iMCP crashes repeatedly during normal use. The crash is not a user-triggered quit or an OS-level termination — it's an internal Swift concurrency error triggered when Claude Desktop reconnects to the MCP server.
What happens:
iMCP starts, runs briefly, then disappears from the menu bar. Inspecting system logs reveals the crash occurs during a burst of incoming TCP connections from Claude Desktop. iMCP opens 20–30 connections in rapid succession, then begins cancelling earlier ones while still trying to set handlers on them — resulting in a fatal Swift error:
SWIFT TASK CONTINUATION MISUSE: connect() tried to resume its continuation
more than once, throwing [-32603] Internal error: Connection cancelled!
Preceded by repeated warnings:
nw_connection_set_queue_block_invoke [Cx] Error in client:
nw_connection_set_queue called after nw_connection_start
And:
SWIFT TASK CONTINUATION MISUSE: connect() leaked its continuation
without resuming it. This may cause tasks waiting on it to remain
suspended forever.
Pattern:
The crash happens consistently when Claude Desktop (re)connects — likely on startup or after iMCP has been idle. The app then auto-relaunches (via Login Items) and the cycle repeats.
Environment:
- iMCP version: 1.4.0
- macOS: Sequoia 15.x
- Mac Mini M4
- Claude Desktop connected as MCP client
Reproduction:
Launch iMCP, open Claude Desktop (or reconnect it). Observe iMCP disappearing from the menu bar within seconds to minutes.
Relevant log excerpt (anonymized):
iMCP[XXXX] [co.dododo.iMCP:server] Starting MCP server for connection: <private>
iMCP[XXXX] [co.dododo.iMCP:server] Connection <private> successfully established
...
[repeated ~25 times for connections C1–C27]
...
iMCP[XXXX] nw_connection_set_queue_block_invoke [C1] Error in client:
nw_connection_set_queue called after nw_connection_start
iMCP[XXXX] (libswiftCore.dylib) SWIFT TASK CONTINUATION MISUSE:
connect() leaked its continuation without resuming it.
...
iMCP[XXXX] (libswiftCore.dylib) Fatal error: SWIFT TASK CONTINUATION MISUSE:
connect() tried to resume its continuation more than once,
throwing [-32603] Internal error: Connection cancelled!
The app exits immediately after the fatal error, with no crash report generated.
iMCP crashes repeatedly during normal use. The crash is not a user-triggered quit or an OS-level termination — it's an internal Swift concurrency error triggered when Claude Desktop reconnects to the MCP server.
What happens:
iMCP starts, runs briefly, then disappears from the menu bar. Inspecting system logs reveals the crash occurs during a burst of incoming TCP connections from Claude Desktop. iMCP opens 20–30 connections in rapid succession, then begins cancelling earlier ones while still trying to set handlers on them — resulting in a fatal Swift error:
Preceded by repeated warnings:
And:
Pattern:
The crash happens consistently when Claude Desktop (re)connects — likely on startup or after iMCP has been idle. The app then auto-relaunches (via Login Items) and the cycle repeats.
Environment:
Reproduction:
Launch iMCP, open Claude Desktop (or reconnect it). Observe iMCP disappearing from the menu bar within seconds to minutes.
Relevant log excerpt (anonymized):
The app exits immediately after the fatal error, with no crash report generated.