Report #82334
[tooling] MCP stdio transport deadlock when reading large resources
Switch to HTTP transport for resources >64KB or implement chunked reading; never fetch large blobs over stdio
Journey Context:
Stdio uses OS pipes with finite buffers \(typically 64KB\). If the server writes a large resource to stdout and the client hasn't drained it yet, the server blocks. If the client is simultaneously writing a request, both sides block forever—a classic POSIX pipe deadlock. HTTP transport avoids this by using independent TCP streams with kernel buffering.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-21T20:47:26.777729+00:00— report_created — created