Skip to main content
This guide covers all configuration options available in the SKU Semantic Search application. Configuration is managed through environment variables using Pydantic Settings.

​
Configuration file

All configuration is defined in app/core/config.py using Pydantic’s BaseSettings class, which automatically loads values from environment variables and .env files.

​
Environment variables

​
Application settings

​
PROJECT_NAME
string
default:"SKU Semantic Search"
The name of the application. This value is used in the FastAPI documentation and API metadata.Status: Optional (has default value)
​
VERSION
string
default:"1.0.0"
The version of the application. Displayed in API documentation.Status: Optional (has default value)

​
Database configuration

​
DATABASE_URL
string
required
PostgreSQL connection string with pgvector extension support.Status: Required
Format: postgresql://username:password@host:port/database
Example: postgresql://user:pass@localhost:5435/skudb
The database must have the pgvector extension installed for vector similarity search to work.

​
AI provider configuration

​
GEMINI_API_KEY
string
required
Google Gemini API key for embeddings and text generation.Status: Required
Used for:
  • Generating embeddings via gemini-embedding-001
  • Primary LLM for answer generation (Gemini 2.5 Flash, Gemini Flash Latest, Gemini 2.5 Pro)
Get your key: Google AI Studio
This is the primary API key. If not provided or invalid, the system will fall back to Anthropic, but embeddings will fail.
​
ANTHROPIC_API_KEY
string
required
Anthropic API key for fallback LLM generation.Status: Required
Used for:
  • Fallback LLM when Gemini fails (Claude 3 Haiku, Claude 3.5 Sonnet)
Get your key: Anthropic Console
The system implements intelligent failover. If Gemini models fail due to rate limits or network issues, Claude models are automatically used.

​
Security configuration

​
JWT_SECRET
string
required
Secret key for signing JWT tokens used in authentication.Status: Required
Security: Must be a strong, random string. Never commit this value to version control.Generate a secure secret:
openssl rand -hex 32
Use a cryptographically secure random string. Weak secrets compromise your entire authentication system.
​
ALGORITHM
string
default:"HS256"
Algorithm used for JWT token encoding and decoding.Status: Optional (has default value)
Default: HS256
Supported values: HS256, HS384, HS512
​
ACCESS_TOKEN_EXPIRE_MINUTES
integer
default:"30"
Expiration time for JWT access tokens in minutes.Status: Optional (has default value)
Default: 30 minutes
Shorter expiration times are more secure but require users to log in more frequently. Adjust based on your security requirements.

​
Setting up your environment

​
Create a .env file

Create a .env file in the root directory of your project: 
# Application
PROJECT_NAME="SKU Semantic Search"
VERSION="1.0.0"

# Database
DATABASE_URL="postgresql://user:password@localhost:5435/skudb"

# AI Providers
GEMINI_API_KEY="your_google_gemini_api_key_here"
ANTHROPIC_API_KEY="your_anthropic_api_key_here"

# Security
JWT_SECRET="your_secure_random_secret_here"
ALGORITHM="HS256"
ACCESS_TOKEN_EXPIRE_MINUTES=30
Never commit your .env file to version control. Add it to .gitignore.

​
Configuration loading order

Pydantic Settings loads configuration in the following order (later sources override earlier ones):
  1. Default values defined in Settings class
  2. Values from .env file
  3. Environment variables from the system
This allows you to:
  • Use .env for local development
  • Override with system environment variables in production
  • Keep sensible defaults for optional settings

​
Accessing configuration

Configuration is available throughout the application via the settings singleton: 
from app.core.config import settings

# Access any setting
print(settings.PROJECT_NAME)
print(settings.DATABASE_URL)

​
Production considerations

In production environments (Docker, Kubernetes, cloud platforms), set environment variables directly rather than using .env files. This is more secure and follows twelve-factor app principles.Docker example:
environment:
  - DATABASE_URL=postgresql://user:pass@db:5432/skudb
  - GEMINI_API_KEY=${GEMINI_API_KEY}
  - ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY}
  - JWT_SECRET=${JWT_SECRET}
Regularly rotate your JWT_SECRET and API keys. When rotating JWT secrets:
  1. Generate a new secret
  2. Update the configuration
  3. Restart the application
  4. Users will need to log in again (existing tokens become invalid)
For production deployments, consider using dedicated secret management:
  • AWS Secrets Manager
  • Google Cloud Secret Manager
  • Azure Key Vault
  • HashiCorp Vault
These services provide encryption, access control, and audit logging for sensitive configuration.

​
Validation

Pydantic automatically validates configuration on startup. If required variables are missing or invalid, the application will fail to start with clear error messages. Example error: 
ValidationError: 1 validation error for Settings
DATABASE_URL
  field required (type=value_error.missing)
This fail-fast behavior prevents running the application with incomplete configuration.

Build docs developers (and LLMs) love

Get started for freeTalk to us