Which client?
All / shared core
The problem
There's no way to point the Inspector at a server and generate concurrent/sustained request load to see how it behaves under stress (connection limits, backpressure, timeout tuning). Today that means writing a separate script outside the Inspector entirely, which loses all the connection/auth/transport setup the Inspector already has.
This is more speculative and architectural than a typical parity fix, so flagging it as a discussion rather than a fully-scoped request -- it sits outside the current roadmap's frame.
Solution you have in mind (optional)
A CLI mode that fires N concurrent tool calls (or a configurable rate) for a set duration, reporting a latency/error-rate summary at the end.
Alternatives or workarounds you have tried
No response
Already built it locally?
No response
Before you submit
Which client?
All / shared core
The problem
There's no way to point the Inspector at a server and generate concurrent/sustained request load to see how it behaves under stress (connection limits, backpressure, timeout tuning). Today that means writing a separate script outside the Inspector entirely, which loses all the connection/auth/transport setup the Inspector already has.
This is more speculative and architectural than a typical parity fix, so flagging it as a discussion rather than a fully-scoped request -- it sits outside the current roadmap's frame.
Solution you have in mind (optional)
A CLI mode that fires N concurrent tool calls (or a configurable rate) for a set duration, reporting a latency/error-rate summary at the end.
Alternatives or workarounds you have tried
No response
Already built it locally?
No response
Before you submit