Input & Output

Output Ports

insert / update Output

query Output

Result Item Schema

Example: query Output

{
  "query_text": "What is the company's policy on remote work?",
  "result_count": 3,
  "sources": ["policy-remote-work-v4", "policy-flexible-hours-v2", "faq-wfh-001"],
  "context_text": "---\nEmployees are eligible for remote work up to 3 days per week after completing 90 days of employment...\n---\nFlexible working hours are available between 7am and 7pm...\n---\nWork-from-home equipment allowance is $500 per calendar year...\n---",
  "results": [
    {
      "item_id": "policy-remote-work-v4",
      "text": "Employees are eligible for remote work up to 3 days per week after completing 90 days of employment. A minimum of 2 days per week must be spent in the office. Remote work requires manager approval and a compliant home office setup.",
      "score": 0.924,
      "metadata": { "category": "hr", "policy_type": "remote-work", "version": "4", "effective_date": "2025-01-01" }
    },
    {
      "item_id": "policy-flexible-hours-v2",
      "text": "Flexible working hours are available between 7am and 7pm local time. Core hours of 10am–3pm must be observed for team collaboration.",
      "score": 0.812,
      "metadata": { "category": "hr", "policy_type": "working-hours", "version": "2" }
    },
    {
      "item_id": "faq-wfh-001",
      "text": "Work-from-home equipment allowance is $500 per calendar year for approved peripheral devices and ergonomic equipment.",
      "score": 0.774,
      "metadata": { "category": "hr", "type": "faq", "topic": "wfh-expenses" }
    }
  ]
}

Handling Zero Results

When no results meet the min_score threshold, the node routes to the success port with result_count: 0 and an empty results array. Always check result_count before attempting to access results[0] — accessing an index on an empty array will cause a downstream expression error.

// Recommended pattern in downstream IfCondition node:
{@ $node.ragNode.result_count } > 0

// True path: use retrieved context in AI prompt
// False path: use a generic fallback response or escalate to human

Expression Reference for Downstream Nodes

Example Output — insert

{
  "item_id": "policy-remote-work-v4",
  "operation": "inserted",
  "collection_name": "hr-policies",
  "embedding_dimensions": 1536,
  "timestamp": "2026-05-23T09:14:22.481Z"
}

Example Output — query (zero results)

{
  "query_text": "What is the policy for company vehicles?",
  "result_count": 0,
  "results": [],
  "sources": [],
  "context_text": "",
  "collection_name": "hr-policies"
}

When result_count is 0, either no items exist in the collection matching the query above the min_score threshold, or no items matching the metadata_filter were found at all. Handle this in a downstream IfCondition to route to a fallback response.

Example: Building a RAG Prompt in a DataMapping Node

// Use a DataMapping node between FlowRag query and AI node
// to construct a well-formatted prompt:
{
  "prompt": "You are an HR assistant for BizFirstGO. Answer the employee's question using the company policies below. If the answer is not covered by the policies, say so clearly.\n\nCompany Policies:\n{@ $node.ragQuery.context_text }\n\nEmployee Question: {@ $var.employeeQuestion }\n\nAnswer in a friendly, professional tone in 2-4 sentences.",
  "cited_policies": "{@ $node.ragQuery.sources }"
}

insert and update — Additional Output Fields