How does External Access for an Agent Work Internally in Pega? Does it use MCP for Database Access?

Hello Community,

I recently enabled an agent in Pega and also enabled External Access for that agent using the Agent Card configuration. While the configuration part is clear from the UI, I’m trying to better understand what happens internally within Pega when this is enabled.

Specifically, I’m interested in understanding:

  • What happens internally in Pega when External Access is enabled for an agent.

  • How external systems or agents actually invoke or interact with this Pega agent.

  • Whether there are any internal components involved such as queues, listeners, or background processing mechanisms.

  • Additionally, when the agent interacts with Pega’s internal database, does Pega use MCP (Multi-Channel Processing) or any similar mechanism for handling these interactions?

If anyone can explain the internal architecture or share relevant documentation or implementation details, it would be greatly appreciated.

Thank you for your help!

Hello Twinkle,

When external access is enabled the Agent discovery URL aka Agent Card URL is generated. which can be shared with the consumers.

External systems(including external pega gen ai agents) can integrate with pega agents using Google’s A2A protocol. Agent card URL is used as endpoint for Agents to collaborate with A2A compliant agents. A2A uses REST as a base protocol.

Pega Agents have access to Tools → DataPages which are executed by PegaRules. So typically Pega engine does JDBC to fetch data from DB which is available for Agent via Datapage tools.

1 Like