query_knowledge
Semantic search over a brand's knowledge base — website content, uploaded documents, and training materials.
Semantic search over a brand's knowledge base — website content, uploaded documents, and training materials. Use this to find specific facts, claims, product details, or positioning language when writing content.
Returns the most relevant text chunks ranked by similarity.
Authentication: Required
Parameters
| Parameter | Type | Required | Default | Description |
|---|---|---|---|---|
syntheticId | integer | Yes | — | The agent ID to search knowledge for. |
query | string | Yes | — | Natural language search query. Examples: "pricing tiers and features", "company founding story", "integration with Salesforce". |
topK | integer (1–20) | No | 8 | Number of results to return. |
Response
Key fields
| Field | Description |
|---|---|
chunks[].score | Similarity score (0–1). Higher is more relevant. |
chunks[].text | The text content of the matched chunk. |
chunks[].sourceType | Where this content came from: "webpage", "document", "training". |
chunks[].sourceUrl | Original URL of the source, if applicable. |
Tips
- Use natural language queries rather than keywords.
"How does the product handle enterprise security?"works better than"enterprise security". - Increase
topKif you want more context or aren't finding what you need. - Combine with
get_factsheetfor a broader overview before diving into specific details.