mcp-qdrant mcp

v1.0.0 · MCP Tool · database · registry.pascalai.org

Qdrant vector database access via REST API. Operations: list_collections, create_collection, delete_collection, count, upsert, search, get_point, delete_points, scroll. Required: host. Optional: port (default 6333), apiKey.

qdrantvectorembeddingssimilaritydatabase

Input Parameters

Parameter	Type	Description
operationrequired	string	Operation to perform. One of: `list_collections`, `create_collection`, `delete_collection`, `count`, `upsert`, `search`, `get_point`, `delete_points`, `scroll`.
hostrequired	string	Qdrant host (default: localhost). Default: `localhost`.
portoptional	integer	Qdrant port (default: 6333). Default: `6333`.
apiKeyoptional	string	Qdrant API key (optional).
collectionoptional	string	Collection name.
pointsoptional	string	JSON array of points [{id,vector,payload?}] for upsert.
vectoroptional	string	Float array JSON for search.
pointIdoptional	string	Point ID for get_point.
idsoptional	string	JSON array of point IDs for delete_points.
filteroptional	string	JSON filter object.
limitoptional	integer	Max results (default: 10). Default: `10`.
vectorSizeoptional	integer	Vector dimension for create_collection.
distanceoptional	string	Distance metric: Cosine, Euclid, Dot (default: Cosine). Default: `Cosine`.
withPayloadoptional	boolean	Include payload in results (default: true). Default: `True`.
withVectoroptional	boolean	Include vectors in results (default: false). Default: `False`.
scoreThresholdoptional	number	Minimum score threshold for search.
offsetoptional	integer	Scroll offset.

Output Fields

Field	Type	Description
collections	array[object]
results	array[object]
count	integer
ok	boolean
error	string

Examples

Search vectors

// Input
{
  "operation": "search",
  "host": "localhost",
  "collection": "my_vectors",
  "vector": "[0.1,0.2,0.3]",
  "limit": 5
}

// Output
{
  "results": [
    {
      "id": "1",
      "score": 0.95
    }
  ]
}

Install & Discovery

Install

ppm install mcp-qdrant

Get JSON Schema

GET /v1/packages/mcp-qdrant/1.0.0/schema

Discover by keyword

GET /v1/mcp/discover?q=qdrant

Discovery hint: Install with ppm install mcp-qdrant or invoke remotely via POST /v1/invoke/mcp-qdrant on the MCP Service.

PascalAI Usage

uses toolslib;
var Tool := LoadTool('mcp-qdrant');
var R := Tool.Call(JsonObj(['operation','search','host','localhost','collection','my_vecs','vector','[0.1,0.2]','limit',5]));
Writeln(R.ToJSON);