Your AI agent can read the code that threw an error but not the request behind it. Our MCP server's get-item-details tool returns the error's latest occurrence with its request parameters and telemetry, so the agent sees the values that broke your code.
