We're heading to KubeCon Mumbai·June 18–19

Let's meet

Thread

A
Abhinav9:30 AMOpen in Slack
Hi team, do we have MCP sampling on the roadmap anywhere? seems like the one MCP piece that's not implemented yet
feels like a natural fit since the LLM proxy's already there. those completions could run with cost tracking + guardrails, rather than a server hitting a model directly with no oversight.
curious if this is something you'd find valuable.. glad to pick it up if there's interest

2 replies
II
Ildar Iskhakov (archestra team)9:31 AMOpen in Slack
we are not interested in implementing it right now
A
Abhinav11:41 AMOpen in Slack
ah good to know, thanks! makes sense not to chase a deprecated spec.