dagster-community-components-cli

CLI to search, install, and manage Dagster community components in your project.

$ dagster-component add s3_parquet_io_manager
✓ Found s3_parquet_io_manager in the registry
✓ Detected project at /Users/me/myproject
✓ Will install to: /Users/me/myproject/components/io_managers/s3_parquet_io_manager

Files to add (7):
  • component.py (3,412 B)
  • io_manager.py (4,109 B)
  • __init__.py (132 B)
  • README.md (2,201 B)
  • schema.json (1,847 B)
  • example.yaml (256 B)
  • requirements.txt (52 B)

Dependencies (3):
  • s3fs>=2023.1.0
  • pandas>=1.5.0
  • pyarrow>=12.0.0

Continue? [Y/n] y
✓ Wrote 7 files
✓ Dependencies installed

Next steps
  1. Open components/io_managers/s3_parquet_io_manager/example.yaml for a ready-to-use snippet.
  2. Read README.md for full configuration reference.
  3. Run dagster dev to load the new component.

Install

# Zero-install via uvx (recommended):
uvx --from dagster-community-components-cli dagster-component add postgres_resource

# Or pip install:
pip install dagster-community-components-cli
dagster-component add postgres_resource

Don't have uv? curl -LsSf https://astral.sh/uv/install.sh | sh (macOS / Linux) — see uv docs for Windows.

Commands

add <component_id>

Install a component into your project. Auto-detects your project root, picks a sensible install location, and installs the component's pip dependencies.

dagster-component add postgres_resource
dagster-component add s3_parquet_io_manager --target-dir ./resources/io
dagster-component add github_ingestion --auto-install         # don't prompt
dagster-component add great_expectations_check --no-install   # skip pip install
dagster-component add snowflake_resource --manager uv         # use uv even if pip is on PATH

search <query>

Search the registry by id, name, description, or tags.

dagster-component search snowflake
dagster-component search "vector store"
dagster-component search api --category resource

info <component_id>

Show full details for a component before installing.

dagster-component info one_hot_encoding

list

Show components installed in the current project.

dagster-component list
dagster-component list --available             # everything in the registry
dagster-component list --available --category transformation

remove <component_id>

Uninstall a component. Refuses to remove directories that aren't tagged with the install marker .dg-community.json.

dagster-component remove postgres_resource

update <component_id>

Re-fetch the component's files from the registry, overwriting in place.

dagster-component update postgres_resource

Where components get installed

By default, add installs to <project-root>/components/<category>/<id>/. Project root is auto-detected by walking upward from the current directory looking for dg.toml, pyproject.toml, or a defs/ folder.

Override with --target-dir if you have a different convention. Each install drops a .dg-community.json marker file so the CLI can later list / update / remove only its own installs without touching hand-written code.

Configuration

The manifest is cached at ~/.cache/dagster-community-components/manifest.json for one hour. Force a refresh with --refresh:

dagster-component --refresh search clickhouse

Registry

This CLI reads the manifest from:

https://raw.githubusercontent.com/eric-thomas-dagster/dagster-component-templates/main/manifest.json

Browse the components in the web UI:

https://dagster-component-ui.vercel.app/

Source repo for the components themselves:

https://github.com/eric-thomas-dagster/dagster-component-templates

Status

This is a community / prototype project — not officially supported by Dagster Labs. Components are best-effort and provided as templates to copy and adapt into your project. Issues and PRs welcome.

License

MIT