Call /v1/experience and /v1/recall from n8n workflows via HTTP nodes.

n8n Integration

n8n's HTTP Request node speaks JSON. Use it to push experiences into CortexDB and to recall context for downstream LLM nodes.

Setup

Store these as n8n Credentials so they're not hard-coded in the workflow:

NameValue
CORTEX_API_URLhttps://api-v1.cortexdb.ai
CORTEX_TOKENPASETO bearer token for the workflow actor
CORTEX_ACTORe.g. service:n8n-flow
CORTEX_SCOPEe.g. org:acme/source:n8n

Capture node — POST /v1/experience

Method:   POST
URL:      {{ $env.CORTEX_API_URL }}/v1/experience
Headers:
  Authorization:  Bearer {{ $credentials.CORTEX_TOKEN }}
  X-Cortex-Actor: {{ $env.CORTEX_ACTOR }}
  Content-Type:   application/json
Body (JSON):
  {
    "scope": "{{ $env.CORTEX_SCOPE }}",
    "modality": "conversation",
    "content": { "kind": "text", "text": "{{ $json.message }}" },
    "context": { "observed_at": "{{ $now.toISOString() }}" },
    "idempotency_key": "{{ $execution.id }}"
  }

The n8n execution ID is a perfect idempotency key — replays of the same execution write once.

Recall node — POST /v1/recall

URL:    {{ $env.CORTEX_API_URL }}/v1/recall
Body:   { "scope": "{{ $env.CORTEX_SCOPE }}", "view": "holistic",
          "query": "{{ $json.question }}",
          "include": ["beliefs", "facts", "episodes"],
          "budgets": { "max_tokens": 3000 } }

Wire the returned context_block into a downstream OpenAI / Anthropic node's system prompt.

See also