Kuberns AI agent
Kuberns uses an agent-first workflow for new application and environment creation. A temporary agent host examines the selected source, determines deployment requirements, and returns a structured configuration to the Kuberns backend.
The agent is a provisioning step, not the long-running application runtime. After it completes or fails, Kuberns cleans up the temporary agent infrastructure.
What the agent detects
The agent produces configuration for the deployment pipeline rather than modifying arbitrary application code.
Source layout
- Framework name.
- Root directory.
- Dependency file name and path.
- Runtime and runtime packages.
Build and runtime
- Base image.
- Web port.
- Pre-build commands.
- Post-build commands.
- Procfile commands for web and worker processes.
- Runtime copy and cleanup instructions.
- Generated Dockerfiles and deployable image references.
Environment variables
The agent can identify environment-variable names and values available to the analysis session. Review all generated or detected values before using them in production.
Do not commit secrets to the repository. Manage runtime secrets from the environment's Environment Variables tab.
Resources
The result can request categories including:
- Server.
- Background worker.
- Database.
- Queue.
- Cache.
Kuberns then maps those requirements to supported managed resource types and plans in the selected region.
Kuberns stores the accepted result, provisions resources, and hands the environment to the standard build and deployment pipeline.