w3-mcp-server-qdrant
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| QDRANT_URL | No | URL of the Qdrant server | http://localhost:6333 |
| QDRANT_API_KEY | No | Optional API key for Qdrant authentication | |
| OLLAMA_BASE_URL | No | URL of the Ollama server | http://localhost:11434 |
| OLLAMA_EMBED_MODEL | No | Embedding model to use for query embedding | bge-m3:latest |
| OLLAMA_RERANK_MODEL | No | LLM model for query expansion, HyDE, and reranking | mistral |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| qdrant_searchA | Search for similar documents in Qdrant. Embeds the query text using Ollama, then searches for similar vectors in the specified Qdrant collection. Returns matching documents with similarity scores. Supports advanced features:
Args: params (SearchInput): Validated parameters: - collection_name (str): Collection to search in - query_text (str): Text to search for (auto-embedded) - limit (int): Max results, 1-100 (default: 5) - score_threshold (float): Min similarity 0.0-1.0 (default: 0.0) - fields (str): Comma-separated metadata fields to return (optional) - response_format (str): 'markdown' or 'json' - expand_query (bool): Enable query expansion (default: False) - expand_query_count (int): Number of variations (default: 3) - use_hyde (bool): Enable HyDE (default: False) - hyde_combine_original (bool): Include original query with HyDE (default: True) - rerank (bool): Enable LLM reranking (default: False) - rerank_top_n (int): Candidates for reranking (default: 10) Returns: str: Formatted search results with document IDs, texts, and scores Errors: - Collection not found: "Collection 'xyz' does not exist" - Embedding failed: "Failed to embed query text" - Connection error: "Cannot connect to Qdrant at {url}" |
| qdrant_list_collectionsA | List all collections in Qdrant. Retrieves metadata about all collections including point counts and vector dimensions. Args: params (ListCollectionsInput): Validated parameters: - response_format (str): 'markdown' or 'json' Returns: str: Formatted list of collections with metadata Errors: - Connection error: "Cannot connect to Qdrant at {url}" |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 2 tools
The two tools have clearly distinct purposes: listing collections vs. searching within a collection. No overlap or ambiguity.
Both tools follow a consistent 'qdrant_verb_noun' pattern (list_collections, search), making it predictable.
Only two tools for a vector database feels thin; typical usage would benefit from additional CRUD operations. However, for a focused query-only server, it might be borderline acceptable.
Significant gaps: no tools to create/delete collections, insert/update/delete points. Agent cannot populate or manage data, only list and search existing content.