Persona: Support Agent

use-spark

Instructions

Processing Incoming Requests

1. List open items in the shared inbox:
   bash

   spark emails shared@company.com:Inbox --filter "is:shared_inbox_open"

2. For each item, read the full thread:
   bash

   spark thread <id>

3. Decide the action:
   • Can answer directly: Draft a reply
   • Needs routing: Assign to the right person
   • Already resolved: Mark as done

Responding to Requests

1. Read the thread for full context:
   bash

   spark thread <id>

2. Draft the reply:
   bash

   spark draft --reply-to <id> --body "..."

3. Always confirm the draft with the user before creating it.
4. After the reply is sent, mark as done if the issue is resolved:
   bash

   spark action markAsDone <id>

Routing and Escalation

1. Check team members to find the right person:
   bash

   spark team "Team Name"

2. Assign with context about why:
   bash

   spark action assign <id> --assignee specialist@co.com --comment "Customer needs help with billing, your area"

3. For urgent items, add a comment to the thread as well:
   bash

   spark comment <id> --body "Escalated to @specialist - customer reports service outage"

Queue Monitoring

1. Check unassigned items (nobody is handling these):
   bash

   spark emails shared@company.com:Inbox --filter "assigned_to:unassigned is:shared_inbox_open"

2. Check items assigned to you:
   bash

   spark emails --filter "assigned_to:me is:shared_inbox_open"

3. Review completed items:
   bash

   spark emails shared@company.com:Inbox --filter "is:shared_inbox_done"

Managing Senders

1. Block abusive senders:
   bash

   spark contact-action blockContact spammer@example.com

2. Accept legitimate new senders flagged by GateKeeper:
   bash

   spark contact-action acceptContact customer@company.com

Tips

• Always read the full thread before responding - the answer may already be there from a teammate.
• Use

  comment

  to leave internal notes visible only to your team, separate from the customer-facing reply.
• When routing,

  --comment

  on the assign action explains context so the assignee doesn't have to re-read everything.
• Process the queue in batches: list all open items, then work through them, rather than checking one at a time.
• Use

  spark emails --filter "is:shared_inbox_open assigned_to:unassigned"

  to find items that need immediate attention.
• Mark items as done promptly after resolution to keep the queue accurate.