この本文は AI(Claude)が読むための原文(英語または中国語)です。日本語訳は順次追加中。

Microsoft Teams

Overview

This skill helps AI agents build integrations with Microsoft Teams — from simple webhook notifications to full-featured bots and workflow automation. It covers the Microsoft Graph API for team/channel/message management, Bot Framework for conversational bots, incoming webhooks for quick notifications, Adaptive Cards for rich UI, and meeting automation.

Instructions

Choose Integration Type

Need	Solution	Complexity
Send notifications to a channel	Incoming Webhook	Low
Read/write messages, manage teams	Graph API	Medium
Interactive bot (commands, dialogs)	Bot Framework	High
No-code automation	Power Automate	Low

Incoming Webhooks (Simplest)

const WEBHOOK_URL = process.env.TEAMS_WEBHOOK_URL;

// Simple text
await fetch(WEBHOOK_URL, {
  method: 'POST', headers: { 'Content-Type': 'application/json' },
  body: JSON.stringify({ text: '**Production Alert**: CPU at 95% on api-server-01' }),
});

// Adaptive Card (rich formatting)
await fetch(WEBHOOK_URL, {
  method: 'POST', headers: { 'Content-Type': 'application/json' },
  body: JSON.stringify({
    type: 'message',
    attachments: [{
      contentType: 'application/vnd.microsoft.card.adaptive',
      content: {
        type: 'AdaptiveCard', version: '1.5',
        body: [
          { type: 'TextBlock', text: 'Deployment Complete', weight: 'Bolder', size: 'Large' },
          { type: 'FactSet', facts: [
            { title: 'Service', value: 'api-gateway' },
            { title: 'Version', value: 'v2.4.1' },
            { title: 'Environment', value: 'Production' },
          ]},
        ],
        actions: [{ type: 'Action.OpenUrl', title: 'View Dashboard', url: 'https://grafana.example.com' }],
      },
    }],
  }),
});

Graph API

import { ClientSecretCredential } from '@azure/identity';
import { Client } from '@microsoft/microsoft-graph-client';
import { TokenCredentialAuthenticationProvider } from '@microsoft/microsoft-graph-client/authProviders/azureTokenCredentials';

// Permissions: Team.ReadBasic.All, ChannelMessage.Send, Chat.ReadWrite.All, OnlineMeetings.ReadWrite.All
const credential = new ClientSecretCredential(
  process.env.AZURE_TENANT_ID, process.env.AZURE_CLIENT_ID, process.env.AZURE_CLIENT_SECRET
);
const authProvider = new TokenCredentialAuthenticationProvider(credential, {
  scopes: ['https://graph.microsoft.com/.default'],
});
const graphClient = Client.initWithMiddleware({ authProvider });

// Create channel
await graphClient.api(`/teams/${teamId}/channels`).post({
  displayName: 'Project Alpha', membershipType: 'standard',
});

// Send message with Adaptive Card
await graphClient.api(`/teams/${teamId}/channels/${channelId}/messages`).post({
  body: { contentType: 'html', content: '' },
  attachments: [{
    id: '1', contentType: 'application/vnd.microsoft.card.adaptive',
    content: JSON.stringify({
      type: 'AdaptiveCard', version: '1.5',
      body: [
        { type: 'TextBlock', text: 'PR #142: Add payment retry logic', weight: 'Bolder' },
        { type: 'FactSet', facts: [{ title: 'Author', value: 'Sarah Chen' }, { title: 'Tests', value: 'All passing' }] },
      ],
      actions: [{ type: 'Action.OpenUrl', title: 'Review PR', url: 'https://github.com/org/repo/pull/142' }],
    }),
  }],
});

// Create online meeting
const meeting = await graphClient.api(`/users/${organizerId}/onlineMeetings`).post({
  subject: 'Sprint Planning',
  startDateTime: '2026-03-01T14:00:00Z', endDateTime: '2026-03-01T15:00:00Z',
  participants: { attendees: [{ upn: 'sarah@example.com', role: 'attendee' }] },
});
console.log('Join URL:', meeting.joinUrl);

Bot Framework

import { ActivityHandler, TurnContext, CardFactory } from 'botbuilder';
import { BotFrameworkAdapter } from 'botbuilder';
import express from 'express';

class TeamBot extends ActivityHandler {
  constructor() {
    super();
    this.onMessage(async (context: TurnContext, next) => {
      const text = context.activity.text?.trim().toLowerCase();
      if (text === 'status') {
        const card = CardFactory.adaptiveCard({
          type: 'AdaptiveCard', version: '1.5',
          body: [
            { type: 'TextBlock', text: 'System Status', weight: 'Bolder', size: 'Large' },
            { type: 'FactSet', facts: [
              { title: 'API', value: 'Healthy (42ms)' },
              { title: 'Database', value: 'Healthy (8ms)' },
              { title: 'Queue', value: 'High (1,247 pending)' },
            ]},
          ],
        });
        await context.sendActivity({ attachments: [card] });
      } else {
        await context.sendActivity('Commands: `status` — system health, `deploy <service>` — deploy');
      }
      await next();
    });
    this.onMembersAdded(async (context, next) => {
      for (const member of context.activity.membersAdded) {
        if (member.id !== context.activity.recipient.id)
          await context.sendActivity(`Welcome, ${member.name}! Type \`help\` to see what I can do.`);
      }
      await next();
    });
  }
}

const adapter = new BotFrameworkAdapter({
  appId: process.env.MICROSOFT_APP_ID, appPassword: process.env.MICROSOFT_APP_PASSWORD,
});
const bot = new TeamBot();
const app = express();
app.post('/api/messages', (req, res) => adapter.process(req, res, (ctx) => bot.run(ctx)));
app.listen(3978);

Examples

Example 1: Set up deployment notifications via webhook

User prompt: "Send a rich notification to our Teams DevOps channel whenever a deployment completes, showing the service name, version, environment, and a link to Grafana."

The agent will:

Get the Incoming Webhook URL from the Teams channel connector settings
Construct an Adaptive Card with a TextBlock header, a FactSet showing service/version/environment/time, and an Action.OpenUrl button linking to the Grafana dashboard
POST the card payload to the webhook URL from the CI/CD pipeline's post-deployment step
Add error handling to log failures without blocking the deployment

Example 2: Build a slash-command bot for system health checks

User prompt: "Create a Teams bot that responds to 'status' with a system health card showing API, database, and queue metrics pulled from our monitoring endpoints."