§ 01 9 tools
Data
Warehouses, lakehouses, open table formats, analytics engines.
SnowflakeBigQueryDatabricksDuckDBClickHouseTrino
§ Catalog
A vendor-neutral index of tools across the five stack layers plus full-stack platforms. Pick a layer to browse its tools, or open the full filterable catalog for search, license filter, and cross-layer comparison.
§ Gut-check
Before reaching for a framework or platform, ask whether you need one at all. A plain Claude or Codex session pointed at your warehouse can answer a surprising range of business questions. Write a tight system prompt, give the model your schema and metric definitions, and you have a working analytics agent with zero infrastructure. Many teams get real value from exactly this setup.
The limitations show up at scale. When you need agents that remember context across sessions, coordinate across domains, run reliably in production, or produce auditable traces, a bare model session starts to break down. That's when dedicated orchestration, a semantic layer, and governance tooling earn their place. The guide on building specialized analytics agents walks through how to structure that transition, starting from a single CLAUDE.md file and adding infrastructure only where it pays off.
The rest of this page maps the landscape for when you're ready to go further.
§ Layers
Each layer has its own filter page with every tool we track, OSS flag, and a one-line description. Tools with multi-layer coverage appear in the full-stack group.
Warehouses, lakehouses, open table formats, analytics engines.
Metric definitions, dimensional models, agent-facing APIs.
Access control, data quality, lineage, audit.
Agent runtime. Planning, tool calls, memory, reasoning loops.
Interfaces that render agent output, traces, and actions.
Platforms spanning multiple layers with bundled NL-to-SQL.
§ Full catalog
The full catalog supports keyword search, layer filter, and license filter. Use it when you know what you need or want to compare across layers.