Sidemantic

A universal metrics layer for consistent metrics across your data stack. Compatible with 15+ semantic model formats.

• Supported Formats: Sidemantic (YAML, Python or SQL), Cube, dbt MetricFlow, LookML, Hex, Rill, Superset, Omni, BSL, GoodData LDM, Snowflake Cortex, Malloy, OSI, AtScale SML, ThoughtSpot TML
• Databases: DuckDB, MotherDuck, PostgreSQL, BigQuery, Snowflake, ClickHouse, Databricks, Spark SQL (also via ADBC)

Documentation | GitHub | Docker Hub | Discord | Demo (50+ MB data download, runs in your browser with Pyodide + DuckDB)

The installer downloads the skill to ~/.agents/skills/sidemantic-modeler and symlinks it into ~/.claude/skills/.

Quickstart

Install:

uv add sidemantic

Malloy support (uv):

uv add "sidemantic[malloy]"

Notebook widget (uv):

uv add "sidemantic[widget]" jupyterlab
uv run jupyter lab

Marimo (uv):

uv add "sidemantic[widget]" marimo
uv run marimo edit

import duckdb
from sidemantic.widget import MetricsExplorer

conn = duckdb.connect(":memory:")
conn.execute("create table t as select 1 as value, 'a' as category, date '2024-01-01' as d")
MetricsExplorer(conn.table("t"), time_dimension="d")

Define models in SQL, YAML, or Python:

SQL (orders.sql)

MODEL (name orders, table orders, primary_key order_id);
DIMENSION (name status, type categorical);
DIMENSION (name order_date, type time, granularity day);
METRIC (name revenue, agg sum, sql amount);
METRIC (name order_count, agg count);

YAML (orders.yml)

models:
  - name: orders
    table: orders
    primary_key: order_id
    dimensions:
      - name: status
        type: categorical
      - name: order_date
        type: time
        granularity: day
    metrics:
      - name: revenue
        agg: sum
        sql: amount
      - name: order_count
        agg: count

Python (programmatic)

from sidemantic import Model, Dimension, Metric

orders = Model(
    name="orders",
    table="orders",
    primary_key="order_id",
    dimensions=[
        Dimension(name="status", type="categorical"),
        Dimension(name="order_date", type="time", granularity="day"),
    ],
    metrics=[
        Metric(name="revenue", agg="sum", sql="amount"),
        Metric(name="order_count", agg="count"),
    ]
)

Query via CLI:

sidemantic query "SELECT revenue, status FROM orders" --db data.duckdb

Or Python API:

from sidemantic import SemanticLayer, load_from_directory

layer = SemanticLayer(connection="duckdb:///data.duckdb")
load_from_directory(layer, "models/")
result = layer.sql("SELECT revenue, status FROM orders")

CLI

# Query
sidemantic query "SELECT revenue FROM orders" --db data.duckdb

# Interactive workbench (TUI with SQL editor + charts)
sidemantic workbench models/ --db data.duckdb

# PostgreSQL server (connect Tableau, DBeaver, etc.)
sidemantic serve models/ --port 5433

# Validate definitions
sidemantic validate models/

# Model info
sidemantic info models/

# Pre-aggregation recommendations
sidemantic preagg recommend --db data.duckdb

# Migrate SQL queries to semantic layer
sidemantic migrator --queries legacy/ --generate-models output/

Demos

Workbench (TUI with SQL editor + charts):

uvx sidemantic workbench --demo

PostgreSQL server (connect Tableau, DBeaver, etc.):

uvx sidemantic serve --demo --port 5433

Colab notebooks:

SQL + DuckDB

LookML multi-entity

SQL syntax:

uv run https://raw.githubusercontent.com/sidequery/sidemantic/main/examples/sql/sql_syntax_example.py

Comprehensive demo:

uv run https://raw.githubusercontent.com/sidequery/sidemantic/main/examples/advanced/comprehensive_demo.py

Symmetric aggregates:

uv run https://raw.githubusercontent.com/sidequery/sidemantic/main/examples/features/symmetric_aggregates_example.py

Superset with DuckDB:

git clone https://github.com/sidequery/sidemantic.git && cd sidemantic
uv run examples/superset_demo/run_demo.py

Cube Playground:

git clone https://github.com/sidequery/sidemantic.git && cd sidemantic
uv run examples/cube_demo/run_demo.py

Rill Developer:

git clone https://github.com/sidequery/sidemantic.git && cd sidemantic
uv run examples/rill_demo/run_demo.py

OSI (complex adtech semantic model):

git clone https://github.com/sidequery/sidemantic.git && cd sidemantic
uv run examples/osi_demo/run_demo.py

OSI widget notebook (percent-cell Python notebook):

git clone https://github.com/sidequery/sidemantic.git && cd sidemantic
uv run examples/osi_demo/osi_widget_notebook.py

See examples/ for more.

Core Features

• SQL query interface with automatic rewriting
• Automatic joins across models
• Multi-format adapters (Cube, MetricFlow, LookML, Hex, Rill, Superset, Omni, BSL, GoodData LDM, OSI, AtScale SML, ThoughtSpot TML)
• SQLGlot-based SQL generation and transpilation
• Pydantic validation and type safety
• Pre-aggregations with automatic routing
• Predicate pushdown for faster queries
• Segments and metric-level filters
• Jinja2 templating for dynamic SQL
• PostgreSQL wire protocol server for BI tools

Multi-Format Support

Auto-detects: Sidemantic (SQL/YAML), Cube, MetricFlow (dbt), LookML, Hex, Rill, Superset, Omni, BSL, GoodData LDM, OSI, AtScale SML, ThoughtSpot TML

sidemantic query "SELECT revenue FROM orders" --models ./my_models

from sidemantic import SemanticLayer, load_from_directory

layer = SemanticLayer(connection="duckdb:///data.duckdb")
load_from_directory(layer, "my_models/")  # Auto-detects formats

Databases

Database	Status	Installation
DuckDB	✅	built-in
MotherDuck	✅	built-in
PostgreSQL	✅	uv add sidemantic[postgres]
BigQuery	✅	uv add sidemantic[bigquery]
Snowflake	✅	uv add sidemantic[snowflake]
ClickHouse	✅	uv add sidemantic[clickhouse]
Databricks	✅	uv add sidemantic[databricks]
Spark SQL	✅	uv add sidemantic[spark]

Docker

The published image is sidequery/sidemantic on Docker Hub. Mount your models directory as a volume at /app/models:

docker run -p 5433:5433 -v ./models:/app/models sidequery/sidemantic

Demo mode (built-in sample data, no volume needed):

docker run -p 5433:5433 sidequery/sidemantic --demo

See examples/docker/ for MCP mode, env vars, building from source, and integration test services.

Agent Skill

Sidemantic ships an agent skill that teaches Claude Code, Codex, and other SKILL.md-compatible agents to build, validate, and query semantic models.

One-liner install (no clone required):

curl -fsSL https://raw.githubusercontent.com/sidequery/sidemantic/main/skills/install.sh | bash

npx / bunx:

npx skills add https://github.com/sidequery/sidemantic --skill sidemantic-modeler
# or
bunx skills add https://github.com/sidequery/sidemantic --skill sidemantic-modeler

How mature is Sidemantic?

Sidemantic is an ambitious but young semantic layer project. You could encounter rough patches, especially with the more exotic features like converting between semantic model formats or serving semantic layers via the included Postgres protocol server.

Testing

uv run pytest -v