openclaw-lark-integration

name	openclaw-lark-integration
description	Official Lark/Feishu plugin for OpenClaw that enables AI agents to interact with Lark workspaces including messages, docs, bases, calendars, and tasks
triggers	["integrate OpenClaw with Lark","set up Feishu plugin for OpenClaw","connect my AI agent to Lark workspace","configure OpenClaw Lark permissions","create a Lark bot with OpenClaw","troubleshoot OpenClaw Lark integration","manage OpenClaw Lark group settings","use OpenClaw with Feishu documents"]

OpenClaw Lark Integration

Skill by ara.so — Hermes Skills collection.

Official Lark/Feishu plugin for OpenClaw that seamlessly connects AI agents to Lark workspaces. Enables reading and writing messages, managing docs, bases, sheets, calendars, and tasks with built-in security controls and interactive cards.

What It Does

The OpenClaw Lark plugin allows AI agents to:

• Messaging: Read message history (groups/DMs/threads), send/reply to messages, search, download attachments
• Documents: Create, update, and read Lark docs
• Base: Full CRUD operations on bases, tables, fields, records with advanced filtering
• Sheets: Create, edit, and view spreadsheets
• Calendar: Manage calendars, events, attendees, and check free/busy status
• Tasks: Create, query, update, complete tasks and manage subtasks/comments
• Interactive Cards: Real-time status updates with streaming responses
• Security: Built-in permission policies and per-group configuration

Installation

Prerequisites

• Node.js v22 or higher
• OpenClaw version 2026.2.26 or higher

Check OpenClaw version:

openclaw -v

If below required version, upgrade:

npm install -g openclaw

Install Plugin

npm install -g @larksuite/openclaw-lark

Or with pnpm:

pnpm add -g @larksuite/openclaw-lark

Configuration

1. Create Lark/Feishu App

1. Go to Lark Open Platform or Feishu Open Platform
2. Create a new custom app
3. Configure bot capabilities and permissions
4. Obtain App ID and App Secret

2. Required Permissions

The app needs these permission scopes:

Messaging:

• im:message - Send messages
• im:message:read_as_user - Read messages as user
• im:chat - Access chat information

Documents:

• docx:document - Manage docs
• drive:drive - Access drive files

Base:

• bitable:app - Manage base apps
• bitable:record - Manage records

Sheets:

• sheets:spreadsheet - Manage spreadsheets

Calendar:

• calendar:calendar - Manage calendars
• calendar:event - Manage events

Tasks:

• task:task - Manage tasks

3. Configure Environment Variables

Create or update your OpenClaw config file with Lark credentials:

// openclaw.config.ts
export default {
  channels: {
    lark: {
      appId: process.env.LARK_APP_ID,
      appSecret: process.env.LARK_APP_SECRET,
      verificationToken: process.env.LARK_VERIFICATION_TOKEN,
      encryptKey: process.env.LARK_ENCRYPT_KEY,
      
      // Optional: Security policies
      policies: {
        allowPrivateChat: true,
        allowGroupChat: false, // Disable by default for security
        groupAllowlist: [], // Whitelist specific groups
      },
      
      // Optional: Default settings
      enableInteractiveCards: true,
      enableStreamingResponse: true,
    }
  }
}

Set environment variables:

export LARK_APP_ID="your_app_id"
export LARK_APP_SECRET="your_app_secret"
export LARK_VERIFICATION_TOKEN="your_verification_token"
export LARK_ENCRYPT_KEY="your_encrypt_key"

4. Start OpenClaw with Lark Channel

openclaw start --channel lark

Or specify config file:

openclaw start --config openclaw.config.ts

Key Commands

OpenClaw CLI Commands

# Start with Lark channel
openclaw start --channel lark

# Check version
openclaw -v

# View help
openclaw --help

# Stop OpenClaw
openclaw stop

Managing the Bot

Once running, interact with your bot in Lark/Feishu by:

• Sending direct messages
• @mentioning in allowed groups
• Using configured skills and prompts

API Usage Patterns

Sending Messages

The plugin automatically handles message sending through OpenClaw's unified interface:

// Example skill that sends a Lark message
import { SkillContext } from 'openclaw';

export async function sendLarkMessage(context: SkillContext, message: string) {
  // OpenClaw automatically routes to Lark channel
  await context.channel.sendMessage({
    text: message,
    chatId: context.chatId
  });
}

Reading Messages

export async function getRecentMessages(context: SkillContext, limit: number = 10) {
  const messages = await context.channel.getMessages({
    chatId: context.chatId,
    limit: limit
  });
  
  return messages.map(msg => ({
    sender: msg.sender,
    content: msg.content,
    timestamp: msg.timestamp
  }));
}

Working with Documents

export async function createDocument(
  context: SkillContext, 
  title: string, 
  content: string
) {
  const doc = await context.channel.lark.createDoc({
    title: title,
    content: content,
    folderToken: context.workspace.defaultFolder
  });
  
  return {
    docId: doc.docToken,
    url: doc.url
  };
}

Managing Base Records

export async function addBaseRecord(
  context: SkillContext,
  baseId: string,
  tableId: string,
  fields: Record<string, any>
) {
  const record = await context.channel.lark.base.createRecord({
    appToken: baseId,
    tableId: tableId,
    fields: fields
  });
  
  return record;
}

export async function queryBaseRecords(
  context: SkillContext,
  baseId: string,
  tableId: string,
  filter?: string
) {
  const records = await context.channel.lark.base.listRecords({
    appToken: baseId,
    tableId: tableId,
    filter: filter, // e.g., "AND(CurrentValue.[Status] = 'Active')"
    pageSize: 100
  });
  
  return records.items;
}

Calendar Operations

export async function createCalendarEvent(
  context: SkillContext,
  summary: string,
  startTime: string,
  endTime: string,
  attendees?: string[]
) {
  const event = await context.channel.lark.calendar.createEvent({
    summary: summary,
    startTime: { timestamp: startTime },
    endTime: { timestamp: endTime },
    attendees: attendees?.map(email => ({ email }))
  });
  
  return {
    eventId: event.eventId,
    htmlLink: event.htmlLink
  };
}

Task Management

export async function createTask(
  context: SkillContext,
  summary: string,
  description: string,
  dueDate?: string
) {
  const task = await context.channel.lark.task.createTask({
    summary: summary,
    description: description,
    due: dueDate ? { date: dueDate } : undefined
  });
  
  return task;
}

Security Configuration

Permission Policies

Configure access control in your config file:

export default {
  channels: {
    lark: {
      // ... credentials
      
      policies: {
        // Allow private chats (default: true)
        allowPrivateChat: true,
        
        // Disable all group chats (recommended for security)
        allowGroupChat: false,
        
        // Or allow specific groups only
        allowGroupChat: true,
        groupAllowlist: [
          'oc_xxxxxxxxxxxxx', // Group chat ID
          'oc_yyyyyyyyyyyyy'
        ],
        
        // Require confirmation for sensitive operations
        requireConfirmation: {
          deleteDocument: true,
          deleteBaseRecord: true,
          sendMessageToGroup: true
        }
      }
    }
  }
}

Per-Group Configuration

export default {
  channels: {
    lark: {
      // ... credentials
      
      groupSettings: {
        'oc_xxxxxxxxxxxxx': {
          enabled: true,
          allowedSkills: ['search', 'summarize'], // Restrict skills
          customSystemPrompt: 'You are a helpful assistant for the engineering team.',
          maxTokens: 4000
        }
      }
    }
  }
}

Interactive Cards and Streaming

Enable Streaming Responses

export default {
  channels: {
    lark: {
      // ... credentials
      enableStreamingResponse: true,
      enableInteractiveCards: true
    }
  }
}

Streaming automatically shows:

• 🤔 Thinking indicator
• 📝 Generating status with live text
• ✅ Complete notification

Custom Interactive Cards

export async function sendCardWithActions(context: SkillContext) {
  await context.channel.sendCard({
    header: {
      title: 'Confirm Action',
      template: 'blue'
    },
    elements: [
      {
        tag: 'div',
        text: {
          tag: 'plain_text',
          content: 'Do you want to proceed with this operation?'
        }
      },
      {
        tag: 'action',
        actions: [
          {
            tag: 'button',
            text: { tag: 'plain_text', content: 'Confirm' },
            type: 'primary',
            value: { action: 'confirm' }
          },
          {
            tag: 'button',
            text: { tag: 'plain_text', content: 'Cancel' },
            type: 'default',
            value: { action: 'cancel' }
          }
        ]
      }
    ]
  });
}

Common Patterns

Message Handler Skill

import { Skill } from 'openclaw';

export const messageHandlerSkill: Skill = {
  name: 'lark-message-handler',
  description: 'Handle incoming Lark messages',
  
  async execute(context) {
    const { message, sender } = context;
    
    // Process message
    if (message.includes('help')) {
      return await context.reply('How can I assist you?');
    }
    
    // Search message history
    if (message.startsWith('search:')) {
      const query = message.substring(7);
      const results = await context.channel.searchMessages({ query });
      return results;
    }
    
    // Default response
    return await context.reply('Message received');
  }
};

Document Automation

export async function createWeeklyReport(context: SkillContext) {
  const today = new Date();
  const title = `Weekly Report - ${today.toISOString().split('T')[0]}`;
  
  // Gather data from Base
  const tasks = await context.channel.lark.base.listRecords({
    appToken: process.env.LARK_TASK_BASE_ID!,
    tableId: 'tblxxxxxxxx',
    filter: "AND(CurrentValue.[CompletedAt] >= DATE_SUB(TODAY(), 7))"
  });
  
  // Create formatted document
  const content = `
# ${title}

## Completed Tasks
${tasks.items.map(t => `- ${t.fields.Name}`).join('\n')}

## Summary
Total tasks completed: ${tasks.items.length}
`;
  
  const doc = await context.channel.lark.createDoc({
    title,
    content
  });
  
  return doc.url;
}

Batch Operations

export async function batchUpdateRecords(
  context: SkillContext,
  baseId: string,
  tableId: string,
  updates: Array<{ recordId: string; fields: Record<string, any> }>
) {
  // Lark API supports batch operations
  const result = await context.channel.lark.base.batchUpdateRecords({
    appToken: baseId,
    tableId: tableId,
    records: updates
  });
  
  return {
    updated: result.records.length,
    records: result.records
  };
}

Troubleshooting

Bot Not Responding

Issue: Bot doesn't reply to messages

Solutions:

1. Verify OpenClaw is running: openclaw status
2. Check credentials are set correctly
3. Ensure bot has required permissions in Lark admin console
4. Verify group is in allowlist if allowGroupChat is true
5. Check logs: openclaw logs --channel lark

Permission Errors

Issue: "Permission denied" errors

Solutions:

1. Review required scopes in Lark app settings
2. Re-authorize the app after adding permissions
3. Check if user has necessary workspace permissions
4. Verify app is published (not in development mode)

Message Not Sent

Issue: Messages fail to send

Solutions:

1. Check if chat ID is valid
2. Verify bot is added to the group chat
3. Ensure im:message permission is granted
4. Check message format and size limits

Configuration Not Loading

Issue: Config changes not taking effect

Solutions:

1. Restart OpenClaw after config changes
2. Verify config file path: openclaw start --config ./path/to/config.ts
3. Check for TypeScript/JSON syntax errors
4. Ensure environment variables are exported

Streaming Not Working

Issue: Streaming responses not appearing

Solutions:

1. Verify enableStreamingResponse: true in config
2. Check Lark client version supports interactive cards
3. Test with simple message first
4. Review network/firewall settings

Rate Limiting

Issue: "Rate limit exceeded" errors

Solutions:

1. Implement exponential backoff in custom skills
2. Cache frequently accessed data
3. Use batch operations where possible
4. Monitor API quota in Lark admin console

Best Practices

1. Security First: Never disable default security policies without understanding risks
2. Use Private Chats: Recommended for personal assistants to avoid permission abuse
3. Whitelist Groups: If using in groups, explicitly whitelist trusted chats only
4. Environment Variables: Always use env vars for credentials, never hardcode
5. Error Handling: Wrap API calls in try-catch blocks
6. Confirmation Dialogs: Use interactive cards for destructive operations
7. Rate Limiting: Respect API limits, implement backoff strategies
8. Logging: Enable detailed logs during development for debugging
9. Skill Restrictions: Limit available skills per group to reduce attack surface
10. Regular Updates: Keep OpenClaw and plugin updated for security patches

Additional Resources

• Official Documentation
• Lark Open Platform
• Feishu Open Platform
• OpenClaw Official Website
• GitHub Repository