Faz Docs

Connect faz to a Qdrant vector database. Intent-based JSON queries — search, scroll, count, recommend.

Qdrant's connector talks to the REST API. Queries are JSON with an intent field that picks the operation. Filters use Qdrant's payload language; vectors are caller-supplied.

Default port	Query language	Write support	Schema discovery	Driver
6333	Intent-based JSON	Yes (RBAC-gated)	Collection enumeration	`qdrant-client`

Quick example

faz.yaml

databases:
  - name: <database>          # used as --database <database> in queries
    type: qdrant
    host: localhost
    port: 6333
    ssl: false
    extra:
      api_key: <api-key>      # leave empty if your Qdrant doesn't require auth

permissions:
  - database: <database>      # must match `name:` above
    access: R
    # access codes:
    # R    read only           — search, scroll, count, recommend
    # W    write only          — upsert, delete (no reads — blocked at connector)
    # RW   read + write
    # RA   read + append       — reads + upsert
    # RWA  read + write, no DELETE
    # A    admin (incl. collection DDL — blocked at connector)

See Permissions for the full model, per-table overrides, and the operation matrix.

faz query '{"intent":"search","vector":[<your-vector>],"limit":5}' --database <database> --table <collection> --language vector

uv run faz query '{"intent":"search","vector":[<your-vector>],"limit":5}' --database <database> --table <collection> --language vector

python -m faz query '{"intent":"search","vector":[<your-vector>],"limit":5}' --database <database> --table <collection> --language vector

Configuration

Field	Type	Default	Notes
`host`	string	`localhost`	Qdrant host.
`port`	integer	`6333`	REST API port. gRPC port (6334) isn't exposed by the connector.
`database`	string	`""`	Not used. Collections are addressed via the IR step's `table`.
`username`	string	`""`	Not used.
`password`	string	`""`	Not used.
`ssl`	boolean	`false`	When `true`, uses HTTPS.
`extra`	mapping	`{}`	`api_key` for managed Qdrant Cloud or hosted setups.

Capabilities

Supported intents:

Intent	Op class	Notes
`search`	SELECT	Vector similarity. `vector`, `filter`, `limit`, `score_threshold`.
`scroll`	SELECT	Pagination through a collection without similarity scoring.
`count`	SELECT	Optional `filter`.
`recommend`	SELECT	Provide positive/negative point ids; finds similar candidates.

The query body's filter field uses Qdrant's payload-filter language directly; the connector passes it through to the client.

Schema discovery enumerates collections and reports their vector dimensions and payload schema.

Limitations

Write intents are blocked at the connector level: upsert, delete, create_collection, delete_collection. Qdrant's data plane is reachable through faz only for reads. Use Qdrant's own client for write workflows.
No batch search. A single search call submits one query vector. Batch queries need multiple calls (or a federated query).
Filter language is Qdrant-native — different from Weaviate's filters. Reads on Weaviate aren't directly portable.

Troubleshooting

ConnectionError / unable to connect — Qdrant isn't listening on host:port. curl http://host:port/ should return a JSON banner.

401 Unauthorized — extra.api_key is missing or wrong. Test via curl: curl -H "api-key: $QDRANT_API_KEY" http://host:port/collections.

Collection not found — the table you submitted doesn't exist. List collections: curl http://host:port/collections.

Vector dimension mismatch — your vector length doesn't match the collection's configured dimension. Qdrant returns this directly; faz passes it through.

For the broader troubleshooting flow, see Connection failed.

Databases overview — cross-connector basics.
Weaviate — sibling vector database with a richer intent vocabulary.
Milvus — sibling vector database.
faz.yaml — the full config schema.

Qdrant

Quick example

Configuration

Capabilities

Limitations

Troubleshooting

Related

On this page