1Lxxx"

# 2. Parse the JSON response to extract text blocks
# 3. Summarize the content

Log Data to a Bitable

# 1. Get table structure
lark_cli get-table --app-token "appXXX" --table-id "tblYYY"

# 2. Create a new record
lark_cli create-record --app-token "appXXX" --table-id "tblYYY" --fields '{"Task":"Review PR","Status":"Pending","Due":"2024-03-15"}'

Update Spreadsheet from User Request

# 1. Read current data
lark_cli read-range --spreadsheet-id "shtXXX" --sheet-id "Sheet1" --range "A1:D10"

# 2. Append new rows
lark_cli append-rows --spreadsheet-id "shtXXX" --sheet-id "Sheet1" --values '[["New Item",100,"Active"]]'

Error Handling

Common errors:

• Permission denied: User needs to grant the Lark app access to the document/table
• Not found: Check the token is correct and the resource exists
• Invalid fields: Ensure field names match exactly (case-sensitive) and values match field types

Response Format

All commands output JSON. Parse the response to extract:

• data - The actual content/records
• code - 0 for success, non-zero for errors
• msg - Error message if failed