Executor locks up on large loops, requires full MCP server reinstall to recover

[BasilAtBentley]

Description

Running a large nested loop off of API calls causes the executor to become permanently unresponsive. This is inconsistent on a small model it
returns a clean timeout error, but on a larger model it jams permanently. Restarting Claude Desktop does not fix it. A full MCP server reinstall is required to recover.

Environment

• STAAD.Pro: 25.00.00.539
• OpenSTAAD MCP Server

Steps to Reproduce

The permanent lockup was reproduced on a larger model (over 2,400 nodes) which can be shared privately with the dev team on request.

geo = staad.Geometry
out = staad.Output

node_ids  = list(geo.GetNodeList())
plate_ids = list(geo.GetPlateList())

for nid in node_ids:
    for pid in plate_ids:
        try:
            r = out.GetPlateCornerForces(pid, nid, 1)
        except:
            pass

Expected Behaviour

Script always returns a clean timeout error regardless of model size.

Actual Behaviour

• Small model (176,400 calls): returns "Code execution timed out" recovers cleanly
• Large model (850,000 calls): hangs permanently full reinstall required

Impact

The Exception class issue recovery from a long-running script is unreliable. On a large model it leaves the executor in a broken state with no way to recover short of reinstalling the MCP server.

[BasilAtBentley]

Another thing that will lock up the executor is if STAAD has a warning displayed. Ability to recognize a warning message would be much smoother.