qdrant/skills

Reference skill for the officially supported Qdrant client SDKs — Python, JavaScript/TypeScript, Rust, Go, .NET, and Java — plus REST and gRPC API guidance for integrating with a Qdrant deployment.

Diagnose and improve Qdrant search relevance — isolate whether the issue is the embedding model, chunking, HNSW tuning, quantization, or query strategy, with sub-skills for diagnosis/tuning and hybrid/reranking search strategies.

Guides coding agents through Qdrant performance tuning — search speed, memory usage, and indexing throughput — with concrete decisions around quantization, HNSW parameters, and payload indexing.

Official Qdrant skill for **monitoring & observability** — splits into Monitoring Setup (Prometheus scraping, health probes, Hybrid Cloud specifics, alerting, log centralization) and Debugging with Metrics (optimizer stuck, memory growth, slow requests); branches based on whether the user is setting up monitoring or diagnosing an active production issue.

Official Qdrant skill for **scaling decisions** — first nails down the goal (data volume / query throughput (QPS) / query latency / query volume), then routes to the right playbook via sub-skills (`scaling-data-volume`, `scaling-qps`, `minimize-latency`, `scaling-query-volume`) since throughput and latency tuning pull in opposite directions.

Official Qdrant skill for **deployment options** — guides coding agents through choosing between local development, self-hosted (Docker/Kubernetes), Qdrant Cloud, and hybrid deployment modes, with configuration and trade-off analysis for each approach.

Official Qdrant skill for **embedding-model migration** without downtime — enforces "you must create a new collection, named vectors can't be added later", offers zero-downtime alias swap, side-by-side multi-vector A/B, Matryoshka-dimensions shortcut, dense→hybrid (BM25) path, and bulk-re-embed tips (`update_mode: insert`, disabled HNSW, parallel batches, ColBERT co-location warning).

Official Qdrant skill covering **safe version upgrades** — compatibility guarantees, rolling upgrade procedures, and upgrade path validation so coding agents can confidently navigate Qdrant version transitions without downtime or data loss.