MCP streamable-HTTP transport: 'Session not found' from server permanently breaks tool calls (no auto-reconnect)

[iggerask]

Description

When a remote MCP server invalidates a streamable-HTTP session (server restart, idle timeout, periodic rotation, etc.), opencode keeps sending the same stale mcp-session-id. The server returns:

Streamable HTTP error: Error POSTing to endpoint: {"error":"Session not found"}

Once this happens, every subsequent tool call against that MCP server fails with the same error for the rest of the opencode session. The only fix is opencode mcp logout && opencode mcp auth and a process restart. There's no retry, no reconnect, no onclose handler.

This is in the same family as #17099 (no retry / reconnect for listTools() failures) but on a different code path: client.callTool() in convertMcpTool (packages/opencode/src/mcp/index.ts). #15209 reported essentially the same bug back in February but was auto-closed for not following the issue template — reopening here properly.

Steps to reproduce

1. Configure a remote streamable-HTTP MCP server.
2. Use it normally — agent calls a few tools successfully.
3. Restart that MCP server (or wait for any server-side session invalidation).
4. Have the agent call any tool on that server again.
5. → Streamable HTTP error: Error POSTing to endpoint: {"error":"Session not found"}. All subsequent calls to that server fail until opencode is restarted.

OpenCode version

1.14.30

Operating System

macOS 14 (Sequoia)

Expected

When the server reports the session is gone, opencode rebuilds the transport and replays the call once. If the retry also fails, surface the error normally.

[github-actions]

This issue might be a duplicate of existing issues. Please check:

• Bug: MCP tools permanently lost mid-session after single transient listTools() failure — no retry, no reconnect #17099: Same root cause — MCP tools permanently lost after a single transient failure with no retry or reconnect logic (same code path in mcp/index.ts, same symptom of permanent failure until restart)
• Remote MCP servers disconnected by server-side idle timeout lack automatic reconnection #15209: Same issue specifically for remote/streamable-HTTP transports with server-side session invalidation and no auto-reconnect (closed for template non-compliance, but covers the same bug)
• fix(mcp): MCP server loses connection mid-session causing intermittent failures #23997: Related — MCP server loses connection mid-session causing intermittent failures
• MCP server connection lost after context compaction — tools unavailable until new session #23556: Related — MCP connection lost after context compaction, tools unavailable until new session

Since you've already referenced #17099 and #15209 in your report, the maintainers may want to consolidate these under one of those issues or confirm this as a distinct sub-case (streamable-HTTP callTool() path vs. listTools() path in #17099).