Cannot configure Cloud Agent environment

Support Bug Reports
1

Where does the bug appear (feature/product)?

Background Agent (GitHub, Slack, Web, Linear)

Describe the Bug

I’m unable to configure a cloud agent environment interactively. I get the following error whenever I attempt to connect to the remote machine to create a machine snapshot:

Failed to connect to the remote extension host server (Error: [internal] Pod exists but exec-daemon is unreachable after 8 retries)

I’ve tried this on multiple projects, including one that I never previously attempted to setup a cloud agent environment for.

I’ve tried with an environment.json & without one. I’ve tried the “Rebuild environment” option, I’ve tried switching to a custom dockerfile (which seems to work?) and back again. I’ve restarted Cursor, run network diagnostics (everything is green), and I’m not on any kind of VPN.

Steps to Reproduce

Cursor Settings > Cloud Agents > Base Environment > Re-configure Manual Snapshot (or Configure Interactively)

Screenshots / Screen Recordings

Screenshot 2025-12-17 at 9.50.21 AM.png
Screenshot 2025-12-17 at 9.50.21 AM.png670×286 19.1 KB

Operating System

MacOS

Current Cursor Version (Menu → About Cursor → Copy)

Version: 2.2.20
VSCode Version: 1.105.1
Commit: b3573281c4775bfc6bba466bf6563d3d498d1070
Date: 2025-12-12T06:29:26.017Z
Electron: 37.7.0
Chromium: 138.0.7204.251
Node.js: 22.20.0
V8: 13.8.258.32-electron.0
OS: Darwin arm64 24.6.0

Does this stop you from using Cursor

Yes - Cursor is unusable

Cloud Agents Setup: "No remote extension installed to resolve background-composer" error during Machine Snapshot creation
2

I’ve also contacted support a couple days ago but have gotten no response or acknowledgement so far.

3

Hey, thanks for the report. The “Pod exists but exec-daemon is unreachable after 8 retries” error is a confirmed infrastructure issue with Cloud Agent that’s being actively investigated.

This is NOT your config error - the pod creates successfully, but the exec-daemon inside the container is unreachable due to Cursor infrastructure problems.

Try these for now:

  • Wait 10-15 minutes and retry setup - often these are temporary glitches that resolve themselves
  • Disable VPN/proxy completely
  • Check internet stability: Settings > Network > Run Diagnostics
  • Try at a different time of day (possible regional AWS issues)
  • If you have environment.json, delete it and start with basic config
  • Try Dockerfile instead of Manual Snapshot (works better for some)

The team knows about the issue and is working on Cloud Agent infrastructure stability.

4

@deanrie Thanks for the quick reply!

Wait 10-15 minutes and retry setup - often these are temporary glitches that resolve themselves

I’ve tried occasionally for ~3 days. Hasn’t worked once yet.

Disable VPN/proxy completely

As I mentioned above, I have no VPN or proxy setup at all on this machine.

Check internet stability: Settings > Network > Run Diagnostics

As mentioned above, I did, and it’s all passing / green.

Try at a different time of day (possible regional AWS issues)

I’ve tried mornings, afternoons, and evenings

If you have environment.json, delete it and start with basic config

As mentioned above, I tried with and without an environment.json.

Try Dockerfile instead of Manual Snapshot (works better for some)

As mentioned above, I tried Dockerfile and it seems to work, but unfortunately Docker isn’t an option for our use case (we use docker-compose for our app’s supporting services in development, which means we need a true Docker in Docker setup, which is quite complex and edge-case-ridden).

Is there any way to publicly track progress here? This is a showstopper for us to adopt background agents, so we’ll need to start exploring other options (i.e. Codex, Devin, etc) if we can’t get any clarity on when the service is expected to be restored.

1 Like
5

Thanks for the detailed summary of what you’ve already tried. I understand your situation, this is a critical issue.

Current status: the infrastructure team is actively working on a fix for the exec-daemon.

About a public tracker: we don’t have a public roadmap for infrastructure bugs, but I can confirm this is a high-priority item for the team.

I’ll keep you posted if I get any updates from the infrastructure team.

6

Dealing with the exact same situation to a tee.

  • Ultra user, with an uncomfortably high limit set on usage pricing.
  • Brand new repo, published to github
  • Cloud agent settings all show as valid
  • Have tried across 48hrs every few hours or so
  • No VPN, no proxy
  • Network diagnostics show fine on HTTP/2 and 1.1
  • Tried both 2 and 1.1
  • Local + local worktree agents work fine
  • Have tried a from-scratch “Cloud Agent Environment Setup” (cannot ever get to taking a snapshot)
  • Have tried a custom environment.json and Dockerfile
  • Using base Ubuntu image with no startup scripts

In no case can I get the environment spun up. It always results in

Failed to connect to the remote extension host server (Error: [internal] Pod exists but exec-daemon is unreachable after 8 retries)

I suspect there’s a wider spread issue.

8

@deanrie any updates on when this is expected to be resolved?

9

Same problem here.

10

@pauluslimma @atgrant thanks for the additional reports. Confirming this is a known issue and the infra team is actively working on it.

This is a high-priority exec-daemon connectivity problem that impacts Cloud Agent users when setting up the environment. I don’t have an ETA for the fix yet, but the team is working on a resolution.

11

For anyone else stuck here unable to use background agents, we ended up going with Claude Code. Support for our workflow isn’t quite as good on paper (no Linear integration), but it actually works, which is the more important aspect.

@deanrie we’d love to use Cursor someday if the team ever figures out the infrastructure stuff, but taking a month to solve a showstopping infrastructure bug doesn’t instill the confidence needed to build our team process around this going forward.