Airframe

Generate production-ready Airflow DAGs from battle-tested templates.

Airframe is a CLI tool that generates complete, deployable Airflow DAGs from Jinja2 templates — not just code skeletons, but the full working system with connections, retry policies, observability hooks, and optional Terraform infrastructure for self-hosted Airflow on Linode.

Quick Start

pip install xtremedp-airframe

# Initialize a project
airframe init

# Generate a DAG from a template
airframe generate postgres-cdc --dag-id orders_cdc --table orders --schedule "@hourly"

# List available templates
airframe list-templates

# Deploy Airflow infrastructure (requires Terraform + Linode token)
airframe deploy

# Check version and environment
airframe version
airframe info

Commands

airframe init

Scaffolds a new airframe project in the current directory:

• dags/ — generated DAG files
• tests/ — DAG tests
• airframe.yaml — project configuration
• .gitignore

airframe generate <template>

Generates a DAG file from a template. Reads defaults from airframe.yaml, overrides with CLI flags:

airframe generate postgres-cdc \
  --dag-id orders_cdc \
  --table orders \
  --schedule "@hourly" \
  --batch-size 2000 \
  --source-conn source_postgres \
  --warehouse-conn warehouse_conn

airframe list-templates

Shows available templates with pricing.

airframe deploy

Runs Terraform to deploy self-hosted Airflow on Linode. Requires Terraform installed and Linode API token configured.

airframe version

Prints the installed airframe version.

airframe info

Shows version, Python version, project config (if airframe.yaml exists), and available templates with pricing.

Examples

See the examples/ directory for generated DAGs from every template:

airframe generate data-quality --dag-id dq_validation
airframe generate postgres-cdc --dag-id orders_cdc --table orders
airframe generate api-to-warehouse --dag-id orders_api_sync --endpoint /api/v1/orders

Templates

Configuration

airframe.yaml stores project defaults:

project_name: my-pipelines
owner: airframe
schedule: "@daily"
start_date: "2025-01-01"
catchup: false
retries: 1
retry_delay_minutes: 5
tags:
  - airframe
output_dir: dags

Development

git clone <repo-url>
cd airframe
uv pip install -e ".[dev]"
pytest

Architecture

src/airframe/
├── cli/main.py          — Typer CLI (init, generate, list-templates, deploy)
├── config.py            — ProjectConfig (Pydantic, airframe.yaml schema)
├── templates/
│   ├── __init__.py      — TemplateConfig, render_template(), template registry
│   ├── data_quality.py.j2 — Standalone DQ validation template (Free)
│   ├── file_batch_ingest.py.j2  — S3 batch ingest template ($29)
│   ├── postgres_cdc.py.j2  — Postgres CDC DAG template ($49)
│   ├── api_to_warehouse.py.j2  — REST API → warehouse template ($99)
│   ├── multi_source_elt.py.j2  — Multi-source ELT template ($149)
│   └── dag_test.py.j2   — Generated DAG test template
terraform/airflow/
├── main.tf              — Linode instance, firewall, DNS
├── variables.tf         — Configurable inputs
├── outputs.tf           — Airflow URL, IP, SSH command
└── install.sh           — StackScript provisioning

License

The CLI and free data-quality template are MIT-licensed. Paid template packs are sold separately under a commercial license — see MARKETING.md for purchasing.