helm repo add gravity https://gravitycloudai.github.io/helm
helm repo update
helm upgrade --install matter-ai-enterprise gravity/matter-ai-enterprise -f matter-enterprise-values.yaml -n matterai --create-namespace
The Matter AI MCP component is disabled by default. To enable it, set matterMCP.enabled
to true
in your values file:
matterMCP:
enabled: true
The MCP deployment uses the gravitycloud/matter-ai-mcp
image with the following default settings:
- 1 replica
- Image tag: latest
- Resource requests: 256Mi memory, 250m CPU
- Resource limits: 512Mi memory, 500m CPU
You can customize these settings in your values file:
matterMCP:
enabled: true
deployment:
replicas: 2
image:
repository: gravitycloud/matter-ai-mcp
tag: v1.0.0
resources:
requests:
memory: "512Mi"
cpu: "500m"
limits:
memory: "1Gi"
cpu: "1000m"
The MCP service will automatically connect to the Matter Backend service through a dynamically constructed endpoint. The MATTER_API_ENDPOINT
environment variable is automatically set to http://matter-backend-service.matterai.svc.cluster.local:8080
(or the appropriate service URL based on your configuration), so no manual configuration is required.
The MCP service is configured as a ClusterIP service on port 9000 by default:
matterMCP:
enabled: true
service:
name: matter-mcp-server-service
type: ClusterIP
port: 9000
targetPort: 9000
To expose the MCP service externally, configure its ingress by setting matterMCP.ingress.enabled
to true
:
matterMCP:
enabled: true
ingress:
enabled: true
className: "nginx"
annotations:
nginx.ingress.kubernetes.io/ssl-redirect: "true"
hosts:
- host: matterai-mcp.example.com
paths:
- path: /
pathType: Prefix
tls:
- secretName: matterai-mcp-server-tls
hosts:
- matterai-mcp.example.com
The MCP deployment uses a ConfigMap volume for runtime configuration:
matterMCP:
enabled: true
deployment:
volumes:
- name: runtime-config
configMap:
name: matter-ai-mcp-config
This ConfigMap is automatically created when the MCP component is enabled.