@unwarkz/n8n-nodes-qdrant

N8N community node for Qdrant vector store — embed and store documents, perform semantic search, and manage collections from n8n AI Agent workflows.

Features

Qdrant Store AI Tools Node

Embed text and binary files into a Qdrant vector collection using a connected Embedding model sub-node. Designed for building knowledge bases that AI Agents can write to.

| Tool | Description | |------|-------------| | qdrant_store_text | Embed a text string and store it as a vector point | | qdrant_store_binary_file | Send a binary file directly to the multimodal embedder and store the result | | qdrant_batch_store_texts | Embed and store multiple texts in a single batch operation | | qdrant_create_collection | Create a new Qdrant collection with vector configuration | | qdrant_delete_collection | Permanently delete a collection and all its data |

Node inputs:

• Embedding (ai_embedding, required) — any n8n-compatible embedding model sub-node (OpenAI, Cohere, HuggingFace, Ollama, etc.)

Binary file embedding strategy (tried in order):

1. Raw Buffer — native multimodal embedders (CLIP, Vertex AI multimodal)
2. data:<mimeType>;base64,<data> URI string — embedders that accept image data-URIs
3. filename: …\nfiletype: …\ncontent_base64: … text — any text embedder as fallback

Qdrant AI Tools Node

Full Qdrant operations node for reading, querying, and managing a Qdrant vector store from an AI Agent.

| Tool | Description | |------|-------------| | qdrant_search | Semantic vector search (requires Embedding sub-node); auto-reranked if Reranker is connected | | qdrant_scroll | Page through all points in a collection with optional filter | | qdrant_get_points | Retrieve specific points by their IDs | | qdrant_upsert_points | Insert or update points with pre-computed vectors | | qdrant_delete_points | Delete points by ID list or filter expression | | qdrant_count_points | Count points matching a filter | | qdrant_recommend | Find similar points using positive/negative example IDs | | qdrant_list_collections | List all collections in the Qdrant instance | | qdrant_get_collection_info | Get detailed config and stats for a collection | | qdrant_create_collection | Create a new collection | | qdrant_delete_collection | Permanently delete a collection | | qdrant_update_collection | Update optimizer, HNSW, and quantization settings | | qdrant_create_snapshot | Create a collection snapshot for backup/export | | qdrant_list_snapshots | List available snapshots for a collection |

Node inputs (both optional):

• Embedding (ai_embedding) — required for qdrant_search; embeds the query vector
• Reranker (ai_reranker) — when connected, search results are automatically reranked using the LangChain BaseDocumentCompressor interface before being returned

Prerequisites

1. A running Qdrant instance:

   docker run -p 6333:6333 qdrant/qdrant

2. n8n version ≥ 1.0.0

Installation

npm install @unwarkz/n8n-nodes-qdrant

Or install via the n8n UI: Settings → Community Nodes → Install → enter @unwarkz/n8n-nodes-qdrant.

Configuration

Create a Qdrant API credential with:

• Qdrant URL: URL of your Qdrant instance (default: http://localhost:6333; for Qdrant Cloud use https://<cluster-id>.cloud.qdrant.io)
• API Key: Qdrant API key (required for Qdrant Cloud; leave empty for unauthenticated local instances)

Usage

Store text from an AI Agent

1. Add a Qdrant Store AI Tools node
2. Connect an Embedding model sub-node (e.g., OpenAI Embeddings) to its Embedding input
3. Create a credential for your Qdrant instance
4. Set Default Collection to the target collection name
5. Connect the Tool output to an AI Agent node's Tools input

The agent can now call qdrant_create_collection to create the collection, then qdrant_store_text or qdrant_batch_store_texts to populate it.

Semantic search with reranking

1. Add a Qdrant AI Tools node
2. Connect an Embedding sub-node to its Embedding input
3. Optionally connect a Reranker sub-node (e.g., Cohere Reranker) to its Reranker input
4. Connect the Tool output to an AI Agent node's Tools input

The agent can now call qdrant_search. If a reranker is connected, results are automatically reranked before being returned.

Store a binary file (image, document, media)

The qdrant_store_binary_file tool reads a binary property from the global binary registry (compatible with files stored by Telegram Bot AI Tools, Gotenberg AI Tools, etc.) and passes the raw buffer directly to the embedding model:

AI Agent → qdrant_store_binary_file(binary_property_name="telegram_file_0")
         → QdrantStoreAiTools reads buffer from _n8nBinaryRegistry
         → sends Buffer to embedder.embedDocuments([buffer])
         → stores resulting vector in Qdrant

Cross-Tool Binary Interoperability

This package uses the same global._n8nBinaryRegistry pattern as other @unwarkz AI tool nodes. Binary files produced by one AI tool (e.g., telegram_get_file) are automatically accessible to this node via their binaryPropertyName, with no manual file passing required.

Part of the Assistant Set

This package is also included in the full @unwarkz/n8n-nodes-assistant-set package.

License

MIT © unwarkz