Ready-to-use patterns and YAML manifests for orchestrating AI agents with Kelos. These examples demonstrate how to combine Tasks, Workspaces, and TaskSpawners into functional AI workflows.
- Kubernetes cluster (1.28+) with Kelos installed (
kelos install) kubectlconfigured
| Example | Description |
|---|---|
| 01-simple-task | Run a single Task with an API key, no git workspace |
| 02-task-with-workspace | Run a Task that clones a git repo and can create PRs |
| 03-taskspawner-github-issues | Automatically create Tasks from labeled GitHub issues |
| 04-taskspawner-cron | Run agent tasks on a cron schedule |
| 05-task-with-agentconfig | Inject reusable instructions and plugins via AgentConfig |
| 06-fork-workflow | Discover upstream issues and work in a fork |
| 07-task-pipeline | Chain Tasks with dependsOn and pass results between stages |
| 08-task-with-kelos-skill | Give an agent the Kelos skill for authoring and debugging resources |
- Pick an example directory.
- Read its
README.mdfor context. - Edit the YAML files and replace every
# TODO:placeholder with your real values. - Apply the resources:
kubectl apply -f examples/<example-directory>/- Watch the Task progress:
kubectl get tasks -w- Secrets first — always create Secrets before the resources that reference them.
- Namespace — all examples use the
defaultnamespace. Changemetadata.namespaceif you use a different one. - Cleanup — delete resources with
kubectl delete -f examples/<example-directory>/. Owner references ensure that deleting a Task also cleans up its Job and Pod.