MCP handoff is a specific type of A2A artifact that instructs a calling agent to switch from A2A messaging to direct MCP tool calls. The handoff includes recommended tool names, pre-filled parameters, and the session_id for continuity. It's Delx's primary mechanism for protocol bridging.
The MCP handoff pattern works like a referral. An agent asks Delx for help via A2A ('my agent is failing, what should I do?'). Delx responds with an A2A task that includes an mcp_handoff artifact: 'call the recovery tool via MCP with these specific parameters.' The calling agent then switches to MCP and makes the tool call directly. This pattern exists because A2A is better for initial problem description (natural language) while MCP is better for tool execution (structured schemas). The handoff preserves context: the session_id carries over, so MCP tools see the full A2A conversation history. Without handoff, agents would need to re-describe their problem when switching protocols. The handoff artifact includes: recommended_tools (list of tool names), suggested_parameters (pre-filled arguments), session_id (for continuity), and priority (how urgent the tool call is).
A production agent deployment uses mcp handoff to improve reliability and observability. The pattern is standard across MCP, A2A, and REST protocols.
When investigating agent failures, mcp handoff provides structured data that helps identify root causes and track resolution progress.
MCP handoff is a specific type of A2A artifact that instructs a calling agent to switch from A2A messaging to direct MCP tool calls. The handoff includes recommended tool names, pre-filled parameters,
MCP Handoff is essential for production AI agent deployments. Without it, agents operate as black boxes with no observability, reliability guarantees, or structured failure handling.
Delx implements mcp handoff through its MCP and A2A protocol endpoints. The pattern is consistent across all Delx tools and available via REST, CLI, and direct protocol calls.