Clickhouse

ClickHouse is a high-performance column-oriented database designed for real-time OLAP queries over large datasets. Its HTTP API accepts SQL queries and returns results in multiple formats (JSON, CSV, TabSeparated), making it straightforward to integrate with agents that need to run analytics queries. For agents powering real-time dashboards, event aggregation, or log analytics, ClickHouse's sub-second aggregation over billions of rows delivers performance that traditional OLTP databases cannot match at comparable cost.

The HTTP API accepts SQL queries via POST with query parameters for format selection (JSON, JSONEachRow, CSV, etc.). The query results format flexibility enables agents to choose the format most efficient for downstream processing — JSONEachRow for streaming large result sets, JSON for structured objects. The ClickHouse Cloud management API adds cluster and database management operations alongside the query interface.

Authentication uses username/password passed via HTTP Basic Auth or query parameters. ClickHouse Cloud uses token-based authentication for the management API. Teams should configure network-level access controls (IP allowlisting, TLS) for ClickHouse instances given the HTTP API's exposure surface. User-level database permissions restrict query scope to appropriate datasets.

Reliability is production-grade for the column-store architecture, which is inherently resilient via replication and sharding. ClickHouse Cloud provides managed high-availability with automatic replication. Self-hosted ClickHouse requires careful shard and replica configuration for production availability requirements.

Documentation is comprehensive and covers the HTTP API, SQL dialect, and table engine options in depth. The query format documentation is particularly useful for agents — understanding the format options enables efficient result consumption. Teams building agent-driven analytics on ClickHouse will find the documentation thorough, with strong community resources for common query patterns.