Skip to main content
The MCP server turns your AI tool into an OpenCX operator. Here are three workflows it handles well.

​
Debug a Ticket From Your IDE

You’re triaging a ticket that a customer escalated. Instead of switching to the dashboard, pull the full investigation into your editor.
1

Look up the ticket

Ask your AI assistant:“Investigate ticket #4521 — why did the AI hand off?”
2

Read the AI's reasoning

Ask:“Show me the debug data for that session, including the AI’s reasoning steps and citations.”
3

Add an internal note

Ask:“Add a comment on the last message: ‘Checked — handoff was correct, agent lacked refund policy training.’”
What the MCP does under the hood:
  • investigate_ticket fetches the session summary, history, handoff details, tags, and diagnostics in one call.
  • investigate_session_debug returns the full debug JSON with per-message AI reasoning and source citations.
  • add_session_comment posts an internal comment visible only to your team.

​
Bulk-Train Your Agent From Unresolved Sessions

Your AI agent handed off 30 sessions last week. You want to find patterns and create training scenarios so it handles those topics next time.
1

Find unresolved sessions

Ask:“List sessions from the last 7 days that were closed as unresolved.”
2

Analyze handoff reasons

Ask:“Show me handoff analytics for the last 7 days — what are the top reasons?”
3

Create training drafts

Ask:“Create a draft training scenario titled ‘Subscription cancellation flow’ that instructs the AI to walk the customer through cancellation steps and offer a retention discount before confirming.”
4

Review and publish

Review the draft in your OpenCX dashboard, then ask:“Publish the training scenario ‘Subscription cancellation flow’.”
What the MCP does under the hood:
  • filter_sessions finds sessions by status and date range.
  • get_handoff_analytics returns reason breakdowns, sentiment, and referenced resources.
  • create_training_scenario saves a new scenario as a draft for review.
  • publish_training_scenario makes it active for the AI agent.

​
Pull Customer Insights for a Sprint Review

Your product team meets weekly. You need a summary of what customers asked about most, grouped by theme and ready to share.
1

List insight categories

Ask:“What customer insight categories do we have?”
2

Fetch recent insights

Ask:“List customer insights — show me the most recent ones with their categories and session counts.”
3

Drill into a theme

Ask:“Show me the details for the top insight — which sessions does it reference?”
4

Assign follow-ups

Ask:“Assign the billing-related insight to the Billing team.”
What the MCP does under the hood:
  • list_insight_categories returns available category groupings.
  • list_insights retrieves insights sorted by recency with pagination.
  • get_insight shows full details including related sessions and resolution status.
  • assign_insight routes an insight to a specific team for follow-up.

Tools & Capabilities

Full reference of every available tool

Installation

Set up the MCP server in your AI tool

Authentication

API key configuration and security

Troubleshooting

Common issues and fixes