Actors are the software components or Agents, that compute Resources. Actors are running units with their own lifecycle.
Actors can be GUIs, Tools, Middleware, indexers, Agents?
Another option is Application Code , but IMO this is limiting.
Developers should be able to design their actors with extreme freedom, but they need to follow the protocol's resource schema. However, what baseline must all Actors follow?
An actor is always invoked uniformly, in exactly the same way, whether it behaves as a recursive function, data structure, or process.
Oh, we already have some definitions:
actor: who did something, recorded on every event in the log. The declared values are user, agent, system, and trigger. Only user and system are ever stamped on an event; the agent's own messages and tool calls derive agent from their shape, and trigger is reserved and never written, so a message a trigger started reads as user. Calls a person runs from the wrench palette carry user.
An actor is a running unit with its own lifecycle. A machine describes its states and transitions. A job is serializable data describing work, not a running actor. Keeping jobs separate from actors lets an app restart with fresh actors rather than attempting to resume an old network promise.
Do you like what you are reading? Subscribe to receive updates.
Unsubscribe anytime