Python Project Radar
Current Python projects, reviewed — with an AI-familiarity grade.
Every entry below was reviewed by a maintainer against source, maintenance state, license, release history, and practical value. The ai_familiarity grade says how well mainstream coding models know the project’s current API — the number one hidden risk when an agent writes your code. Latest review: 2026-09-02.
Reviewed entries
7 projects on the radar
Descriptions and grades are human-written and dated — never generated. Machine-readable data: radar.json in the pinned repository export.
fastapi/fastapi
Production-standard ASGI framework with automatic OpenAPI docs, Pydantic validation, and dependency injection.
02567-labs/instructor
Production standard for extracting structured JSON from LLMs using Pydantic models with retry validation.
03marimo-team/marimo
Reactive, pure-Python notebook stored as standard executable .py files with deterministic state execution.
04pola-rs/polars
High-performance DataFrame library built in Rust on Apache Arrow with lazy query optimization.
05pydantic/pydantic-ai
Model-agnostic agent framework prioritizing type-safe structured outputs, dependency injection, and testability.
06astral-sh/ruff
10-100x faster linter and formatter that unifies Flake8, Black, isort, and pyupgrade rules in a single configuration.
07astral-sh/uv
Extremely fast Rust-based package and project manager that replaces pip, pip-tools, venv, and pyenv with lockfile determinism.
Method
What a Radar entry proves — and does not
Lifecycle states are new, rising, stable, major-update, experimental, archived. “New” means recently reviewed, not pre-vetted hype.
ai_familiarity grades whether mainstream model training data covers the project and its current API: low (post-dates common cutoffs), medium (known but often generated with outdated APIs), high (models reliably produce current-API code). It is a dated maintainer judgment, not a benchmark.
A Radar entry is a review record, not an endorsement to adopt blindly — each entry names when not to use the project. Propose a project through the repository’s proposal form; discovery candidates are machine-generated, verdicts never are.