| name | dagster-expert |
| description | Expert guidance for working with Dagster and the dg CLI. Use before doing any task that requires knowledge specific to Dagster, or that references assets, materialization, components, data tools or data pipelines. Common tasks may include creating a new project, adding new definitions, understanding the current project structure, answering general questions about the codebase (finding asset, schedule, sensor, component or job definitions), debugging issues, or providing deep information about a specific Dagster concept. |
Core Dagster Concepts
Brief definitions only (see reference files for detailed examples):
- Asset: Persistent object (table, file, model) produced by your pipeline
- Component: Reusable building block that generates definitions (assets, schedules, sensors, jobs, etc.) relevant to a particular domain.
Integration Workflow
When integrating with ANY external tool or service, read the Integration libraries index. This contains information about which integration libraries exist, and references on how to create new custom integrations for tools that do not have a published library.
dg CLI
The dg CLI is the recommended way to programmatically interact with Dagster (adding definitions, launching runs, exploring project structure, etc.). It is installed as part of the dagster-dg-cli package. If a relevant CLI command for a given task exists, always attempt to use it.
ONLY explore the existing project structure if it is strictly necessary to accomplish the user's goal. In many cases, existing CLI tools will have sufficient understanding of the project structure, meaning listing and reading existing files is wasteful and unnecessary.
Almost all dg commands that return information have a --json flag that can be used to get the information in a machine-readable format. This should be preferred over the default table output unless you are directly showing the information to the user.
UV Compatibility
Projects typically use uv for dependency management, and it is recommended to use it for dg commands if possible:
uv run dg list defs
uv run dg launch --assets my_asset
CRITICAL: Always Read Reference Files Before Answering
NEVER answer from memory or guess at CLI commands, APIs, or syntax. ALWAYS read the relevant reference file(s) from the Reference Index below before responding.
For every question, identify which reference file(s) are relevant using the index descriptions, read them, then answer based on what you read.
Reference Index
- Asset Selection Syntax — filtering assets by tag, group, kind, upstream, or downstream; AssetSelection in Python, UI search bar, or CLI
- Environment Variables — configuring environment variables across different environments
- Asset Patterns — defining assets, dependencies, metadata, partitions, or multi-asset definitions
- Choosing an Automation Approach — deciding between schedules, sensors, and declarative automation
- Schedules — time-based automation with cron expressions
- Declarative Automation — asset-centric condition-based automation using AutomationCondition
- Asset Sensors — triggering on asset materialization events
- Basic Sensors — event-driven automation with file watching or custom polling
- Run Status Sensors — reacting to run success, failure, or other status changes
- dg check — validating project configuration or definitions
- create-dagster — creating a new Dagster project from scratch
- dg dev — starting a local Dagster development instance
- dg launch — materializing assets or executing jobs locally
- dg list components — seeing available component types for scaffolding
