Python

Python Project Standards

When to Use

• Creating a new Python project or module
• Setting up pyproject.toml and dependency management
• Writing Python functions, classes, or scripts
• Scaffolding Python-based AWS Lambda functions
• Auditing an existing Python project for type hint coverage, missing Ruff config, or dependency management gaps
• Upgrading Python projects to modern conventions (union types, pathlib, pydantic v2)

Python Version

Target Python 3.12+ unless constraints require otherwise.

Language Features

Use the latest Python language features:

• Type hints on all function signatures and return types
• f-strings for string formatting
• List and dict comprehensions where they improve readability
• Union types using X | Y syntax (Python 3.10+)
• Match statements where appropriate (Python 3.10+)
• Walrus operator (:=) where it improves readability

Package Management

Use uv as the package and project manager. It is substantially faster than pip and handles virtual environments, lockfiles, and workspaces in a single tool.

# Create a new project
uv init my-project && cd my-project

# Add runtime dependencies
uv add fastapi mangum pydantic pydantic-settings

# Add dev-only dependencies (not bundled in Lambda)
uv add --dev pytest ruff mypy boto3

# Run tools within the project environment
uv run pytest
uv run ruff check .

# Sync environment from lockfile (CI / fresh checkout)
uv sync

• Use pyproject.toml as the single source for all project metadata and dependencies.
• The uv.lock lockfile must be committed to version control for reproducible installs.
• For AWS Lambda deployment, packages already in the Lambda runtime (boto3, botocore) belong in [dependency-groups] dev — not bundled in the artifact.

[project]
name = "my-project"
version = "0.1.0"
requires-python = ">=3.12"
dependencies = [
    "fastapi",
    "mangum",
    "pydantic",
    "pydantic-settings",
]

[dependency-groups]
dev = [
    "boto3",
    "pytest",
    "ruff",
    "mypy",
]

Data Modeling with Pydantic

Use Pydantic BaseModel as the default for all structured data that crosses a function or module boundary: API request/response models, configuration, and any externally-sourced data.

Request and Response Models

from datetime import datetime
from pydantic import BaseModel, Field

class CreateProjectRequest(BaseModel):
    name: str = Field(..., min_length=1, max_length=100)
    description: str = Field(default="", max_length=500)
    is_public: bool = False

class ProjectResponse(BaseModel):
    id: str
    name: str
    description: str
    is_public: bool
    created_at: datetime

Configuration with BaseSettings

Use pydantic-settings for environment-based configuration. It reads env vars automatically and validates types at startup — fail fast before any work begins:

from pydantic_settings import BaseSettings

class Settings(BaseSettings):
    service_name: str = "my-api"
    stage: str = "dev"
    database_url: str        # required — no default
    api_key: str             # required — no default
    max_retries: int = 3

    model_config = {"env_file": ".env", "env_prefix": "APP_"}

settings = Settings()  # reads APP_DATABASE_URL, APP_API_KEY, etc.

Choosing the Right Type

| Use | When | |-----|------| | BaseModel | Validation needed, API boundaries, config, JSON parsing | | @dataclass | Pure data containers with no validation or serialization | | TypedDict | Typing only — no runtime instances, plain dict interop |

Never use plain dict for structured data that crosses a function or module boundary.

Documentation

• Write docstrings following Google docstring standards.

def process_items(items: list[str], max_count: int = 10) -> dict[str, int]:
    """Process a list of items and return frequency counts.

    Args:
        items: List of item names to process.
        max_count: Maximum number of items to include in results.

    Returns:
        Dictionary mapping item names to their frequency counts.

    Raises:
        ValueError: If items list is empty.
    """

Error Handling

• Wrap main logic in try/except blocks.
• Log errors before re-raising or returning error codes.
• Use specific exception types rather than bare except.

Logging

• Use the logging module, never print() for operational output.
• Use appropriate log levels: DEBUG for detail, INFO for progress, WARNING for recoverable issues, ERROR for failures.

Linting and Formatting

• Use Ruff for both linting and formatting.
• Configure Ruff in pyproject.toml:

[tool.ruff]
line-length = 100
target-version = "py312"

[tool.ruff.lint]
select = ["E", "F", "I", "N", "UP", "B", "SIM"]

[tool.mypy]
strict = true

Testing

• Use pytest as the test framework.
• Name test files test_.py and test functions test_.
• Use descriptive test class and method names that read as specifications.
• Mock external dependencies (APIs, file systems, databases).
• Use tmp_path fixture for file system tests.
• Test edge cases: empty inputs, missing config, error conditions, boundary values.

IDE Integration

For VS Code / Cursor configuration with Pylance type checking, Ruff format-on-save, uv virtual environment discovery, and pytest test runner, see the ide-setup skill. For FastAPI web server patterns and Lambda deployment, see the python-web-server skill. For general testing strategy and coverage thresholds, see the testing skill. For data pipeline development with dlt, see the data-pipelines skill. For DataFrame analysis workflows with pandas, Polars, and DuckDB, see the data-analysis skill.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: michaelsvanbeek
• Source: michaelsvanbeek/personal-agent-skills
• License: MIT

Install and usage instructions live in the source repository linked above.