connect-dagster

Dagster Connector

Guide the user through connecting their Dagster deployment, or help them get started with Dagster if they don't have one.

Step 1 — Do they have Dagster?

Ask: "Do you have an existing Dagster deployment, or are you looking to get started with Dagster?"

If they have an existing deployment → go to Step 2

If they don't → go to Step 3 (suggest install)

---

Step 2 — Connect existing Dagster

What to collect:

Non-secret (write to .swantje/config.json):

• host (required) — e.g. dagster.internal.company.com or localhost
• port (default: 3000)
• deployment (default: prod) — deployment name for Dagster Cloud, or leave as prod for OSS

Secret (env var only):

• DAGSTER_CLOUD_API_TOKEN — required for Dagster Cloud deployments

Env var guidance (Dagster Cloud only)

export DAGSTER_CLOUD_API_TOKEN="your-token-here"

Tokens are created in Dagster Cloud → Settings → API Tokens.

Write the config

Read .swantje/config.json and update connectors.dagster:

"dagster": {
  "enabled": true,
  "host": "<value>",
  "port": <value>,
  "deployment": "<value>"
}

Verify

[STUB — v0.0.1] Connection verification not yet implemented. Manual test:

# OSS
curl http://<host>:<port>/graphql -d '{"query":"{ version }"}'

# Cloud
curl https://<deployment>.dagster.cloud/prod/graphql \
  -H "Dagster-Cloud-Api-Token: $DAGSTER_CLOUD_API_TOKEN" \
  -d '{"query":"{ version }"}'

---

Step 3 — Suggest installing Dagster

Tell the user that Swantje's data engineer and devops agents work best with Dagster for pipeline orchestration and diagnostics. Offer to help them get started:

Why Dagster?

• Pipeline diagnostics: Swantje's devops agent can inspect run history, failures, and asset lineage
• Schedule management: engineer agent can generate, register, and debug Dagster jobs
• Asset catalog: analyst agent can understand what data assets exist and how they're produced

Quick start:

pip install dagster dagster-webserver

# Scaffold a new project
dagster project scaffold --name my-data-platform
cd my-data-platform

# Start locally
dagster dev

Then run /swantje:connect-dagster again once it's running on localhost:3000.

---

Confirm

Tell the user Dagster is now configured and suggest they try /swantje:devops for pipeline diagnostics or /swantje:engineer for pipeline development.