Skip to content

ADR 003: Integration of Model Context Protocol (MCP) and Agent-to-Agent (A2A) Standards

  • Status: Accepted
  • Date: 2026-08-06
  • Deciders: Software Architecture Team

Context and Problem Statement

To make the AI Operator extensible and product-ready, the platform needs a standardized way to integrate external tools, data sources, and inter-agent communication across different clusters or services without writing custom skill code for every integration.

Decision Drivers

  • Extensibility: Support standard plug-and-play tools (PostgreSQL, GitHub, Slack, local files).
  • Interoperability: Allow agents running in different Kubernetes clusters or microservices to communicate using open standards.
  • Security: Enforce strict policy controls and sandboxing on external tools.

Decision Outcome

Chosen Option: Adopt Model Context Protocol (MCP) for external tool/data server integration and A2A Protocol Adapter for inter-agent communication.

Implementation Details

  1. MCP Server Integration:
  2. Platform implements an MCP Client registry.
  3. Agents discover and invoke tools exported by standard MCP servers over SSE or Stdio.
  4. A2A Protocol Adapter:
  5. Expose standardized gRPC/REST endpoints for agent discovery, handoff contracts, and remote execution.

Positives

  • Ecosystem Compatibility: Instantly access hundreds of community-built MCP servers.
  • Cluster-to-Cluster Handoff: Agents can delegate sub-tasks to remote agents in other clusters using A2A.

Negatives

  • Network latency overhead for remote MCP/A2A calls must be mitigated with connection pooling and timeouts.