Skip to main content

Troubleshooting

Start here, then follow the link — most of these have a page of their own.

I cannot launch anything

The launcher saysFix
Connect a coding agent firstAdd a credential
Connect an environment firstYou are in a workspace you created yourself — those start empty. Create an environment
Deploy a controller for the environmentAttach a controller to your sandbox environment

If the button is enabled but a specific agent definition is greyed out, that is different: it means you have no credential for the agent that definition names. → Why a definition can be unlaunchable for you

My launch started and never finished

A launch that will not finish — the steps one by one. Where it stopped is what tells you why.

I added a valid credential and the checklist still says to connect one

Two things satisfy that step, and both have to hold:

  1. The credential has to reach the workspace you are in. A shared credential is either available to all workspaces or assigned to specific ones — if it is assigned and this workspace is not on the list, it does not resolve here. Fix it in the connection's settings.
  2. It has to be for your organization's default coding agent. Once a default is set, a credential for the other coding agent will not satisfy the step — deliberately, so you are not told everything is ready and then fail inside the container.

Also check that the credential has not expired and is still active. → Credentials and secrets

I rotated a credential and it is still using the old one

Expected. A credential is frozen into the container at launch. Launch a fresh runtime; nothing else will pick it up.

My script waits forever for a runtime that is clearly up

You are almost certainly reading a field that answers a different question — or one the endpoint you called does not return at all. → Knowing when a runtime is ready

My environment went red

The controller disconnected. Usually the host slept or rebooted, the container was removed, or egress to PlatformSmith closed. → When a controller goes offline

Clones fail but everything else works

The Git connection has lost access to that repository — the App installation was revoked, or it is a selected-repositories installation that does not include it. Use Reauthorise on the connection, or add the repository on GitHub's side. → Connect GitHub

I cannot find my Forgemate history

It is in the Forgemate panel, as a thread. It is not under Conversations in the sidebar — different noun entirely. → The four execution nouns

A launch is refused with "no company default is set"

Your organization has two agent definitions and neither is flagged as the default — the expected state until a credential is connected. Connect one through the setup wizard, or set the default from the definition's own page. → No default coding agent yet

I switched an environment to Kubernetes and it was refused

Either a controller has already connected to it — in which case the type is settled and you need a new environment — or its derived namespace name collides with another Kubernetes environment in the organization, which renaming clears. The message distinguishes them. → Choosing the type

A teammate can see everything but cannot do anything

They are probably a member, which is read-across-the-organization and cannot launch. Give them a workspace role. → The five roles

A screen I read about is not in my sidebar

That capability is not enabled on this deployment. → Not yet enabled

Still stuck

Collect the workspace, the project, the environment, the phase or screen, and roughly when — those are the first things anyone helping will ask for. → Support