BestPaid guides · Microsoft 365 · Microsoft Teams

Teams Incoming Webhook connectors stopped in May 2026: move every Office 365 Connector to Workflows, a bot or Microsoft Graph

Microsoft disabled Office 365 Connectors in Teams between May 18 and May 22, 2026. Scripts, monitoring tools and CI pipelines that still post to a connector webhook URL no longer reach their channel. Here is how to find each sender, move it to a Workflows webhook, or pick a bot or Microsoft Graph instead.

Sources: Microsoft's retirement post (updated April 14, 2026) and Message Center post MC1181996; Microsoft Learn's Create Incoming Webhooks, Create and send messages, card reference, card formatting, MessageCard reference and Teams connector pages; the Adaptive Card schema; Microsoft Support's webhook templates; Power Automate's trigger authentication, limits, trigger URL changes, SAS key and IP address pages; Microsoft Graph's channel message, chat message and activity feed pages; the Teams proactive messages and bot rate limits pages; the Invoke-RestMethod, ConvertTo-Json and curl manuals; and a Microsoft Q&A moderator's answer (Microsoft External Staff, July 27, 2026). Checked on November 21, 2026. This page is our own summary; follow the vendor's documentation for every detail.

What happened and when

DateWhat Microsoft announced
July 3, 2024Retirement announced; the deadline later moved to the end of 2025, March 31, 2026 and April 30, 2026.
November 30, 2025Workflows URLs containing logic.azure.com stop working in environments on the Logic Apps architecture; the flow's trigger card shows the new one.
February 5, 2026Workflows webhooks accept MessageCards and can post to shared channels.
April 14, 2026Final schedule; private channels supported through the Power Automate portal.
May 18–22, 2026Connectors disabled in stages; after that they no longer function.

What stopped working

1. Find every sender

2. Create a Workflows webhook for each channel or chat

  1. Allow the Workflows app in the Teams admin center.
  2. In Teams, select More options (...) next to the channel, then Workflows, and pick a template such as Send webhook alerts to a channel. There are also chat versions, and variants for alerts from specific people or from people in an org; Microsoft says each template has a different authentication type and none needs a premium license.
  3. Choose the channel or chat, save and copy the URL; each URL posts to one place. Test private channels first: the Teams docs say they work, but the connector reference says they don't.
  4. Add a co-owner. A workflow is tied to its owner (generally its creator), not to the channel, and uses that person's plan. Without co-owners it can be orphaned when the owner leaves, and Microsoft says a flow whose original owner leaves the organization reverts to the Low performance profile.
  5. Treat the URL as a secret: with Anyone, whoever has it can trigger the flow (Microsoft documents regenerating the sig= key of HTTP trigger URLs). Allow for URLs over 255 characters, and in the commercial cloud open *.api.powerplatform.com and *.logic.azure.com on outbound firewalls.

3. Decide how senders authenticate

The trigger accepts Anyone, Any user in my tenant or Specific users in my tenant. For token claims, the Teams connector reference points to Power Automate's docs for its HTTP request trigger, which call Anyone a legacy setting. The two tenant options need a bearer token whose aud is exactly https://service.flow.microsoft.com/ (public cloud, trailing slash included), plus iss and tid, and oid for specific users, which can include a service principal's object ID. With Anyone, send no Authorization header, or requests fail.

4. Convert the payload: MessageCard or Adaptive Card

The trigger takes Adaptive Card attachments or a single MessageCard, so a MessageCard without buttons may need only the new URL. To convert:

Before, a MessageCard (its button won't render on Workflows):

{
  "@type": "MessageCard",
  "@context": "http://schema.org/extensions",
  "themeColor": "D70000",
  "summary": "Disk space low on SRV01",
  "sections": [{
    "activityTitle": "Disk space low on SRV01",
    "facts": [ { "name": "Drive", "value": "D:" }, { "name": "Free", "value": "4%" } ],
    "markdown": true
  }],
  "potentialAction": [{
    "@type": "OpenUri",
    "name": "Open dashboard",
    "targets": [ { "os": "default", "uri": "https://monitor.example.com/srv01" } ]
  }]
}

After, as an Adaptive Card:

{
  "type": "message",
  "attachments": [{
    "contentType": "application/vnd.microsoft.card.adaptive",
    "contentUrl": null,
    "content": {
      "$schema": "http://adaptivecards.io/schemas/adaptive-card.json",
      "type": "AdaptiveCard",
      "version": "1.2",
      "msteams": { "width": "Full" },
      "body": [
        { "type": "TextBlock", "text": "Disk space low on SRV01", "weight": "Bolder", "size": "Medium", "wrap": true },
        { "type": "FactSet", "facts": [ { "title": "Drive", "value": "D:" }, { "title": "Free", "value": "4%" } ] }
      ],
      "actions": [
        { "type": "Action.OpenUrl", "title": "Open dashboard", "url": "https://monitor.example.com/srv01" }
      ]
    }
  }]
}

5. Update the sender

Microsoft recommends exponential backoff on HTTP 429. If a request succeeds but no card appears, check the flow's run history in Power Automate.

using System.Text;

// The Workflows URL is a secret: keep it in configuration, not in source code.
var webhookUrl = Environment.GetEnvironmentVariable("TEAMS_WORKFLOW_URL")
    ?? throw new InvalidOperationException("TEAMS_WORKFLOW_URL is not set");
var cardJson = await File.ReadAllTextAsync("card.json");   // the Adaptive Card message above

using var http = new HttpClient();
for (var attempt = 1; ; attempt++)
{
    using var content = new StringContent(cardJson, Encoding.UTF8, "application/json");
    using var response = await http.PostAsync(webhookUrl, content);
    if (response.IsSuccessStatusCode) break;                // request succeeded

    if ((int)response.StatusCode == 429 && attempt < 5)     // throttled: back off and retry
    {
        await Task.Delay(response.Headers.RetryAfter?.Delta ?? TimeSpan.FromSeconds(Math.Pow(2, attempt)));
        continue;
    }
    response.EnsureSuccessStatusCode();                     // throws with the status code
}

In PowerShell 7, Invoke-RestMethod retries only if you set -MaximumRetryCount. It then retries on a 304 or any 400–599 status; on a 429 with a Retry-After header, it waits that long instead of -RetryIntervalSec. Windows PowerShell 5.1 has neither parameter. Set -Depth: ConvertTo-Json stops at two levels by default.

$card = @{
    type        = 'message'
    attachments = @(@{
        contentType = 'application/vnd.microsoft.card.adaptive'
        contentUrl  = $null
        content     = @{
            '$schema' = 'http://adaptivecards.io/schemas/adaptive-card.json'
            type      = 'AdaptiveCard'
            version   = '1.2'
            body      = @(
                @{ type = 'TextBlock'; text = "Backup failed on $([Environment]::MachineName)"; weight = 'Bolder'; wrap = $true }
                @{ type = 'FactSet'; facts = @(@{ title = 'Time'; value = (Get-Date -Format u) }) }
            )
        }
    })
}

Invoke-RestMethod -Method Post -Uri $env:TEAMS_WORKFLOW_URL `
    -ContentType 'application/json; charset=utf-8' `
    -Body ($card | ConvertTo-Json -Depth 10) `
    -MaximumRetryCount 4 -RetryIntervalSec 5

In CI, add --fail: by default curl doesn't treat HTTP errors as failures.

curl --fail --silent --show-error --retry 4 \
  -H "Content-Type: application/json; charset=utf-8" \
  --data-binary @card.json "$TEAMS_WORKFLOW_URL"

The limits Microsoft documents

LimitValue
Message sizeAbout 28 KB; larger posts fail with Request Entity too large.
Burst rateOver four requests a second gets the client throttled.
Flow bot posts25 non-GET calls per connection per 300 seconds.
Trigger calls4,500 per 5 minutes on the Low profile (Microsoft 365 licenses), 45,000 on higher ones.
Daily actionsAbout 10,000 Power Platform requests a day on Low; each action counts.
Idle or failing flowsNo trigger for 90 days: may be turned off after 30 days' notice (premium-licensed owners and Process-licensed flows are exempt), a risk for rare alerts. Failing for 14 days straight: turned off.
IdentityThe Flow bot or a user; no custom name or icon. No Flow bot in GCC, GCC High or DoD.

When a bot or Microsoft Graph is the better target

Alerts went quiet in May and nobody knows which scripts sent them? We move apps and scripts like these to Microsoft Graph and Microsoft Entra ID at a fixed price: a script or scheduled job from $690, an app or service from $1,490. One app gets a free estimate within 2 business days. We never ask for passwords, secrets or admin access: your admin registers the app and grants consent.
See the Microsoft 365 migration offer