Brief account owners when incidents hit their customers

Build me an agent workflow that fans out customer-impacting incidents from incident.io into HubSpot and Slack so account owners and CS leads are never the last to know.

Trigger: an incident.io outgoing webhook on the incident_updated event. Configure the trigger so it accepts the webhook payload incident.io sends when an incident is declared or escalated.

When the webhook fires, the agent should:

1. Pull the incident id out of the webhook payload and call incident.io Show Incident to get the full record. From it, read the severity, current status, name and summary, whether it is marked internal, and any custom field listing affected customers or accounts. If the incident is internal-only, or the severity is below a configurable threshold (default: major and above), stop and do nothing.

2. For each affected customer name on the incident, call HubSpot Search Companies to find the matching company record. Match on company name, and fall back to domain if the custom field includes one. Keep track of any names that do not resolve to a HubSpot company so they can be flagged in the digest.

3. For each matched HubSpot company, call HubSpot Create Note and attach it to that company. The note should describe the current incident status in plain English, including incident name, severity, current status, a short summary, and a link back to the incident in incident.io.

4. For each matched HubSpot company, call HubSpot Create Task assigned to the company's account owner. The task subject should make it obvious this is a proactive outreach for an active incident, with a due date of today and high priority. If a company has no owner, do not create the task and instead flag the company in the digest so it can be routed manually.

5. Once every affected customer has been processed, call Slack Send a Message to post a single digest in a configurable channel (default: a shared support and CS room). The message should list each impacted account with what was logged (note created, task created and assigned to owner X, or skipped because no HubSpot match / no owner), plus the incident link so anyone reading can jump straight to incident.io.

Settings the user should be able to change without editing the workflow: minimum severity threshold, the Slack channel that receives the digest, and the wording or due date used for the HubSpot follow-up task.

Guardrails: never invent or create new HubSpot companies, never duplicate notes or tasks within the same incident update, and skip silently when the incident is internal-only or below the severity threshold.