Quickstart: your first workflow

The fastest useful first workflow is read-only: it gathers the context an engineer would gather and attaches it to a ticket. It cannot break anything, and it saves time from day one.

1. Connect a context source

Start with monitoring, because it answers the questions a first responder asks. Add a connector for your monitoring system and supply read-only credentials. StackHeal stores them in the secret store; they are resolved at execution time and never appear in a workflow definition.

2. Connect your service desk

Add the connector for your ticketing system. You need two capabilities: receiving an event when a ticket is created or tagged, and posting an internal note back.

3. Register the actions you need

For a read-only first workflow, three actions suffice: query current metrics for an asset, look up the asset for an account, and post a note to a ticket. Each action declares its input schema and the credential it resolves.

4. Build the workflow

1. Trigger: ticket created with a tag you control, so you can opt tickets in deliberately.
2. Step: resolve the asset from the ticket's account and any identifier in the body.
3. Condition: if no unique asset matches, post a note saying so and stop.
4. Step: query current metrics for that asset.
5. Step: post an internal note with the asset, the metrics and the timestamp.

5. Run it against one asset

Trigger a manual run with a known ticket, confirm the note content is what you would have wanted as an engineer, then enable the tag-based trigger for one team.

6. Then add an action

Once the context is reliably right, add one state-changing action — scoped to a single asset class, requiring verification, with a blast-radius limit of one asset per run. Widen from there.