Show HN: Plain — The full-stack Python framework designed for humans and agents

What is Plain?

Plain is a full‑stack Python framework published on GitHub that aims for explicit, typed, and predictable application code. It has been reported that the project emphasizes clear, typed models and class‑based views — think Django-style structure but with modern typing and a focus on machine‑readability. Example snippets in the README show typed Postgres models, a QuerySet API, class-based DetailView handlers, and a Router class for URLs.

Agent-friendly tooling

It has been reported that Plain includes built‑in tooling meant to be used by LLM agents as well as humans. The README advertises “rules” — short, always‑on guardrails stored in project files — plus on‑demand docs optimized for LLM consumption (plain docs --api) and a set of slash‑command “skills” for common workflows like installs, upgrades, performance optimization, and bug reporting. The project even demonstrates starting a new app by piping a remote starter MD into Claude — it has been reported that Plain intends to work with Claude, Codex, Amp, OpenCode, or “your agent of choice.”

Stack and packages

Plain is opinionated about its stack: Python 3.13+, Postgres, Jinja2 templates, htmx and Tailwind on the front end, and an Astral‑based tooling chain (uv, ruff, ty) for Python tasks. It also lists roughly 30 first‑party packages — core modules like plain.postgres, auth, sessions, plus utilities for APIs, jobs, email, caching, and production tools such as admin and tracing. All packages ship with built‑in docs, according to the repository.

Why it matters

Tools that nudge codebases toward explicit typing and machine‑readable interfaces are trending fast as LLMs move from toy to teammate. Plain is pitching itself at that intersection — readable for humans, predictable for agents. Will it stick? Time will tell. But if you’ve ever wanted a framework that feels like giving your project a stern, helpful personal assistant, Plain is worth a look; the source lives on GitHub for anyone who wants to try it out.

Sources: github.com/dropseed, Hacker News