Run your first task
Every piece of work in StratoNext is a task: it says what to do, who does it, and what came of it. Your first task is a short, read-only investigation, so you see the whole loop from request to written result in a few minutes.
You get: a completed task with a result you can read in the web app, and the same habit your team will use for every piece of work.
Time: about 5 minutes.
Before you start
Section titled “Before you start”- You finished Connect your cloud account.
- Your agent is opened in the project’s repository folder.
Choose who does the work
Section titled “Choose who does the work”There are two ways to run a task. Try the first one now and the second one when you are ready. See Operating modes for the full picture.
| Your local agent | A background agent | |
|---|---|---|
| Who works | Your coding agent, in this chat | A StratoNext agent running in your AWS account |
| Cloud access | Your own AWS credentials | Short-lived credentials from StratoNext, scoped to the task |
| Interactive | Yes, you can steer it | No, you read the result |
| Best for | Quick checks while you are at the keyboard | Unattended work, accounts you cannot reach locally |
Option A: with your local agent
Section titled “Option A: with your local agent”Paste this prompt:
Using the StratoNext skill, create a task in this project assigned to me: "Check the health of the deployment". Start it, then use my AWS credentials to look for errors, alarms in ALARM state, and failed invocations in the last 24 hours across the deployment's resources.Do not change anything. Complete the task with a short summary of what you found and what you would look at next.The agent follows the task lifecycle that the skill requires.
Example commands the agent might run
stratonext task create --target PROJECT --name "Check the health of the production deployment" \ --description "Errors, alarms and failed invocations in the last 24 hours" --minestratonext task start <task_id># ...investigates with your AWS credentials...stratonext task complete <task_id> --result "No alarms firing. 42 Lambda errors in orders-api, all from one bad request payload at 09:14. Suggest input validation."Option B: with a background agent
Section titled “Option B: with a background agent”A background agent runs inside your AWS account, so it needs the account connected with agent deployment enabled, as in the previous guide. You create it once and it works across every project in the workspace.
Paste this prompt:
Using the StratoNext skill, check whether this workspace has an agent. If not, create a managed agent in the environment. Then create a task in this project assigned to that agent:"Investigate the latest Lambda errors in the last 24 hours and write up the likely cause". Dispatch it, and tell me when it has a result.The agent creates the background agent if the workspace has none, then creates and dispatches the task.
Example commands the agent might run
stratonext agent liststratonext agent create --name "ops-agent" --kind MANAGED --environment-id <environment_id> --setting Region=us-east-1stratonext task create --target PROJECT --name "Investigate the latest Lambda errors" \ --description "Errors in the last 24 hours and their likely cause" \ --assignee-id <agent_id> --assignee-type AGENTstratonext task dispatch <task_id>stratonext task get <task_id>The background agent reads only, using the permissions of the environment’s operations role. If it needs more, it stops and asks for an approval instead of working around it.
Check the result
Section titled “Check the result”stratonext task list --minestratonext task get <task_id>Or open Tasks in the web app. A task that waits on you shows INPUT_REQUIRED: answer it with
stratonext task submit-input <task_id> "...", and the same task carries on.
More quick tasks to try
Section titled “More quick tasks to try”Find any S3 buckets or databases in this deployment that are open to the public, and create a report.List the most expensive resources in this deployment over the last 7 days.Check whether every Lambda function in this deployment has an error alarm, and list the ones that do not.Next step
Section titled “Next step”Watch metrics and alarms with signals so the project page shows what matters at a glance.