Configurations
FluidGrids is a managed service, so there is no server configuration to maintain. What you do configure is how your clients reach FluidGrids, and how your workflows behave when they run.
Connecting a client
| Client | Settings | Notes |
|---|---|---|
| CLI | FLUIDGRIDS_WORKSPACE_ENDPOINT, FLUIDGRIDS_GLOBAL_ENDPOINT, FLUIDGRIDS_OIDC_ISSUER | Pick the workspace with fluidgrids workspace set <id>. See CLI Reference. |
| Node SDK | FLUIDGRIDS_WORKSPACE_ENDPOINT, FLUIDGRIDS_GLOBAL_ENDPOINT | Or pass workspaceEndpoint / globalEndpoint to the constructor. See Node SDK. |
| MCP server | FLUIDGRIDS_WORKSPACE_ENDPOINT, FLUIDGRIDS_GLOBAL_ENDPOINT | Plus tokens or FLUIDGRIDS_API_KEY for unattended use. See MCP Server. |
All clients default to production. See the FAQ for the endpoint values.
Authentication
| Setting | Description |
|---|---|
| App client credentials | For automation: register an app under Developer Portal and sign in as it (fluidgrids auth login --client-credentials, or loginWithClientCredentials in the SDK) |
| API key | Workspace-scoped key created under Developer Portal › API Keys: the SDK's apiKey option, or FLUIDGRIDS_API_KEY for the MCP server |
API keys are scoped to a single workspace and are audit-logged on every use. Rotate them from the same screen that issues them; revocation takes effect immediately.
Keep keys out of source control. Read them from your own secret store or your CI provider's encrypted variables.
Workflow execution settings
These are configured in the product, not by environment variable. Workspace-wide defaults live on the workflow module's Settings screen (/workflows/settings); per-node behaviour is set in the node's Settings tab in the builder:
| Setting | Where | Description |
|---|---|---|
| Default Timeout | Settings › Execution | How long a run may take before it is stopped |
| Max Retries / Retry Delay | Settings › Execution | Automatic retries on failure and the wait between them |
| Concurrent Execution Limit | Settings › Execution | How many runs may execute at once in the workspace |
| Per-node retry and error handling | Node › Settings | Whether a failed node is retried, and whether the run continues on error |
Plan limits cap the maximum values; the workspace's own settings apply below those caps. See Settings and Plans and Quotas.