Skip to content

Standard

Build Python applications with clear module boundaries

Standard is an alpha Python library for shared application infrastructure and agent guidance. Install the standard distribution, import it as std, and use the std command to discover supported APIs and install the packaged Python skill. The core provides configuration, logging, collections, file helpers, and async service lifecycles. FastAPI, AWS, HTTP, Redis cache, and relational database integrations are separate distributions selected with extras. All six distributions use one version from a shared release tag.

The application pattern keeps settings, routes, schemas, data models, repositories, and services with their feature. main.py constructs settings and dependencies once, then injects them into the app and its lifecycle. The FastAPI template and sample consumer show the pattern in code.

Choose a path

If you want to… Start with…
Install Standard or run the first example Get started
Load YAML/TOML and configure Pydantic Settings Configuration reference
Configure application logs by environment Logging guide
Find public helpers and integrations API reference
Inspect a consumer app installed from the published packages FastAPI consumer sample and its evaluation report
Call HTTP services HTTP client guide
Cache data or coordinate work with Redis Cache and locks guide
Manage async database sessions Relational database guide
Compare configuration tools Research notes
Understand architecture boundaries and current decisions Decisions
Run tests or reproduce measurements Testing and benchmarks
Read release changes Changelog
Contribute Contributing
Report a bug or ask for help GitHub Issues
Report a security issue Security policy

Current release

The latest coordinated release is v0.2.4. It includes standard and five provider distributions at the same version. Installing standard alone gives you the core modules; install an extra such as standard[fastapi] to add an integration. See publishing for package links and the tag-based release process. The documentation site is published through GitHub Pages. Configuration benchmark records are exploratory and do not demonstrate the 100× target.

Keep configuration errors and source precedence explicit, keep optional integrations out of the core install, and measure equivalent work before making performance claims.