Evenfire is an intelligence layer platform for running LLM agents that take real actions, on your infrastructure
TypeScript
apiVersion: clerum.io/v1alpha1
kind: WorkflowRecipe
metadata:
name: agentic-workload-template-resolution
spec:
description: Agentic workflow sample that validates workload env/command/args template resolution.
agent:
provider: zai
model: glm-4.7
triggers:
onDemand:
allowedActors:
- user
- autonomous
inputContract:
properties:
db_name:
type: string
default: clerum
shell_bin:
type: string
default: sh
computed:
- name: db_mode
expression: "'readonly'"
workloads:
- id: postgres
type: statefulset
image: postgres:16-alpine
port: 5432
security:
runAsUser: 70
runAsGroup: 70
fsGroup: 70
env:
- name: POSTGRES_DB
value: '{{inputs.db_name}}'
- name: POSTGRES_USER
value: clerum
envSecret:
name: agentic-template-postgres-auth
keys:
- secretKey: password
envVar: POSTGRES_PASSWORD
- id: qa-api
type: deployment
image: busybox:1.36
port: 8080
env:
- name: DATABASE_URL
value: 'postgres://clerum@{{postgres:host}}:{{postgres:port}}/{{inputs.db_name}}'
- name: DB_MODE
value: '{{computed.db_mode}}'
command:
- '{{inputs.shell_bin}}'
- -c
args:
- 'echo "db=$DATABASE_URL host={{postgres:host}} port={{postgres:port}} mode={{computed.db_mode}}"; sleep 3600'
steps:
- id: verify-template-resolution
instruction: >
Confirm that the qa-api workload received resolved PostgreSQL service
host and port values in env.value, command, and args. Workload secrets
stay on envSecret and are not rendered through env.value templates.