Configuration

Configuration

Kernova.config.json

Every Kernova project has a configuration file at its root:

{
  "name": "my-app",
  "version": "0.1.0",
  "runtime": {
    "environment": "development",
    "debug": true
  },
  "memory": {
    "provider": "local",
    "layers": {
      "working": { "maxItems": 50 },
      "session": { "ttlSeconds": 3600 },
      "longterm": { "provider": "local" }
    }
  },
  "models": {
    "default": {
      "provider": "openai",
      "model": "gpt-4o"
    },
    "fallback": {
      "provider": "anthropic",
      "model": "claude-sonnet-5-5"
    }
  },
  "providers": [],
  "budget": {
    "maxTokens": 1000000,
    "maxCostUsd": 50.00,
    "requestsPerMinute": 60,
    "maxTokensPerIntent": 10000
  },
  "studio": {
    "port": 4200,
    "requireAuth": false
  }
}

Configuration Fields

runtime

Field	Type	Default	Description
environment	string	"development"	development, staging, or production
debug	boolean	false	Enable debug logging

models

Field	Type	Description
default.provider	string	Default model provider ID
default.model	string	Default model name
fallback.provider	string	Fallback provider if default fails
fallback.model	string	Fallback model name

providers (custom)

Add any OpenAI-compatible provider:

{
  "providers": [
    {
      "id": "my-provider",
      "name": "My Custom LLM",
      "adapter": "openai",
      "baseUrl": "https://api.example.com/v1",
      "apiKeyEnv": "MY_API_KEY",
      "defaultModel": "my-model",
      "timeoutMs": 30000
    }
  ]
}

Adapter types: openai, anthropic, gemini, ollama, custom

budget

Field	Type	Default	Description
maxTokens	number	unlimited	Total token cap
maxCostUsd	number	unlimited	USD spending cap
requestsPerMinute	number	unlimited	Global rate limit
maxTokensPerIntent	number	unlimited	Per-intent token limit

studio

Field	Type	Default	Description
port	number	4200	Studio HTTP port
requireAuth	boolean	false	Require API key for API access
tlsCert	string	—	Path to TLS certificate
tlsKey	string	—	Path to TLS private key

Environment Variables

Model Provider Keys

Variable	Provider
OPENAI_API_KEY	OpenAI (GPT-4o)
ANTHROPIC_API_KEY	Anthropic (Claude)
GEMINI_API_KEY	Google Gemini
DEEPSEEK_API_KEY	DeepSeek
GROQ_API_KEY	Groq
MISTRAL_API_KEY	Mistral AI
TOGETHER_API_KEY	Together AI
OPENROUTER_API_KEY	OpenRouter
XAI_API_KEY	xAI (Grok)
COHERE_API_KEY	Cohere
FIREWORKS_API_KEY	Fireworks AI
PERPLEXITY_API_KEY	Perplexity
CEREBRAS_API_KEY	Cerebras
SAMBANOVA_API_KEY	SambaNova

Runtime Variables

Variable	Description	Default
PORT	Studio/API port	4200
Kernova_DATA_DIR	Persistence directory	./Kernova-data
Kernova_AUTH	Require auth in production	true
NODE_ENV	Environment mode	development

Data Directory

Kernova stores SQLite databases in ./Kernova-data/:

Kernova-data/
  Kernova.db       ← Events, memories, API keys, identities
  Kernova.db-wal   ← Write-ahead log (performance)
  Kernova.db-shm   ← Shared memory (performance)

Override the location:

new KernovaKernel({ dataDir: '/custom/path' });

Disable persistence entirely:

new KernovaKernel({ ephemeral: true });

TypeScript Configuration

Kernova uses ESM modules. Your tsconfig.json should include:

{
  "compilerOptions": {
    "target": "ES2022",
    "module": "NodeNext",
    "moduleResolution": "NodeNext",
    "strict": true,
    "esModuleInterop": true
  }
}