Unable to Configure DataDog MCP with an Automation

[Greg_Farrow]

Where does the bug appear (feature/product)?

Somewhere else…

Describe the Bug

Until recently I had been using the Automation “Investigate top Datadog errors”. However, recently the DataDog MCP broke and the automation stopped working.

I can no longer setup the DataDog MCP on the automation agents.

The Datadog MCP is unusable in automations, which is a big issue for my use case.

Steps to Reproduce

• Go to https://cursor.com/automations and choose the “Incidents & Triage” tab, then select “Investigate top Datadog errors”
• The automation opens and at the bottom under “Tools” I can see “Datadog” with the options “Set up MCP”, edit and delete.
• When I click “Set up MCP” a dialog opens asking to install the Datadog Plugin (Note: the plugin is already installed, both locally and in the cloud, so this shouldn’t show)
• I click “Add Plugin”
• The dialog closes, but the Datadog MCP is not authenticated.

I don’t see any console logs or network activity when I click “add plugin”. There’s no indiation that anythig has happened.

Expected Behavior

The Datadog MCP authentication completes successfully and the Datadog MCP is available to this automation (and any others I setup).

Screenshots / Screen Recordings

Operating System

MacOS

Version Information

Version: 3.8.11 (Universal)
VS Code Extension API: 1.105.1
Commit: e56ad3440df06d22ca7501e65fd518e905486ef0
Date: 2026-06-18T01:40:18.333Z
Layout: glass
Build Type: Stable
Release Track: Default
Electron: 40.10.3
Chromium: 144.0.7559.236
Node.js: 24.15.0
V8: 14.4.258.32-electron.0
xterm.js: 6.1.0-beta.256
OS: Darwin arm64 25.5.0

Does this stop you from using Cursor

No - Cursor works, but with this issue

[Colin]

Hey @Greg_Farrow

Thanks for the report. We’ve had a couple of reports come in about this, and I can also reproduce similar behavior on my personal account!

It looks like something changed in the Datadog plugin that the Plugin install for Cloud Agents/Automations isn’t expecting. Something will have to change on our end or in the plugin! Will keep you posted.

[Greg_Farrow]

Hi @Colin - has the team made any progress on this bug yet? I am still able to replicate it today. Any ideas on an estimated time to fix as this is a major issue for me, I had been relying heavily on this functionality.

Thanks

Greg

[Jason_Luo1 (Jason Luo)]

This is the kind of automation breakage where a visible auth state would help a lot.

If the plugin is installed but the automation agent cannot actually use the MCP connection, the UI should probably distinguish “plugin present” from “tool authenticated and available in this automation run.” Otherwise it looks configured until the job silently loses its core tool.

[Colin]

I’ll ping on this @Greg_Farrow!

[Colin]

@Greg_Farrow We just merged a fix, so next deploy this should be fixed. Check back a little later!

[Greg_Farrow]

TL;DR there is more visibility but the bug remains.

@Colin I’ve tried to connect in the latest version:

Version: 3.9.16 (Universal)
VS Code Extension API: 1.105.1
Commit: 042b3c1a4c53f2c3808067f519fbfc67b72cad80
Date: 2026-06-27T06:41:01.941Z
Layout: editor
Build Type: Stable
Release Track: Default
Electron: 40.10.3
Chromium: 144.0.7559.236
Node.js: 24.15.0
V8: 14.4.258.32-electron.0
xterm.js: 6.1.0-beta.256
OS: Darwin arm64 25.5.0

First I have removed the DataDog Plugin, to attempt to set it up from scratch.

When I attempt to reconnect I see:

and in the output logs I see:

2026-06-30 08:31:38.202 [info] [MCPService] [MCPService] createClient: identifier=“plugin-datadog-datadog”, serverName=“datadog”

2026-06-30 08:31:38.217 [error] [MCPService] [MCPService] createClient completed for server: plugin-datadog-datadog, connected=false, statusType=error, error=net::ERR_NAME_NOT_RESOLVED

The plugin is installed but broken.

In the Agents Window the MCP is broken for Cloud and is failing to authenticate for Local:

In older versions, the plugin would prompt for the DataDog URL to use (i.e. the region). But, in this version it does not prompt, so I suspect it is not setting the URL correctly and that is causing the error?

[Greg_Farrow]

@Colin the DataDog MCP is still broken in cloud. It works locally n, but for Cloud and Automations it is still in Error state.

I have tried reinstalling the plugin and rerunning /ddsetup. But, these actions haven’t helped.

Is there an investigation still underway as this ia ba big issue for me still.

[Greg_Farrow]

@Colin following the latest release the error configuring the DataDog MCP tool is still present. I was able to retrieve the following output logs - I hope these help you diagnose and fix the problem:

```
2026-07-06 01:41:57.130 [info] [Shared MCP process] connecting streamableHttp for “datadog” (plugin-datadog-datadog)

2026-07-06 01:41:57.130 [info] [Shared MCP process] [V2 FSM] connection:connect_start: conn=idle,auth=unknown → conn=connecting,auth=unknown

2026-07-06 01:41:57.527 [warning] [Shared MCP process] MCP HTTP exchange failed

2026-07-06 01:41:57.527 [warning] [Shared MCP process] Transient error connecting to streamableHttp server: fetch failed

2026-07-06 01:42:05.574 [warning] [Shared MCP process] MCP HTTP exchange completed

2026-07-06 01:42:05.685 [info] [Shared MCP process] MCP OAuth refresh lock acquired

2026-07-06 01:42:05.686 [info] [Shared MCP process] MCP OAuth refresh prepared

2026-07-06 01:42:06.353 [info] [Shared MCP process] MCP OAuth tokens persisted

2026-07-06 01:42:06.353 [info] [Shared MCP process] MCP OAuth refresh result

2026-07-06 01:42:06.354 [info] [Shared MCP process] MCP OAuth refresh lock released

2026-07-06 01:42:06.526 [info] [Shared MCP process] Successfully connected to streamableHttp server

2026-07-06 01:42:06.526 [info] [Shared MCP process] [V2 FSM] connection:connect_success: conn=connecting,auth=unknown → conn=connected,auth=unknown

2026-07-06 01:42:06.596 [warning] [Shared MCP process] MCP HTTP exchange completed

2026-07-06 01:42:48.615 [info] [MCPService] [MCPService] createClient: identifier=“plugin-datadog-datadog”, serverName=“datadog”

2026-07-06 01:42:48.616 [info] [MCPService] [MCPService] createClient completed for server: plugin-datadog-datadog, connected=true, statusType=connected

2026-07-06 01:42:49.319 [info] [MCPService] [MCPService] createClient: identifier=“plugin-datadog-datadog”, serverName=“datadog”

2026-07-06 01:42:49.326 [info] [MCPService] [MCPService] createClient completed for server: plugin-datadog-datadog, connected=true, statusType=connected

2026-07-06 01:42:50.273 [info] [MCPService] [MCPService] createClient: identifier=“plugin-datadog-datadog”, serverName=“datadog”

2026-07-06 01:42:52.477 [info] [MCPService] [MCPService] createClient: identifier=“plugin-datadog-datadog”, serverName=“datadog”

2026-07-06 01:42:52.481 [info] [MCPService] [MCPService] createClient completed for server: plugin-datadog-datadog, connected=true, statusType=connected

2026-07-06 01:43:50.640 [error] [MCPService] [MCPService] Error creating client: Timeout waiting for EverythingProvider with command ‘mcp.createClient’

2026-07-06 01:43:50.646 [info] [MCPService] [MCPService] createClient: identifier=“plugin-datadog-datadog”, serverName=“datadog”

2026-07-06 01:44:51.633 [error] [MCPService] [MCPService] Error creating client: Timeout waiting for EverythingProvider with command ‘mcp.createClient’
```

[Greg_Farrow]

Hey @Colin its been over 2 weeks and this is still broken for me and is causing me serious issues with my production workload. It would be really helpful to know if this is still being worked on as this thread has gone cold. Thanks

[Colin]

Sorry for the delay, @Greg_Farrow. I’ll reflag this for the team.

[Colin]

Hi again @Greg_Farrow!

@Greg_Farrow

In older versions, the plugin would prompt for the DataDog URL to use (i.e. the region). But, in this version it does not prompt, so I suspect it is not setting the URL correctly and that is causing the error?

After installing the plugin, do you see a Configure option? This should let you configure the URL correctly.

IDE:

Web:

This was unfortunately not too obvious to me, and probably something we should improve the visibility of! Let me know if you’re still facing issues after the URL is configured.

[Javier_Valencia]

Hey @Colin ! I am also having an issue with the Automations + Datadog.
Even when I set the Datadog plugin URL it shows as this on the MCP and wont let me connect:
https://${dd_mcp_domain:-not-setup}/v1/mcp?referrer_ide=cursor-plugin&plugin_version=0.7.14&toolsets=${dd_mcp_toolsets:-core,visualizations}

[naveen.narayanan (Naveen)]

Same issue for me as well, getting this error each time with all the same setup as others when running it on cursor automation + cloud.

Datadog MCP is set up, but it is not working right now
Please run /ddconfig to repair the Datadog MCP server

[Greg_Farrow]

Hi @Colin thanks so much for the reply and the info - following those instructions I was able to successfully re-configure

You are right, the user experience around this functionality is really poor. I particularly struggle with differences between browser-based behaviour being similar but subtly different to the Agent Window behaviour - for example the configure button is not visible in the browser…

Its also very difficult to understand from the UI where configuration is applied only to local, only to cloud, or both. I hope you are able to feed this back t the team so they can improve this.

TL;DR - Using the Agent Window, I was able to navigate tot he Plugin page fo Datadog and click on the greyed out configure button (it looks like it is disabled) and click on it to select the region.