What problem does it solve?
When an n8n-mcp account runs in multi-instance mode, every workflow, datatable, credential, and execution call silently routes to whichever n8n instance the session currently targets. A wrong target produces wrong data, misleading NOT_FOUND errors, or secrets written to the wrong instance, usually with no error at all.
Core Features & Use Cases
- Instance discovery and switching: List available instances with
n8n_instances({mode:"list"}) and bind the session by name with mode:"switch", always in its own turn.
- Pre-write verification: Re-confirm the
current instance immediately before credential create/update/delete operations, and handle the INSTANCE_AMBIGUOUS fail-close by switching on the session and retrying.
- Misroute recovery playbook: Diagnose unexpected NOT_FOUND or empty reads as wrong-instance routing, recover without recreating objects, and detect silent fallback to the default instance.
- Use Case: You have prod and staging n8n instances and need to create a credential on staging. The skill guides you to list instances, switch to staging in its own call, verify
current, then write the credential safely.
Quick Start
Ask the assistant to list your n8n instances, switch this session to the staging instance, and confirm the target before creating a credential there.
