System Requirements¶

Python¶

KelpMesh requires Python 3.11 or later.

Python	Status
3.13	✅ Supported
3.12	✅ Supported
3.11	✅ Supported
3.10	❌ Not supported (no `match/case` with guards)
< 3.10	❌ Not supported

Operating System¶

OS	Status
Windows 11 / Windows Server 2022+	✅ Supported
Ubuntu 20.04 LTS / Debian 11+	✅ Supported
macOS 12 Monterey+	✅ Supported
Other Linux (RHEL 8+, Alpine)	✅ Generally works

Hardware¶

Component	Minimum	Recommended
RAM	512 MB	2 GB+
Disk	200 MB (core + DuckDB)	2 GB+ (for large state DBs)
CPU	Any	2+ cores (for `--threads 4`)

Core Dependencies (auto-installed)¶

KelpMesh has intentionally minimal dependencies — all installed automatically via pip:

typer>=0.12       CLI framework
sqlglot>=25.0     SQL parsing, linting, transpilation
networkx>=3.0     DAG / dependency graph
duckdb>=1.0       State engine + local warehouse
pydantic>=2.0     Config validation
rich>=13.0        Terminal output
jinja2>=3.0       Template compatibility layer
pyyaml>=6.0       Config files
sqlparse>=0.5     SQL formatting
cryptography>=42  State encryption

Warehouse Driver Extras¶

Install only what you need:

# Local / DuckDB (no extra needed — built-in)
pip install KelpMesh

# PostgreSQL / Amazon Redshift
pip install "KelpMesh[postgres]"

# Snowflake
pip install "KelpMesh[snowflake]"

# Google BigQuery
pip install "KelpMesh[bigquery]"

# Databricks
pip install "KelpMesh[databricks]"

# Microsoft Fabric (requires pyodbc + Azure SDK)
pip install "KelpMesh[fabric]"

# MySQL / MariaDB
pip install "KelpMesh[mysql]"

# ClickHouse
pip install "KelpMesh[clickhouse]"

# Trino / Presto
pip install "KelpMesh[trino]"

# Apache Spark
pip install "KelpMesh[spark]"

# Amazon Athena
pip install "KelpMesh[athena]"

# Apache Hive
pip install "KelpMesh[hive]"

# SQL Server / Azure Synapse  (also requires OS-level ODBC driver)
pip install "KelpMesh[sqlserver]"

# All major warehouses at once
pip install "KelpMesh[all-warehouses]"

OS-Level Driver Requirements¶

Some adapters require an OS-level driver beyond the Python package:

SQL Server / Azure Synapse — ODBC Driver 18

# Windows: download from https://aka.ms/odbc18
# Ubuntu/Debian:
curl https://packages.microsoft.com/keys/microsoft.asc | sudo apt-key add -
sudo apt install msodbcsql18
# macOS:
brew install msodbcsql18

Microsoft Fabric — pyodbc

# Same ODBC driver as SQL Server above

Apache Hive — SASL/Kerberos (optional)

# Ubuntu/Debian (for Kerberos-secured HiveServer2):
sudo apt install libsasl2-dev libkrb5-dev
pip install "KelpMesh[hive]"

Optional Integrations¶

Integration	Package	Min Version
Apache Airflow	`apache-airflow`	2.8
Dagster	`dagster dagster-webserver`	1.6
Prefect	`prefect`	3.0
MkDocs (for docs site)	`mkdocs mkdocs-material`	1.6 / 9.5

Install a specific orchestrator:

pip install "KelpMesh[dagster]"
pip install "KelpMesh[airflow]"
pip install "KelpMesh[prefect]"

kelpmesh-studio Additional Requirements¶

KelpMesh Studio adds a web UI on top of the core:

pip install kelpmesh-studio

Studio requires: - All requirements above - A browser (Chrome 120+, Firefox 122+, Safari 17+, Edge 120+) - Port 8501 available (configurable)

Network Requirements¶

KelpMesh itself makes zero outbound network calls. All communication is with your own warehouse.

KelpMesh Studio telemetry is also disabled by default and can be verified at startup with kelpmesh debug.

Verify Installation¶

# Check Python version
python --version

# Verify kelpmesh installed correctly
kelpmesh --version

# Check connection to your warehouse
kelpmesh debug

# Run self-test (no warehouse needed — uses in-memory DuckDB)
kelpmesh run --project-dir examples/quickstart