Clickhouse Best Practices

Comprehensive guidance for ClickHouse covering schema design, query optimization, data ingestion, and AI agent connectivity. Contains 31 rules across 4 main categories (schema, query, insert, agent), prioritized by impact.

Official docs: ClickHouse Best Practices (opens in a new tab)

IMPORTANT: How to Apply This Skill

Before answering ClickHouse questions, follow this priority order:

1. Check for applicable rules in the rules/ directory
2. If rules exist: Apply them and cite them in your response using “Per rule-name…”
3. If no rule exists: Use the LLM’s ClickHouse knowledge or search documentation
4. If uncertain: Use web search for current best practices
5. Always cite your source: rule name, “general ClickHouse guidance”, or URL

Why rules take priority: ClickHouse has specific behaviors (columnar storage, sparse indexes, merge tree mechanics) where general database intuition can be misleading. The rules encode validated, ClickHouse-specific guidance.

Agent Connectivity & Query Workflow

Before querying ClickHouse, agents must establish a connection and follow the discovery workflow:

1. rules/agent-connect-mcp.md - Connection setup (MCP + CLI), credential discovery, output format selection
2. rules/agent-discovery-schema.md - CRITICAL: 7-step schema discovery workflow
3. rules/agent-query-safety.md - CRITICAL: LIMIT, timeouts, progressive exploration

Every agent session should follow this sequence:

1. Connect — establish connection via MCP or CLI (see agent-connect-mcp)
2. Discover — databases → tables → columns + comments → sort keys → skip indexes → sample → EXPLAIN
3. Plan — use sort key and skip index knowledge to write efficient WHERE clauses
4. Execute — run queries with LIMIT and timeouts
5. Recover — on timeout/memory errors, narrow filters and retry (see agent-query-safety)

Subagent architecture notes

If your system dispatches ClickHouse tasks to specialized subagents:

• Schema discovery + query execution: any model — the steps are procedural
• EXPLAIN analysis + query optimization: benefits from mid-tier reasoning
• Schema design review against all 28 rules: benefits from mid-tier reasoning

Review Procedures

For Schema Reviews (CREATE TABLE, ALTER TABLE)

Read these rule files in order:

1. rules/schema-pk-plan-before-creation.md - ORDER BY is immutable
2. rules/schema-pk-cardinality-order.md - Column ordering in keys
3. rules/schema-pk-prioritize-filters.md - Filter column inclusion
4. rules/schema-types-native-types.md - Proper type selection
5. rules/schema-types-minimize-bitwidth.md - Numeric type sizing
6. rules/schema-types-lowcardinality.md - LowCardinality usage
7. rules/schema-types-avoid-nullable.md - Nullable vs DEFAULT
8. rules/schema-partition-low-cardinality.md - Partition count limits
9. rules/schema-partition-lifecycle.md - Partitioning purpose

Check for:

• Not done: PRIMARY KEY / ORDER BY column order (low-to-high cardinality)
• Not done: Data types match actual data ranges
• Not done: LowCardinality applied to appropriate string columns
• Not done: Partition key cardinality bounded (100-1,000 values)
• Not done: ReplacingMergeTree has version column if used

For Query Reviews (SELECT, JOIN, aggregations)

Read these rule files:

1. rules/query-join-choose-algorithm.md - Algorithm selection
2. rules/query-join-filter-before.md - Pre-join filtering
3. rules/query-join-use-any.md - ANY vs regular JOIN
4. rules/query-index-skipping-indices.md - Secondary index usage
5. rules/schema-pk-filter-on-orderby.md - Filter alignment with ORDER BY

Check for:

• Not done: Filters use ORDER BY prefix columns
• Not done: JOINs filter tables before joining (not after)
• Not done: Correct JOIN algorithm for table sizes
• Not done: Skipping indices for non-ORDER BY filter columns

For Insert Strategy Reviews (data ingestion, updates, deletes)

Read these rule files:

1. rules/insert-batch-size.md - Batch sizing requirements
2. rules/insert-mutation-avoid-update.md - UPDATE alternatives
3. rules/insert-mutation-avoid-delete.md - DELETE alternatives
4. rules/insert-async-small-batches.md - Async insert usage
5. rules/insert-optimize-avoid-final.md - OPTIMIZE TABLE risks

Check for:

• Not done: Batch size 10K-100K rows per INSERT
• Not done: No ALTER TABLE UPDATE for frequent changes
• Not done: ReplacingMergeTree or CollapsingMergeTree for update patterns
• Not done: Async inserts enabled for high-frequency small batches

Output Format

Structure your response as follows:

## Rules Checked
- `rule-name-1` - Compliant / Violation found
- `rule-name-2` - Compliant / Violation found
...

## Findings

### Violations
- **`rule-name`**: Description of the issue
  - Current: [what the code does]
  - Required: [what it should do]
  - Fix: [specific correction]

### Compliant
- `rule-name`: Brief note on why it's correct

## Recommendations
[Prioritized list of changes, citing rules]

Rule Categories by Priority

Quick Reference

Schema Design - Primary Key (CRITICAL)

• schema-pk-plan-before-creation - Plan ORDER BY before table creation (immutable)
• schema-pk-cardinality-order - Order columns low-to-high cardinality
• schema-pk-prioritize-filters - Include frequently filtered columns
• schema-pk-filter-on-orderby - Query filters must use ORDER BY prefix

Schema Design - Data Types (CRITICAL)

• schema-types-native-types - Use native types, not String for everything
• schema-types-minimize-bitwidth - Use smallest numeric type that fits
• schema-types-lowcardinality - LowCardinality for <10K unique strings
• schema-types-enum - Enum for finite value sets with validation
• schema-types-avoid-nullable - Avoid Nullable; use DEFAULT instead

Schema Design - Partitioning (HIGH)

• schema-partition-low-cardinality - Keep partition count 100-1,000
• schema-partition-lifecycle - Use partitioning for data lifecycle, not queries
• schema-partition-query-tradeoffs - Understand partition pruning trade-offs
• schema-partition-start-without - Consider starting without partitioning

Schema Design - JSON (MEDIUM)

• schema-json-when-to-use - JSON for dynamic schemas; typed columns for known

Query Optimization - JOINs (CRITICAL)

• query-join-choose-algorithm - Select algorithm based on table sizes
• query-join-use-any - ANY JOIN when only one match needed
• query-join-filter-before - Filter tables before joining
• query-join-consider-alternatives - Dictionaries/denormalization vs JOIN
• query-join-null-handling - join_use_nulls=0 for default values

Query Optimization - Indices (HIGH)

• query-index-skipping-indices - Skipping indices for non-ORDER BY filters

Query Optimization - Materialized Views (HIGH)

• query-mv-incremental - Incremental MVs for real-time aggregations
• query-mv-refreshable - Refreshable MVs for complex joins

Insert Strategy - Batching (CRITICAL)

• insert-batch-size - Batch 10K-100K rows per INSERT

Insert Strategy - Async (HIGH)

• insert-async-small-batches - Async inserts for high-frequency small batches
• insert-format-native - Native format for best performance

Insert Strategy - Mutations (CRITICAL)

• insert-mutation-avoid-update - ReplacingMergeTree instead of ALTER UPDATE
• insert-mutation-avoid-delete - Lightweight DELETE or DROP PARTITION

Insert Strategy - Optimization (HIGH)

• insert-optimize-avoid-final - Let background merges work

Agent Integration - Discovery (CRITICAL)

• agent-discovery-schema - Always discover schema before querying

Agent Integration - Safety (CRITICAL)

• agent-query-safety - LIMIT, timeouts, progressive exploration

Agent Integration - Connectivity + Formats (HIGH)

• agent-connect-mcp - MCP + CLI setup, credential discovery, output format selection

When to Apply

This skill activates when you encounter:

• AI agent connecting to ClickHouse (MCP, CLI, HTTP)

• Agent workflow design for ClickHouse

• Schema discovery or exploration requests

• CREATE TABLE statements

• ALTER TABLE modifications

• ORDER BY or PRIMARY KEY discussions

• Data type selection questions

• Slow query troubleshooting

• JOIN optimization requests

• Data ingestion pipeline design

• Update/delete strategy questions

• ReplacingMergeTree or other specialized engine usage

• Partitioning strategy decisions

Rule File Structure

Each rule file in rules/ contains:

• YAML frontmatter: title, impact level, tags
• Brief explanation: Why this rule matters
• Incorrect example: Anti-pattern with explanation
• Correct example: Best practice with explanation
• Additional context: Trade-offs, when to apply, references

Full Compiled Document

For the complete guide with all rules expanded inline: AGENTS.md

Use AGENTS.md when you need to check multiple rules quickly without reading individual files.