Restack Autoscaling Agent Example

A sample repository with an agent doing memory intensive operations

Requirements

• Node 20+

Start Restack

To start Restack, use the following Docker command:

docker run -d --pull always --name restack -p 5233:5233 -p 6233:6233 -p 7233:7233 -p 9233:9233 ghcr.io/restackio/restack:main

Install dependencies and start services

npm install
npm run dev

This will start a Node.js app with Restack Services. Your code will be running and syncing with Restack to execute agents.

Run agents

from UI

You can run agents from the UI by clicking the "Run" button.

from API

You can run agents from the API by using the generated endpoint:

POST http://localhost:6233/api/agents/agentMemoryIntensive

from any client

You can run agents with any client connected to Restack, for example:

npm schedule-agent

executes scheduleAgent.ts which will connect to Restack and execute the agentMemoryIntensive agent.

Send events to the Agent

from UI

You can send events like message or end from the UI.

And see the events in the run:

from API

You can send events to the agent by using the following endpoint:

PUT http://localhost:6233/api/agents/AgentChat/:agentId/:runId

You can send following payload

{
  "eventName": "end"
}

to end the conversation with the agent.

from any client

You can send event to the agent with any client connected to Restack, for example:

Modify agentId and runId in eventAgent.ts and then run:

pnpm event-agent

It will connect to Restack and send 2 events to the agent, one to generate another agent and another one to end the conversation.

Deploy on Restack Cloud

To deploy the application on Restack, you can create an account at https://console.restack.io

Scaling on Restack cloud

This example is to showcase how Restack Cloud will handle autoscaling to handle operations for your agent. Once the app is deployed you can spawn several agents using the agent POST request on the Restack developer ui and then trigger the agentEvent event on each of the agents.

On Restack cloud console under the logs section, you will notice how more pods are spawned to handle the load.