dbt Model Catalog¶
The Gnosis Analytics data platform is powered by dbt-cerebro, a dbt project that transforms raw Gnosis Chain data into approximately 1,200 curated models organized across 14 thematic modules. These models are materialized in a ClickHouse Cloud warehouse and serve as the backbone for both the Cerebro MCP server and the metrics dashboard. Models tagged for API exposure become REST endpoints automatically — see the Endpoint Catalog for the endpoint-side view of the same data.
Naming Convention¶
All models follow a structured naming pattern:
For example:
api_execution_transactions_daily-- API-layer, execution module, transaction data, daily granularityint_consensus_validators_status_daily-- Intermediate-layer, consensus module, validator status, daily granularitystg_execution__blocks-- Staging-layer, execution module, blocks (double underscore indicates direct source mapping)
Model Layers¶
Models are organized into four layers, each building on the previous one. Always prefer higher-tier models for query performance.
| Layer | Prefix | Purpose | Performance |
|---|---|---|---|
| API | api_* | Pre-aggregated daily/weekly views designed for fast reads | Fastest |
| Fact | fct_* | Immutable event records with business-level grain | Fast |
| Intermediate | int_* | Business logic joins, enrichment, deduplication | Moderate |
| Staging | stg_* | Minimal cleaning and type casting of raw source tables | Slowest (raw scans) |
Query Performance
When using Cerebro MCP or the dashboard API, always query api_* models first. Fall back to fct_* or int_* only when the API layer does not expose the fields you need. Avoid stg_* models for ad-hoc analytics unless you need raw-level granularity.
Module Overview¶
| Module | Models | Description |
|---|---|---|
| Bridges | ~17 | -- |
| Celo (Gnosis Pay) | ~48 | -- |
| Consensus | ~88 | -- |
| Contracts | ~102 | -- |
| Crawlers | ~16 | -- |
| ESG | ~16 | -- |
| Execution | ~722 | -- |
| Web Analytics (Mixpanel/GA) | ~59 | -- |
| MMM (Marketing Mix) | ~7 | -- |
| MTA (Attribution) | ~27 | -- |
| P2P Network | ~25 | -- |
| ProbeLab | ~5 | -- |
| Quarterly Data | ~27 | -- |
| Revenue | ~50 | -- |
Source Databases¶
Models are built from raw data stored across multiple ClickHouse databases:
| Database | Content |
|---|---|
execution | Blocks, transactions, logs, traces, contracts, transfers |
consensus | Validators, attestations, rewards, deposits, blobs, specs |
crawlers_data | Dune labels, Dune prices, bridge flows, GNO supply, Gnosis Pay wallets |
nebula | P2P network crawls and visit records |
nebula_discv4 | Discovery v4 protocol data (variant schema) |
dbt | Materialized dbt models (~1,200 tables) |
Using Models with Cerebro MCP¶
To explore models programmatically through the MCP server:
search_models("transactions daily") -- Find models by keyword
search_models(module="execution") -- Filter by module
get_model_details("api_execution_transactions_daily") -- Full schema + SQL
describe_table("api_execution_transactions_daily") -- Column types + descriptions
See the MCP Tools Reference for the complete tool catalog.