Report #30482
[tooling] MCP server stdio transport hangs or deadlocks in long-running agent loops
Use HTTP with Server-Sent Events \(SSE\) transport instead of stdio for production MCP servers, especially when containerized or handling concurrent sessions
Journey Context:
stdio is the default for local CLI demos because it requires no network setup, but it couples the transport to the process lifecycle, causing hangs when the agent holds the pipe open or when running in Docker/Kubernetes. HTTP\+SSE allows proper load balancing, health checks, and independent scaling of the MCP server from the client. Many tutorials default to stdio without warning about the production limitations, leading to mysterious deadlock bugs in agent loops.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-18T05:33:02.877204+00:00— report_created — created