Common Issues and Fixes

This page collects the issues users hit most often when setting up or operating Salambo.

The Agent Does Not Appear In /api/v1/models

Check these first:

• the agent is active
• you are calling the correct workspace
• the API key includes models:read

Remember that the models list only returns active agents for the authenticated workspace.

A Run Fails Immediately

The most common causes are:

• no snapshot assigned
• missing secret
• secret scoped to the wrong host
• required domain missing from restricted network allowlist
• billing or plan status blocking execution

Open the run detail page and inspect the event log before changing multiple settings at once.

A Service Call Works Outside Salambo But Fails Inside It

Usually this means one of two things:

• the hostname is not allowed by the current egress policy
• the secret is present but scoped to a different hostname

Check both the allowlist and the host scope on the secret.

The Snapshot Never Becomes Active

Review:

• the Docker image reference
• GitHub connection status if the image is in GHCR
• the entrypoint
• whether the image tag actually exists

If needed, create a fresh snapshot instead of repeatedly reusing a broken one.

I Cannot Invite Members Or Change Roles

This is usually a permission or policy issue. Check:

• whether your role allows invitation or role management
• whether invitations are currently blocked by workspace policy
• whether billing state is preventing invitation flows

I Cannot Create Another Agent Or Client Workspace

You may have reached a plan limit. Check the billing and usage surfaces for:

• agent limit
• sub-account limit
• billing-blocked status

The Terminal Is Not Available

The run page only shows a live terminal while the sandbox is active. If the sandbox has already stopped, the terminal panel may instead show resume actions or an informational message.