Qdrant MCP Server
Official Qdrant MCP server that gives LLM agents a semantic memory layer: agents can store information with `qdrant-store` and retrieve semantically similar content with `qdrant-find`, backed by Qdrant vector database and FastEmbed embeddings.
Score Breakdown
⚙ Agent Friendliness
🔒 Security
API key auth for Qdrant Cloud. Self-hosted: API key optional. TLS configurable. Collection-level access control via API keys in newer versions. Qdrant stores vector embeddings — collection-level isolation important for multi-tenant.
⚡ Reliability
Best When
You need an agent to accumulate and semantically recall knowledge across sessions, and you already run (or are willing to run) Qdrant locally or in Qdrant Cloud.
Avoid When
Your retrieval needs are primarily keyword-based or structured, or you want a zero-infrastructure managed memory solution without operating a vector database.
Use Cases
- • Persistent semantic memory for AI coding assistants (store and retrieve code snippets, decisions, notes)
- • RAG augmentation: let an agent store ingested documents and search them during reasoning
- • Cross-session knowledge retention for long-running agentic workflows
Not For
- • Exact keyword search (use Elasticsearch or similar full-text engines)
- • Structured relational queries requiring SQL-style filtering
- • Teams who do not already operate a Qdrant instance or are not willing to run one
Interface
Authentication
Optional QDRANT_API_KEY env var for Qdrant Cloud or secured self-hosted instances. No auth required for local/open Qdrant.
Pricing
Apache-2.0 open source. Qdrant itself is open source for self-hosting; Qdrant Cloud has its own pricing.
Agent Metadata
Known Gotchas
- ⚠ Cannot specify both QDRANT_URL and QDRANT_LOCAL_PATH simultaneously.
- ⚠ Default embedding model (all-MiniLM-L6-v2) is downloaded at first run; initial startup may be slow.
- ⚠ qdrant-store does not deduplicate; repeated writes create duplicate vectors.
- ⚠ COLLECTION_NAME must be set or passed per-call; omitting both causes an error.
Alternatives
Full Evaluation Report
Detailed scoring breakdown, competitive positioning, security analysis, and improvement recommendations for Qdrant MCP Server.
Scores are editorial opinions as of 2026-03-06.