pydantic-settings-python

Use for writing, reviewing, debugging, migrating, or testing Python application configuration built with pydantic-settings. Trigger for BaseSettings, SettingsConfigDict, environment names, dotenv, sec

SKILL.md

GitHub

Pydantic Settings source contracts

Treat settings as two ordered phases:

configured sources -> candidate field values -> Pydantic validation -> Settings

BaseSettings is an application configuration boundary, not a global service locator. Instantiate it once near the composition root and inject the validated result.

Workflow

  1. Inspect the installed pydantic-settings and Pydantic versions separately. Do not infer one package's API from the other.
  2. Define the settings ownership boundary, field types, required values, defaults, environment prefix, case policy, nested delimiter, dotenv policy, secrets source, CLI use, and extra-key behavior.
  3. List every enabled source and its required priority. Use the documented defaults only when they match the application contract; otherwise implement and test settings_customise_sources explicitly.
  4. Use nested Pydantic models for nested configuration. Use Pydantic secret types for sensitive values, but do not mistake redacted representation for storage or access control.
  5. Keep I/O, network secret retrieval, and business logic out of validators. Add a narrow custom source only when built-in sources cannot express the input contract.
  6. Test each source alone, source collisions, malformed values, missing required fields, unknown dotenv keys, nested overrides, secret redaction, and isolated environment/filesystem state.

Canonical application boundary

from pydantic import BaseModel, SecretStr
from pydantic_settings import BaseSettings, SettingsConfigDict


class DatabaseSettings(BaseModel):
    host: str
    port: int = 5432
    password: SecretStr


class Settings(BaseSettings):
    model_config = SettingsConfigDict(
        env_prefix="APP_",
        env_nested_delimiter="__",
        env_file=".env",
        extra="forbid",
    )

    debug: bool = False
    database: DatabaseSettings


def load_settings() -> Settings:
    return Settings()

Do not manually call os.getenv inside validators. Complex environment values can use JSON decoding; nested keys override the matching top-level JSON subkey. If a different grammar is required, make it explicit and reject ambiguity.

Source priority

Default priority can include CLI arguments when enabled, then initializer, environment, dotenv, secrets directory, and defaults. Verify current docs and the installed signature before relying on the exact sequence. When policy says environment must override initializer values, return the sources in that exact order from settings_customise_sources and test a collision.

Never commit production secrets in .env, log a settings dump, or claim that SecretStr encrypts data. Secrets-directory file permissions, rotation, and provider authorization remain external controls.

Completion requires isolated tests of names, conversion, nesting, precedence, extras, missing and malformed values, and secret exposure. Read the source model, settings recipes, and verification matrix for nontrivial configurations.