Knowledge Search Node
Search knowledge records (Tier 2) and append matching results to records
Last updated 24 Jun 2026
The Knowledge Search node performs a full-text search against knowledge records (Tier 2 of the tiered knowledge system) and appends the best-matching result's fields to each record. Unlike RAG Query, this node uses keyword search rather than vector similarity.
When to Use
Use Knowledge Search when you want to enrich records with structured knowledge records — for example, looking up product specs, territory rules, or reference data that was previously saved via the Knowledge Records UI.
Configuration
| Field | Type | Description |
|---|---|---|
queryColumn | string | Record field whose value is used as the search query |
queryTemplate | string | Template string for the search query when no queryColumn is set. Supports {{fieldName}} interpolation (\w+ identifiers only) |
contentTypes | string[] | Optional filter by content type (insight, pattern, learning, outcome_insight, decision_rationale) |
topics | string[] | Optional filter by topic tags (records must match all specified topics) |
limit | number | Maximum results to retrieve per record (1–50, default: 5) |
outputMode | enrich | expand | How to attach results to records (see below) |
Output Modes
enrich — Appends four scalar fields from the top matching result to each input record. One output record per input record, even when no results are found (null fields).
Added fields:
| Field | Description |
|---|---|
_kr_title | Title of the best matching knowledge record |
_kr_content | Content of the best matching knowledge record |
_kr_content_type | Content type of the best matching record |
_kr_confidence | Confidence score of the best matching record |
expand — Emits one output record per matched knowledge result, merging its fields into the input record. Input records with no matches are kept with all _kr_* fields set to null.
Added fields (superset of enrich, plus):
| Field | Description |
|---|---|
_kr_id | Knowledge record ID |
_kr_topics | Topics (comma-separated) |
Example
Look up relevant product documentation for each lead's industry:
queryColumn: industry
contentTypes: [product_specs]
limit: 3
outputMode: enrich
Each record gets _kr_title, _kr_content, _kr_content_type, and _kr_confidence fields from the top matching product spec entry. Records with no match get those fields as null.
Notes
- Knowledge records are created via the Knowledge Records UI under Settings → Knowledge (this node only reads them).
- In
expandmode, records with no matches are not dropped — they are kept with null_kr_*fields. - For semantic (vector) search, use RAG Query instead.
Related articles
Still stuck? Browse the FAQ or contact us.