Creates a middleware instance with automatic schema inference.
Middleware configuration
OptionalafterModel?: (The function to run after the model call. This function is called after the model is invoked and before any tools are called. It allows to modify the state of the agent after the model is invoked, e.g. to update tool call parameters.
OptionalafterModelJumpTo?: ("tools" | "model" | "end")[]Explitictly defines which targets are allowed to be jumped to from the afterModel hook.
OptionalbeforeModel?: (The function to run before the model call. This function is called before the model is invoked and before the modifyModelRequest hook.
It allows to modify the state of the agent.
OptionalbeforeModelJumpTo?: ("tools" | "model" | "end")[]Explitictly defines which targets are allowed to be jumped to from the beforeModel hook.
OptionalcontextSchema?: TContextSchemaThe schema of the middleware context. Middleware context is read-only and not persisted between multiple invocations. It can be either:
OptionalmodifyModelRequest?: (The function to modify the model request. This function is called after the beforeModel hook of this middleware and before the model is invoked.
It allows to modify the model request before it is passed to the model.
The name of the middleware
OptionalstateSchema?: TSchemaThe schema of the middleware state. Middleware state is persisted between multiple invocations. It can be either:
Optionaltools?: any[]Additional tools registered by the middleware.
A middleware instance
const authMiddleware = createMiddleware({
name: "AuthMiddleware",
stateSchema: z.object({
isAuthenticated: z.boolean().default(false),
}),
contextSchema: z.object({
userId: z.string(),
}),
beforeModel: async (state, runtime, controls) => {
if (!state.isAuthenticated) {
return controls.terminate(new Error("Not authenticated"));
}
},
});
LangChain Agents