Helm Values Reference

Full reference for the Skyflo Helm chart. Only engine.secrets.llmModel and the corresponding provider API key are required. Everything else has sensible defaults.

Global Settings

yaml

global:
  imageTag: ""           # Override image tag for all components
  nodeSelector: {}       # Placement defaults for all components
  tolerations: []
  affinity: {}
 
nameOverride: ""
fullnameOverride: ""
imagePullSecrets: []

Engine

The Engine is the core backend. It runs the LangGraph workflow, handles auth, and communicates with the MCP server.

yaml

engine:
  image:
    repository: skyfloaiagent/engine
    tag: ""
    pullPolicy: Always
  replicas: 1
  resources:
    requests:
      cpu: 200m
      memory: 512Mi
    limits:
      cpu: 1000m
      memory: 1Gi

Engine Configuration

yaml

  config:
    appName: "Skyflo - Engine"
    appDescription: "Core engine for Skyflo"
    debug: "false"
    apiV1Str: "/api/v1"
    logLevel: "INFO"
    rateLimitingEnabled: "true"
    rateLimitPerMinute: "100"
    jwtAlgorithm: "HS256"
    jwtAccessTokenExpireMinutes: "15"
    jwtRefreshTokenExpireDays: "7"
    llmMaxIterations: "25"

Engine Secrets

These are the most important values. Set llmModel and the API key for your chosen provider.

yaml

  secrets:
    existingSecret: ""        # Use an existing Secret instead of creating one
    jwtSecret: ""             # Auto-generated if empty, preserved across upgrades
    llmModel: ""              # Required: e.g. "gemini/gemini-2.5-pro"
    llmHost: ""               # Optional: self-hosted LLM endpoint
    mcpServerUrl: ""          # Override auto-computed internal URL
    integrationsSecretNamespace: ""
 
    # Reasoning model settings (no-ops unless explicitly set)
    llmReasoningEffort: ""       # low, medium, high (default: high for reasoning models)
    llmThinkingBudgetTokens: "" # Anthropic-specific thinking budget (default: 10000)
    llmMaxTokens: ""            # Max tokens when thinking is enabled (default: 16384)
 
    # Set only the key for your chosen provider
    openaiApiKey: ""
    anthropicApiKey: ""
    geminiApiKey: ""
    groqApiKey: ""
    moonshotApiKey: ""
    deepseekApiKey: ""
    mistralApiKey: ""
    azureApiKey: ""
    ollamaApiKey: ""
    # ... 25+ additional provider keys supported

See Multi-LLM Support for the full list of supported providers.

MCP Server

The MCP server exposes typed infrastructure tools to the Engine.

yaml

mcp:
  image:
    repository: skyfloaiagent/mcp
    tag: ""
    pullPolicy: Always
  replicas: 1
  resources:
    requests:
      cpu: 200m
      memory: 512Mi
    limits:
      cpu: 1000m
      memory: 1Gi
  config:
    appName: "Skyflo - MCP Server"
    appDescription: "MCP Server for Skyflo"
    debug: "false"
    logLevel: "INFO"
    maxRetryAttempts: "3"
    retryBaseDelay: "60"
    retryMaxDelay: "300"
    retryExponentialBase: "2.0"

UI (Command Center)

The Next.js operator interface with an Nginx reverse proxy.

yaml

ui:
  image:
    repository: skyfloaiagent/ui
    tag: ""
    pullPolicy: Always
  proxy:
    image:
      repository: skyfloaiagent/proxy
      tag: ""
      pullPolicy: Always
    resources:
      requests:
        cpu: 100m
        memory: 128Mi
      limits:
        cpu: 200m
        memory: 256Mi
  replicas: 1
  service:
    type: NodePort
  resources:
    requests:
      cpu: 100m
      memory: 256Mi
    limits:
      cpu: 500m
      memory: 512Mi
  config:
    appName: "Skyflo"
    nodeEnv: "production"
    enableAnalytics: "false"
    enableFeedback: "false"
    apiUrl: ""                # Override auto-computed engine URL
    publicApiUrl: "/api/v1"

Kubernetes Controller

The Go operator managing Skyflo's lifecycle via CRD.

yaml

controller:
  image:
    repository: skyfloaiagent/k8s-controller
    tag: ""
    pullPolicy: Always
  replicas: 1
  resources:
    requests:
      cpu: 100m
      memory: 64Mi
    limits:
      cpu: 500m
      memory: 128Mi

PostgreSQL

Built-in PostgreSQL. Disable to use an external instance.

yaml

postgresql:
  enabled: true
  image:
    repository: postgres
    tag: "15.15-alpine"
  auth:
    username: skyflo
    password: ""             # Auto-generated if empty
    database: skyflo
    port: 5432
    existingSecret: ""       # Use existing Secret with POSTGRES_USER, POSTGRES_PASSWORD, etc.
  persistence:
    enabled: true
    size: 5Gi
    storageClass: ""
  resources:
    requests:
      cpu: 200m
      memory: 512Mi
    limits:
      cpu: 1000m
      memory: 1Gi
  external:
    host: ""
    port: 5432
    database: skyflo
    username: skyflo
    password: ""
    url: ""                  # Full connection URL (overrides individual fields)

Redis

Built-in Redis. Disable to use an external instance.

yaml

redis:
  enabled: true
  image:
    repository: redis
    tag: "7.4.7-alpine"
  persistence:
    enabled: true
    size: 1Gi
    storageClass: ""
  resources:
    requests:
      cpu: 100m
      memory: 256Mi
    limits:
      cpu: 500m
      memory: 512Mi
  external:
    url: ""                  # Full Redis URL, e.g. redis://host:6379/0

Network Policy

Restricts MCP ingress to Engine pods only.

yaml

networkPolicy:
  enabled: true