Retrieve
Category: RAG
Node id: retrieve
Performs vector similarity search and retrieval from Qdrant collections with filtering and score thresholds
What it does
- Vector similarity search against Qdrant collections
- Configurable top-K result retrieval
- Minimum score threshold filtering
- Qdrant filter conditions for metadata-based narrowing
- Automatic Qdrant client initialization from config
Use cases
- Retrieving relevant documents for RAG pipelines
- Semantic search across clinical knowledge bases
- Finding similar patient records by embedding similarity
Runtime
- Performance tier:
fast
Input ports
| Name | Type | Required | Description |
|---|---|---|---|
query_vector | json | Yes | Query embedding vector ([]float32 or []interface{}) |
Output ports
| Name | Type | Required | Description |
|---|---|---|---|
result | json | No | Main result containing results array, count, collection, top_k, min_score, and total_results |
success | boolean | No | Whether the operation succeeded |
Settings
Basic
| Setting | Name | Type | Required | Default | Allowed values | Description |
|---|---|---|---|---|---|---|
| Collection | collection | string | Yes | Qdrant collection name | ||
| Top K | top_k | number | No | 5 | min 1; max 100 | Number of results to retrieve |
| Min Score | min_score | number | No | 0 | min 0; max 1 | Minimum similarity score threshold |
| Qdrant URL | qdrant_url | string | No | http://localhost:6333 | Qdrant server URL |
Auth
| Setting | Name | Type | Required | Default | Allowed values | Description |
|---|---|---|---|---|---|---|
| API Key | api_key | secret | No | Qdrant API key |
Advanced
| Setting | Name | Type | Required | Default | Allowed values | Description |
|---|---|---|---|---|---|---|
| Filter | filter | json | No | Qdrant filter conditions |
Errors
| Code | Description |
|---|---|
COLLECTION_REQUIRED | Collection name is not configured |
VECTOR_REQUIRED | No query_vector input was provided |
INVALID_VECTOR | query_vector has invalid element types |
SEARCH_FAILED | Qdrant vector search request failed |
Runtime notes
- total_results is the raw count before min_score filtering; count is after filtering
- query_vector must match the dimensionality of the target Qdrant collection
- Wire result.results to a rerank node's results input for two-stage retrieval
- If qdrant_url is not set in config, defaults to http://localhost:6333
See Build any workflow for how to put nodes together and Ports, types and data flow for how port types connect.
Quelque chose n'est pas clair ? Envoyez vos commentaires sur la documentation par courriel.