Story (Required)

As a Platform Engineer trying to integrate AI capabilities with RHDH I want to create a proof of concept for MCP (Model Context Protocol) integration with Backstage's eventing and notification system

This spike explores how AI clients can safely send and receive events within Red Hat Developer Hub (RHDH) while maintaining data privacy and security. The goal is to understand the technical feasibility and security implications of connecting MCP-based AI systems with Backstage's existing notification infrastructure to enable real-time AI-driven insights and updates for developers.

Background (Required)

Red Hat Developer Hub (RHDH) currently has a robust eventing and notification system that supports both broadcast and entity-specific notifications. The system includes a notifications backend plugin, signals for real-time push notifications, and subscription mechanisms. We need to explore how to integrate Model Context Protocol (MCP) systems with this existing infrastructure to allow AI clients to participate in the event ecosystem while ensuring sensitive data (PII/CII) is properly handled. This integration would enable AI-powered features like automated incident detection, deployment notifications, and proactive alerts.

Out of scope

Implementation of notification/eventing systems for MCP tools delivered in RHDH 1.8.

Approach (Required)

1. Analyze Backstage's current notification system architecture (@backstage/plugin-notifications-backend, @backstage/plugin-signals-backend)
2. Determine potential MCP integration points with existing eventing infrastructure
3. If possible, develop proof of concepts that can integrate with the eventing infrastructure (possible suggestions below):

• Integration points for MCP tool plugin: (MCP tools that use the eventing infrastructure, or creating events when certain actions are performed?) (https://github.com/redhat-ai-dev/community-plugins/tree/main/workspaces/mcp-integrations)
• Integration points for the upstream MCP server plugin? (https://github.com/backstage/backstage/tree/master/plugins/mcp-actions-backend)

Dependencies

Access to RHDH development environment, understanding of Backstage notification system architecture, MCP protocol specification, security review guidelines for PII/CII handling

Acceptance Criteria (Required)

_1. Successfully demonstrate MCP client subscribing to filtered Backstage events
2. Prove AI client can send notifications through existing Backstage notification channels
3. Verify PII/CII data is filtered before reaching AI systems
4. Document integration points and security considerations
5. Create technical feasibility report with recommendations for production implementation
6. Edge cases to consider:

• Event filtering failure scenarios
• MCP client connection loss and reconnection
• Rate limiting for AI-generated notifications
• Audit logging of all AI interactions with event system
• Handling of malformed MCP messages_

documentation updates (design docs, release notes etc)
SOP required
education module update (Filled by RHDHPAI team only)
(✓) R&D label required (Filled by RHDHPAI team only)

Done Checklist

Code is completed, reviewed, documented and checked in
Unit and integration test automation have been delivered and running cleanly in continuous integration/staging/canary environment
Continuous Delivery pipeline(s) is able to proceed with new code included
Customer facing documentation, API docs, design docs etc. are produced/updated, reviewed and published
Acceptance criteria are met
If the Grafana dashboard is updated, ensure the corresponding SOP is updated as well