Configuration - Theta Health

Configuration Overview

Mirobody Health uses YAML-based configuration files for all system settings. This guide covers all configuration options and how to set them up properly.

Configuration changes require restarting the application to take effect.

Configuration File Structure

Mirobody Health uses environment-specific configuration files:

.env - Specifies which configuration to use
config.[env].yaml - Your environment-specific configuration
config.yaml - Base configuration (immutable, do not edit)

Setup Steps

Create .env file

echo "ENV=localdb" > .env

The ENV variable determines which config file to load:

ENV=localdb → loads config.localdb.yaml
ENV=production → loads config.production.yaml
ENV=staging → loads config.staging.yaml

Create your configuration file

Create config.localdb.yaml (or match your ENV value):

touch config.localdb.yaml

Configure essential settings

Add at least one LLM API key and configure agents:

config.localdb.yaml

# ============ LLM API Keys (At least one required) ============
GOOGLE_API_KEY: ''      # For Gemini models
OPENAI_API_KEY: ''      # For GPT models
OPENROUTER_API_KEY: ''  # For Claude, DeepSeek, etc.

# ============ Agent Configuration ============
MCP_TOOL_DIRS:
  - mirobody/pub/tools
  - mirobody/pub/tools_health

MCP_RESOURCE_DIRS:
  - mirobody/pub/resources

AGENT_DIRS:
  - mirobody/pub/agents

Do not edit config.yaml - This is the base configuration file that should remain unchanged. All customizations go in your environment-specific file (e.g., config.localdb.yaml).

Agent Configuration

Mirobody Health includes two agent types: DeepAgent (LangChain-based) and BaselineAgent (native Gemini/MiroThinker).

DeepAgent Configuration

DeepAgent supports multiple LLM providers and is recommended for complex tasks:

config.localdb.yaml

# API Keys for DeepAgent
GOOGLE_API_KEY: 'your_google_api_key'        # For Gemini models
OPENAI_API_KEY: 'your_openai_api_key'        # For GPT models
OPENROUTER_API_KEY: 'your_openrouter_key'    # For Claude, DeepSeek, etc.

# Provider Configurations for DeepAgent
PROVIDERS_DEEP:
  gemini-3-flash:
    llm_type: google-genai
    api_key: GOOGLE_API_KEY
    model: gemini-3-flash-preview
    temperature: 1.0
  
  gpt-5.1:
    llm_type: openai
    api_key: OPENAI_API_KEY
    base_url: https://api.openai.com/v1/
    model: gpt-5.1
    temperature: 0.1
  
  claude-sonnet:
    llm_type: openai
    api_key: OPENROUTER_API_KEY
    base_url: https://openrouter.ai/api/v1
    model: anthropic/claude-sonnet-4.5
    temperature: 0.1

# Optional: Custom prompts
PROMPTS_DEEP:
  - mirobody/pub/agents/deep/prompts/theta_health.jinja

# Optional: Tool filtering
ALLOWED_TOOLS_DEEP: []     # Leave empty to allow all
DISALLOWED_TOOLS_DEEP: []  # Specify tools to disable

BaselineAgent Configuration

BaselineAgent is optimized for lightweight operations:

config.localdb.yaml

# API Keys for BaselineAgent
GOOGLE_API_KEY: 'your_google_api_key'           # For Gemini models
MIROTHINKER_API_KEY: 'your_mirothinker_key'     # For MiroThinker

# Provider Configurations for BaselineAgent
PROVIDERS_BASELINE:
  gemini-3-flash:
    api_key: GOOGLE_API_KEY
    model: gemini-3-flash-preview
  
  gemini-2.5-flash:
    api_key: GOOGLE_API_KEY
    model: gemini-2.5-flash
  
  miro-thinker:
    api_key: MIROTHINKER_API_KEY
    model: miro-thinker

For detailed comparison of agent capabilities, see Tools Overview.

OAuth Provider Configuration

To integrate health device providers, you need OAuth credentials from each provider’s developer portal.

Garmin Connect

Visit Garmin Connect Developer Portal
Sign in with your Garmin account
Navigate to “Applications” and create a new application
Fill in application details and submit for approval

Garmin approval process can take several business days.

Get OAuth Credentials

Once approved, you’ll receive:

Consumer Key (Client ID)
Consumer Secret (Client Secret)

Configure Callback URL

Set your callback URL in the Garmin developer portal:

http://your-domain.com/api/v1/pulse/theta/theta_garmin/callback

For local development:

http://localhost:18080/api/v1/pulse/theta/theta_garmin/callback

Add to your config file

config.localdb.yaml

# Garmin OAuth1 Configuration
GARMIN_CLIENT_ID: 'your_consumer_key'
GARMIN_CLIENT_SECRET: 'your_consumer_secret'
GARMIN_REDIRECT_URI: 'http://localhost:18080/api/v1/pulse/theta/theta_garmin/callback'

# API Endpoints (usually don't need to change)
GARMIN_AUTH_URL: 'https://connect.garmin.com/oauthConfirm/'
GARMIN_TOKEN_URL: 'https://connectapi.garmin.com/oauth-service/oauth/request_token'
GARMIN_ACCESS_TOKEN_URL: 'https://connectapi.garmin.com/oauth-service/oauth/access_token'
GARMIN_API_BASE_URL: 'https://apis.garmin.com/wellness-api/rest'

Whoop

Visit Whoop Developer Portal
Create a developer account
Create a new application
Set required scopes:
- offline (for refresh tokens)
- read:recovery
- read:sleep
- read:cycles
- read:profile
- read:workout
- read:body_measurement

Get OAuth2 Credentials

You’ll receive:

Client ID
Client Secret

Configure Redirect URI

Set your redirect URI in the Whoop developer portal:

http://your-domain.com/api/v1/pulse/theta/theta_whoop/callback

For local development:

http://localhost:18080/api/v1/pulse/theta/theta_whoop/callback

Add to your config file

config.localdb.yaml

# Whoop OAuth2 Configuration
WHOOP_CLIENT_ID: 'your_client_id'
WHOOP_CLIENT_SECRET: 'your_client_secret'
WHOOP_REDIRECT_URI: 'http://localhost:18080/api/v1/pulse/theta/theta_whoop/callback'

# Scopes (space-separated)
WHOOP_SCOPES: 'offline read:recovery read:sleep read:cycles read:profile read:workout read:body_measurement'

# API Endpoints (usually don't need to change)
WHOOP_AUTH_URL: 'https://api.prod.whoop.com/oauth/oauth2/auth'
WHOOP_TOKEN_URL: 'https://api.prod.whoop.com/oauth/oauth2/token'
WHOOP_API_BASE_URL: 'https://api.prod.whoop.com/developer/v2'

Leave the client ID and secret empty ('') to disable a provider. The system will automatically skip disabled providers.

AI Provider Configuration

Mirobody Health supports multiple AI providers for chat functionality.

OpenAI

config.yaml

# OpenAI Configuration
OPENAI_API_KEY: 'sk-...'
OPENAI_API_KEY_RTC: 'sk-...'  # For real-time API (optional)

Get API Key

Sign up at OpenAI Platform
Navigate to API Keys section
Create a new API key

Google AI (Gemini)

config.yaml

# Google AI Configuration
GOOGLE_API_KEY: 'AIza...'

Get API Key

Visit Google AI Studio
Create a new API key

OpenRouter

config.yaml

# OPENROUTER Configuration
OPENROUTER_API_KEY: 'sk-or-...'  # For AI chat and agents

Get OpenRouter API Key

Sign up at OpenRouter
Navigate to Keys section
Create a new API key

Anthropic (Claude)

config.yaml

# Anthropic Configuration
ANTHROPIC_API_KEY: 'sk-ant-...'

Get API Key

Sign up at Anthropic Console
Navigate to API Keys
Create a new API key

Database Configuration

PostgreSQL

config.yaml

# PostgreSQL Configuration
PG_HOST: 'db'                    # 'localhost' for non-Docker setup
PG_PORT: 5432
PG_USER: 'holistic_user'
PG_DBNAME: 'holistic_db'
PG_PASSWORD: 'REPLACE_THIS_VALUE_IN_PRODUCTION'  # Change in production!
PG_SCHEMA: 'theta_ai'
PG_MIN_CONNECTION: 5
PG_MAX_CONNECTION: 20
PG_ENCRYPTION_KEY: 'REPLACE_THIS_VALUE_IN_PRODUCTION'  # Change in production!

For production, always use strong, unique passwords and encryption keys.

Redis

config.yaml

# Redis Configuration
REDIS_HOST: 'redis'              # 'localhost' for non-Docker setup
REDIS_PORT: 6379
REDIS_DB: 0
REDIS_PASSWORD: 'REPLACE_THIS_VALUE_IN_PRODUCTION'  # Change in production!
REDIS_SSL: false
REDIS_SSL_CHECK_HOSTNAME: false
REDIS_SSL_CERT_REQS: 'none'

Server Configuration

HTTP Server Settings

config.yaml

# HTTP Server Configuration
HTTP_SERVER_NAME: 'mirobody'
HTTP_SERVER_VERSION: '1.0.1'
HTTP_HOST: '0.0.0.0'             # Listen on all interfaces
HTTP_PORT: 18080                 # Change if port is in use
HTTP_URI_PREFIX: ''              # Optional URL prefix

# CORS Configuration
HTTP_HEADERS:
  Access-Control-Allow-Origin: '*'
  Access-Control-Allow-Credentials: 'true'
  Access-Control-Allow-Methods: '*'
  Access-Control-Allow-Headers: '*'
  Access-Control-Max-Age: '86400'

For production, restrict Access-Control-Allow-Origin to specific domains instead of '*'.

Public URLs

config.yaml

# Public URLs (adjust for your domain)
MCP_FRONTEND_URL: 'http://localhost:18080'
MCP_PUBLIC_URL: 'http://localhost:18080'
DATA_PUBLIC_URL: 'http://localhost:18080'
QR_LOGIN_URL: ''                 # Optional QR code login URL

Security Configuration

JWT and Encryption

config.yaml

# Security Keys
DATABASE_DECRYPTION_KEY: 'REPLACE_THIS_VALUE_IN_PRODUCTION'  # 32 characters minimum
JWT_KEY: 'myjwtkey'                                 # Strong secret key
JWT_PRIVATE_KEY: ''                                 # Optional RSA private key

Important Security Practices:

Use strong, randomly generated keys
Never commit production keys to version control
Rotate keys periodically
Use environment variables for sensitive values

Generating Secure Keys

# Generate a random 32-character key
openssl rand -hex 32

# Generate RSA key pair
openssl genrsa -out private_key.pem 2048
openssl rsa -in private_key.pem -pubout -out public_key.pem

Authentication Providers

Google OAuth
Apple Sign In

config.yaml

GOOGLE_CLIENT_ID: 'your_client_id.apps.googleusercontent.com'
FIREBASE_PROJECT_ID: 'your_project_id'

Get credentials from Google Cloud Console.

config.yaml

APPLE_TEAM_ID: 'your_team_id'
APPLE_KEY_ID: 'your_key_id'
APPLE_PRIVATE_KEY: 'your_private_key'
APPLE_CLIENT_ID: 'com.yourapp.service'
APPLE_AUTH_CLIENT_ID: 'com.yourapp.client'

Get credentials from Apple Developer Portal.

Logging Configuration

config.yaml

# Logging Configuration
LOG_NAME: ''                     # Leave empty for default
LOG_DIR: ''                      # Leave empty for stdout
LOG_LEVEL: 'INFO'                # DEBUG, INFO, WARNING, ERROR, CRITICAL

Log Levels
Example Configuration

Level	Description	Use Case
DEBUG	Detailed information	Development, debugging
INFO	General information	Production monitoring
WARNING	Warning messages	Production alerts
ERROR	Error messages	Production errors
CRITICAL	Critical failures	System failures

For development:

LOG_LEVEL: 'DEBUG'
LOG_DIR: './logs'

For production:

LOG_LEVEL: 'INFO'
LOG_DIR: '/var/log/mirobody'

Cloud Storage (Optional)

Configure S3-compatible storage for backups and assets:

config.yaml

# S3 Configuration
S3_KEY: 'your_access_key'
S3_TOKEN: 'your_secret_key'
S3_REGION: 'us-east-1'
S3_BUCKET: 'your-bucket-name'
S3_PREFIX: 'mirobody/'
S3_CDN: 'https://cdn.yourdomain.com'

S3 configuration is optional. The system works without cloud storage.

MCP Protocol Configuration

config.yaml

# MCP Tool and Resource Directories
MCP_TOOL_DIRS:
  - 'mirobody/pub/tools'
  - 'tools'                      # Optional custom tools directory

MCP_RESOURCE_DIRS:
  - 'mirobody/pub/resources'

# Agent Directories
AGENT_DIRS:
  - 'mirobody/pub/agents'

# Provider Directories
THETA_PROVIDER_DIRS:
  - 'connect/theta'              # Your custom providers

Email Configuration (Optional)

For sending email notifications and codes:

config.yaml

# Email Configuration
EMAIL_FROM: '[email protected]'
EMAIL_FROM_NAME: 'Mirobody Health'
EMAIL_TEMPLATE: ''               # Path to email templates
EMAIL_SMTP_PASS: 'your_smtp_password'

# Predefined verification codes (development only)
EMAIL_PREDEFINE_CODES:
  [email protected]: '777777'

Never use predefined codes in production!

Additional API Services

Serper (Web Search)

config.yaml

SERPER_API_KEY: ''               # From serper.dev

E2B (Code Execution)

config.yaml

E2B_API_KEY: ''                  # From e2b.dev

FAL (AI Generation)

config.yaml

FAL_KEY: ''                      # From fal.ai

OAuth Behavior Configuration

config.yaml

# OAuth temporary state TTL (seconds)
OAUTH_TEMP_TTL_SECONDS: 900      # 15 minutes

# Provider-specific settings
WHOOP_MAX_DETAIL_RECORDS: 50     # Max records to fetch in detail
WHOOP_CONCURRENT_REQUESTS: 5     # Concurrent API requests
WHOOP_REQUEST_TIMEOUT: 30        # Request timeout in seconds

Tool Management

config.yaml

# Allowed/Disallowed MCP Tools
ALLOWED_TOOLS_SYNERGY:
  - tool_name_1
  - tool_name_2

DISALLOWED_TOOLS_SYNERGY:
  - dangerous_tool

Leave these arrays empty to allow all tools by default.

Environment-Specific Configuration

Development
Production

config.yaml

# Development Settings
LOG_LEVEL: 'DEBUG'
HTTP_HOST: '0.0.0.0'
HTTP_PORT: 18080

# Use local services
PG_HOST: 'localhost'
REDIS_HOST: 'localhost'

# Relaxed CORS
HTTP_HEADERS:
  Access-Control-Allow-Origin: '*'

config.yaml

# Production Settings
LOG_LEVEL: 'INFO'
HTTP_HOST: '0.0.0.0'
HTTP_PORT: 443

# Use production services
PG_HOST: 'your-db-host.rds.amazonaws.com'
REDIS_HOST: 'your-redis.cache.amazonaws.com'

# Strict CORS
HTTP_HEADERS:
  Access-Control-Allow-Origin: 'https://yourdomain.com'

# Strong security
DATABASE_DECRYPTION_KEY: '[generated-secure-key]'
JWT_KEY: '[generated-secure-key]'
REDIS_PASSWORD: '[strong-password]'
PG_PASSWORD: '[strong-password]'

# SSL/TLS
REDIS_SSL: true

Configuration Validation

After updating config.yaml, validate your configuration:

Check syntax

python -c "import yaml; yaml.safe_load(open('config.yaml'))"

No output means the YAML syntax is valid.

Test database connection

docker-compose exec backend python -c "
from mirobody.utils.config import global_config
import asyncio

async def test():
    cfg = global_config()
    db = await cfg.get_db()
    print('Database connection: OK')

asyncio.run(test())
"

Test Redis connection

docker-compose exec redis redis-cli ping

Should return PONG.

Configuration Best Practices

Use Environment Variables

For sensitive values, use environment variables:

config.yaml

OPENAI_API_KEY: ${OPENAI_API_KEY}
PG_PASSWORD: ${POSTGRES_PASSWORD}

Then set in .env:

.env

OPENAI_API_KEY=sk-...
POSTGRES_PASSWORD=secure_password

Version Control

✅ Commit config.example.yaml
❌ Never commit config.yaml with real credentials
✅ Add config.yaml to .gitignore
✅ Document all required configuration keys

Security

Use strong, randomly generated keys
Rotate credentials periodically
Limit CORS origins in production
Enable SSL/TLS for external connections
Use secrets management (AWS Secrets Manager, HashiCorp Vault, etc.)

Performance

Adjust database connection pool sizes based on load
Configure appropriate timeouts
Tune Redis cache settings
Monitor and adjust concurrent request limits

Next Steps

Quick Start

Test your configuration with the quick start guide

Provider Integration

Learn how to integrate additional health providers

API Reference

Explore the API endpoints

Production Deployment

Deploy to production with security best practices

Getting Started

Core Concepts

Using Providers

Provider Examples

Tools & MCP

​Configuration Overview

​Configuration File Structure

​Setup Steps

​Agent Configuration

​DeepAgent Configuration

​BaselineAgent Configuration

​OAuth Provider Configuration

​Garmin Connect

​Whoop

​AI Provider Configuration

​OpenAI

​Google AI (Gemini)

​OpenRouter

​Anthropic (Claude)

​Database Configuration

​PostgreSQL

​Redis

​Server Configuration

​HTTP Server Settings

​Public URLs

​Security Configuration

​JWT and Encryption

​Generating Secure Keys

​Authentication Providers

​Logging Configuration

​Cloud Storage (Optional)

​MCP Protocol Configuration

​Email Configuration (Optional)

​Additional API Services

​Serper (Web Search)

​E2B (Code Execution)

​FAL (AI Generation)

​OAuth Behavior Configuration

​Tool Management

​Environment-Specific Configuration

​Configuration Validation

​Configuration Best Practices

​Next Steps

Quick Start

Provider Integration

API Reference

Production Deployment

Configuration Overview

Configuration File Structure

Setup Steps

Agent Configuration

DeepAgent Configuration

BaselineAgent Configuration

OAuth Provider Configuration

Garmin Connect

Whoop

AI Provider Configuration

OpenAI

Google AI (Gemini)

OpenRouter

Anthropic (Claude)

Database Configuration

PostgreSQL

Redis

Server Configuration

HTTP Server Settings

Public URLs

Security Configuration

JWT and Encryption

Generating Secure Keys

Authentication Providers

Logging Configuration

Cloud Storage (Optional)

MCP Protocol Configuration

Email Configuration (Optional)

Additional API Services

Serper (Web Search)

E2B (Code Execution)

FAL (AI Generation)

OAuth Behavior Configuration

Tool Management

Environment-Specific Configuration

Configuration Validation

Configuration Best Practices

Next Steps