markdown-docs_skill

Overview

This skill provides concise, practical documentation for Materialize: a streaming SQL database for real-time analytics. It covers SQL syntax, data ingestion, core concepts, cluster management, and best practices for building incrementally maintained views and low-latency queries. Use it to get targeted guidance on queries, sources, sinks, views, and operational tasks.

How this skill works

The skill inspects organized documentation by topic: SQL command reference, concepts, ingestion patterns, transformations, and operational guidance. For SQL questions it consults the SQL directory; for concepts it reads the concepts pages; for ingestion and transform workflows it reads the ingest-data and transform-data sections. It synthesizes examples, common commands, and practical steps for real deployments.

When to use it

• You need correct Materialize SQL syntax or examples for CREATE SOURCE, CREATE MATERIALIZED VIEW, CREATE SINK, SELECT, or CREATE INDEX.
• You plan to ingest change data capture (CDC) or streaming data from Kafka, Debezium, MySQL, PostgreSQL, MongoDB, or webhooks.
• You want guidance on clusters, namespaces, replicas, or self-managed vs cloud deployments.
• You need patterns for low-latency, incrementally maintained analytics or view optimization.
• You are troubleshooting dataflow, monitoring ingestion, or preparing disaster recovery and upgrades.

Best practices

• Prefer CREATE MATERIALIZED VIEW for frequently queried aggregations to get incremental updates and low latency.
• Create indexes on materialized views for hot lookup paths rather than indexing raw sources.
• Ingest via CDC connectors (Debezium, Postgres/MySQL connectors) to preserve change semantics for consistent views.
• Monitor freshness and reaction time; set alerts on lag metrics and stalled dataflows.
• Use namespaces and clusters to isolate workloads and enforce resource boundaries in production.

Example use cases

• Build a real-time dashboard with a materialized view that aggregates events from Kafka and serves SELECT and SUBSCRIBE queries.
• Stream CDC from MySQL via Debezium into Materialize, create views for business metrics, and export results to a downstream sink.
• Optimize a slow analytical query by adding an index on a materialized view instead of scanning raw streaming tables.
• Deploy a self-managed Materialize cluster with operator configuration and upgrade guidelines for production.