Block: Qdrant Block — Zelaxy Docs

The Qdrant block lets you store, search, and retrieve vector embeddings using a Qdrant vector database. Reach for it when you need semantic similarity search or to manage vector collections directly from a workflow.

Overview

Property	Value
Type	`qdrant`
Category	tools
Color	`#1A223F`

When to Use

Upsert embeddings (with optional payloads) into a Qdrant collection after generating them with an Agent or embedding provider.
Run semantic similarity searches over a collection using a query vector to power RAG pipelines.
Filter search results by payload metadata (e.g. only points where city matches London).
Fetch specific points by their IDs to retrieve stored payloads or raw vectors on demand.
Write embeddings from one step and read them back in a later step of the same workflow.
Connect to a self-hosted Qdrant instance or Qdrant Cloud with an API key.

Configuration

Operation

Dropdown that selects the action to perform. Defaults to upsert.

Label	ID	Description
Upsert	`upsert`	Insert or update points in a collection
Search	`search`	Find similar vectors using a query vector
Fetch	`fetch`	Retrieve specific points by their IDs

Upsert operation fields

Sub-block	ID	Type	Required	Description
Qdrant URL	`url`	short-input	Yes	Qdrant server endpoint, e.g. `http://localhost:6333`
Collection	`collection`	short-input	Yes	Name of the target collection
Points	`points`	long-input	Yes	JSON array of points, each with `id`, `vector`, and optional `payload`

Search operation fields

Sub-block	ID	Type	Required	Description
Qdrant URL	`url`	short-input	Yes	Qdrant server endpoint
Collection	`collection`	short-input	Yes	Name of the target collection
Query Vector	`vector`	long-input	Yes	Float array used as the similarity query, e.g. `[0.1, 0.2]`
Limit	`limit`	short-input	No	Maximum number of results to return (defaults to 10 in the tool)
Filter	`filter`	long-input	No	Qdrant filter object, e.g. `{"must":[{"key":"city","match":{"value":"London"}}]}`
With Payload	`with_payload`	switch	No	When enabled, includes point payload in results
With Vector	`with_vector`	switch	No	When enabled, includes the raw vector in results

Fetch operation fields

Sub-block	ID	Type	Required	Description
Qdrant URL	`url`	short-input	Yes	Qdrant server endpoint
Collection	`collection`	short-input	Yes	Name of the target collection
IDs	`ids`	long-input	Yes	JSON array of point identifiers to retrieve, e.g. `["370446a3-310f-58db-8ce7-31db947c6c1e"]`
With Payload	`with_payload`	switch	No	When enabled, includes point payload in results
With Vector	`with_vector`	switch	No	When enabled, includes the raw vector in results

API Key

Sub-block	ID	Type	Required	Description
API Key	`apiKey`	short-input	Yes	Qdrant API key used to authenticate requests. Stored as a secret; use `{{QDRANT_API_KEY}}`. Pass empty string if your instance has no auth.

Inputs & Outputs

Inputs:
- operation (string) — Operation to perform (upsert, search, or fetch)
- url (string) — Qdrant server URL
- apiKey (string) — Qdrant API key
- collection (string) — Collection name
- points (json) — Points to upsert (upsert operation)
- vector (json) — Query vector (search operation)
- limit (number) — Result limit (search operation)
- filter (json) — Search filter (search operation)
- ids (json) — Point identifiers to fetch (fetch operation)
- with_payload (boolean) — Include payload in response (search and fetch)
- with_vector (boolean) — Include vectors in response (search and fetch)
Outputs:
- matches (json) — Search matches
- upsertedCount (number) — Number of points upserted
- data (json) — Response data (search results array or fetched points array)
- status (string) — Operation status reported by Qdrant

Tools

Qdrant Upsert Points (qdrant_upsert_points) — Sends a PUT request to /collections/{collection}/points to insert or update an array of points. Each point must have an id, a vector, and an optional payload map. Returns status and data (the raw result from Qdrant).
Qdrant Search Vector (qdrant_search_vector) — Sends a POST request to /collections/{collection}/points/query with a query vector and optional limit, filter, with_payload, and with_vector params. Returns data (array of scored result points) and status.
Qdrant Fetch Points (qdrant_fetch_points) — Sends a POST request to /collections/{collection}/points with an array of IDs and optional with_payload / with_vector flags. Returns data (array of matching points) and status.

YAML Example

qdrant_1:
  type: qdrant
  name: "Qdrant"
  inputs:
    operation: "search"
    url: "http://localhost:6333"
    collection: "my-collection"
    vector: "{{embedder.vector}}"
    limit: 10
    with_payload: true
    apiKey: "{{QDRANT_API_KEY}}"
  connections:
    outgoing:
      - target: next-block-id

Qdrant Block