Cloud Agents
Cloud Agents are AI-powered workflows you can launch on-demand or on a schedule from the Nirmata Control Hub. They run as short-lived jobs or tasks, execute a specific analysis or remediation task, and produce a detailed report — with no persistent infrastructure required.
Common use cases include cost analysis, workload troubleshooting, RBAC security auditing, policy recommendations, and compliance checks.
Accessing Cloud Agents
Navigate to Agent Hub → Cloud Agents in the Nirmata Control Hub.
Agent Catalog
In the Agent Catalog you can:
- View all available agents and see what each agent does, what inputs it requires, and what permissions it needs on the target cluster.
- View recent runs for a specific agent.
- Launch an agent immediately or schedule it to run on a recurring basis.
Launching an Agent
When launching an agent you can:
- Select a target cluster (required for agents that run against a cluster).
- Provide agent-specific inputs such as namespace, resource names, or other parameters.
After launch, the run starts immediately and you can track its progress in real time.
Note
If the selected cluster is disconnected, the run will be retried automatically before being marked as failed.Monitoring Agent Runs
In Agent Runs you can:
- View all agent executions across all agents and trigger sources.
- Filter runs by status, agent name, or how the run was triggered (manual, schedule, Copilot, webhook, or event).
- View the full details of any run — including who or what triggered it, which cluster it targeted, when it started and completed, and any error details if it failed.
- Follow live execution logs and step-by-step progress for runs that are currently in progress.
- View the report produced by a completed run, share it, or download it.
- Delete runs you no longer need.
Scheduling Agents
In Schedules you can:
- View all configured schedules and see which are active.
- Filter schedules by agent, status, or target cluster.
- Create a new schedule by selecting an agent, a target cluster, the input parameters, and a recurring interval (hourly, every 12 hours, daily, weekly, or a custom cron expression). You can also set a timezone and enable or disable the schedule at creation time.
- Configure notifications so that an email or Slack message is sent when each scheduled run completes.
- View the full history of runs and reports produced by a specific schedule.
- Edit, enable, disable, or delete an existing schedule.
Agent Types
Cloud Agents are one of two types:
- Workflow — Multi-step agents where steps execute sequentially or in parallel. You can follow the progress of each step in real time as the agent runs.
- Task — Single-task agents that execute a focused operation and produce a report on completion.
Trigger Sources
Agent runs can be triggered in multiple ways:
| Source | Description |
|---|---|
| Manual | Launched directly from the Agent Catalog |
| Schedule | Triggered automatically by a configured schedule |
| Copilot | Launched by the NCH Copilot in response to a request |
| Webhook | Triggered by an external event (e.g., a GitHub PR or Jira issue) |
| Event | Triggered as a follow-on from another agent run |
Notifications
For scheduled runs, you can configure notifications to be sent on completion:
- Email — sends a summary to one or more recipients when the run finishes.
- Slack — posts a message to a Slack channel. Requires a Slack workspace integration configured under Settings → Integrations.
Permissions
Some agents require specific Kubernetes RBAC permissions on the target cluster. Check the agent details before launching to ensure the required permissions are in place. The agent runs under a scoped service account that is provisioned automatically and cleaned up after the run completes.