Hey HN, I am Álvaro and I am working on ObsessionDB, a managed ClickHouse infra as a service. We run CH at about 100B rows scale and we spent the last few years learning what works and what doesn't at that scale.

These are open-source skills that teach Claude Code, Cursor, or any agent the ClickHouse patterns that matter.

The problem is that AI agents write syntactically correct ClickHouse SQL, but they don't know your query patterns. They'll suggest `ORDER BY (timestamp, user_id)` because that's what most examples online show. Any ClickHouse dev knows to put your filter column first, but the agent doesn't. Every conversation starts fresh.

I got tired of reviewing the same issues. The agent isn't stupid, it's just uninformed. So I wrote down what I'd tell a new engineer: ORDER BY column selection, why PREWHERE matters, when to use LowCardinality, partition strategies, materialized view gotchas.

Three skills so far: - Schema Design - ORDER BY selection, partitioning, engine choice - Query Optimization - PREWHERE, column selection, join strategies - Materialized Views - Aggregation patterns, refresh timing

Each pattern has a priority (CRITICAL/HIGH/MEDIUM) so the agent knows what to check first.

npx skills add obsessiondb/clickhouse-skills

Curious what patterns I'm missing or what other devs have experienced. What ClickHouse lessons should your agent know?