n8n

Prerequisites

• An n8n instance, self-hosted or on n8n Cloud
• An email address for the knowmind account
• An access key with write if the workflow saves; read is enough for looking things up

Steps

1. 1

   Create your knowmind account

   Enter your email address at knowmind.de. You receive a sign-in link by mail and use it to sign in. There is no password.

   Then create your workspace: name, short name and plan. The Private plan costs 0 € and is enough for this guide; the pricing page lists the others.

   Outcome: The welcome screen is open, offering “start by writing” and “connect my AI tool”. Three starter cards are already in your workspace.

2. 2

   Create the credential in n8n

   Create a key in the Cockpit under keys. In n8n, use it for a header authentication credential:

   textcopy

   Credentials → New → HTTP Header Auth
   Name:           knowmind
   Header Name:    Authorization
   Header Value:   Bearer kmt_YOUR_KEY

   The key then lives in n8n rather than in single nodes; to rotate it later, the credential is all you touch.

   Outcome: The credential “knowmind” is ready in n8n.

3. 3

   Set up the HTTP Request node

   One node covers every tool: you call the MCP address over JSON-RPC and pick the tool in the name field.

   textcopy

   Method:          POST
   URL:             https://knowmind.de/api/mcp/v1
   Authentication:  Generic Credential Type → HTTP Header Auth → knowmind
   Send Body:       yes, Body Content Type: JSON

   jsoncopy

   {
     "jsonrpc": "2.0",
     "id": 1,
     "method": "tools/call",
     "params": {
       "name": "knowmind_recall_memories",
       "arguments": { "query": "{{ $json.subject }}", "k": 5 }
     }
   }

   Outcome: The node answers with plain JSON. The hits sit in result.content[0].text; a following node reads them out.

4. 4

   Activate the workflow and call it

   Save the workflow, activate it and trigger it — with a webhook trigger, through the production URL rather than the test URL. To save instead of search, change only the tool name and the fields:

   jsoncopy

   {
     "jsonrpc": "2.0",
     "id": 2,
     "method": "tools/call",
     "params": {
       "name": "knowmind_store_memory",
       "arguments": {
         "memory_type": "episodic",
         "title": "Mail from {{ $json.from }}",
         "content": "{{ $json.subject }}\n\n{{ $json.text }}",
         "source": "n8n"
       }
     }
   }

   Outcome: The run ends with HTTP 200 and a result field, and the saved content is found in the next recall straight away.

5. 5

   Plan for the rate limit

   Workflows that query knowmind in a loop, such as an import of old mail, run into the rate limit. It applies to the tools that query the memory; saving and linking do not count towards it.

   Over this route the limit does not arrive as a status code: the node receives HTTP 200 with an error field carrying the code -32004. As far as n8n is concerned the call succeeded, so Retry On Fail does not catch it. Check in the next node whether error is set, and slow the workflow down with a wait node. The limits of your plan are on the German limits page.

Verify the result

Troubleshooting