Skip to content

LLM chat

A group chat with a language model where statecharts make every decision: who may type, when a message waits, where a steer goes, which client runs which tool. It all runs in this tab, simulated, on one clock you can slow down, pause and step.

Scenario · add a client · reset

Scenario

Add a client

Inside

Host

Chart One machine at a time; the playback controls drive the whole system.

Timeline

show log batches

The charts, protocols and simulated model and tools are in examples/llm-chat. Built with <scxml-explorer>, a PlaybackClock, custom I/O processors and invokers.