Deployment

Automate-E agents deploy to Kubernetes as Deployments with ConfigMap-mounted character files.

Namespace

All AI Accountant resources live in the ai-accountant namespace:

apiVersion: v1
kind: Namespace
metadata:
  name: ai-accountant

Secrets

Each agent needs a Discord bot token and an Anthropic API key. Use SealedSecrets for GitOps-safe storage:

apiVersion: bitnami.com/v1
kind: SealedSecret
metadata:
  name: book-e-secrets
  namespace: ai-accountant
spec:
  encryptedData:
    discord-bot-token: <sealed>
    anthropic-api-key: <sealed>

For Postgres-backed memory, add a database URL secret:

apiVersion: bitnami.com/v1
kind: SealedSecret
metadata:
  name: book-e-db
  namespace: ai-accountant
spec:
  encryptedData:
    database-url: <sealed>

ConfigMap

Mount the character file as a ConfigMap:

apiVersion: v1
kind: ConfigMap
metadata:
  name: book-e-character
  namespace: ai-accountant
data:
  character.json: |
    {
      "name": "Book-E",
      "bio": "AI accounting assistant",
      ...
    }

Deployment

apiVersion: apps/v1
kind: Deployment
metadata:
  name: book-e
  namespace: ai-accountant
  labels:
    app: book-e
    runtime: automate-e
spec:
  replicas: 1
  selector:
    matchLabels:
      app: book-e
  template:
    metadata:
      labels:
        app: book-e
        runtime: automate-e
    spec:
      imagePullSecrets:
        - name: ghcr-pull-secret
      containers:
        - name: book-e
          image: ghcr.io/stig-johnny/automate-e:<sha>
          env:
            - name: CHARACTER_FILE
              value: /config/character.json
            - name: DISCORD_BOT_TOKEN
              valueFrom:
                secretKeyRef:
                  name: book-e-secrets
                  key: discord-bot-token
            - name: ANTHROPIC_API_KEY
              valueFrom:
                secretKeyRef:
                  name: book-e-secrets
                  key: anthropic-api-key
            - name: DATABASE_URL
              valueFrom:
                secretKeyRef:
                  name: book-e-db
                  key: database-url
          volumeMounts:
            - name: character
              mountPath: /config
              readOnly: true
          resources:
            requests:
              memory: "64Mi"
              cpu: "50m"
            limits:
              memory: "256Mi"
              cpu: "500m"
      volumes:
        - name: character
          configMap:
            name: book-e-character

Note

Always use 1 replica. Multiple replicas would create duplicate Discord responses.

ArgoCD

The k8s manifests in deploy/k8s/ are synced by ArgoCD from the ai-accountant repo:

source:
  repoURL: https://github.com/Stig-Johnny/ai-accountant.git
  targetRevision: HEAD
  path: deploy/k8s

Image tags use git SHAs. When the automate-e repo CI builds a new image, it updates book-e-deployment.yaml in the ai-accountant repo, triggering an ArgoCD sync.

Adding a New Agent

To deploy a second agent on the same runtime:

1. Create a new character.json for the agent
2. Add a ConfigMap with the character data
3. Add a SealedSecret with the agent's Discord token and Anthropic key
4. Add a Deployment referencing the same ghcr.io/stig-johnny/automate-e image
5. Set CHARACTER_FILE to point to the mounted config

Each agent runs as an independent pod with its own Discord connection and memory.

Resource Guidelines

Agent Load	CPU Request	Memory Request	CPU Limit	Memory Limit
Low (< 10 messages/day)	50m	64Mi	500m	256Mi
Medium (10-100/day)	100m	128Mi	1000m	512Mi
High (100+/day)	200m	256Mi	2000m	1Gi

Most of the CPU is spent on HTTP calls to the Claude API (network I/O, not compute).