A new Python web framework marketed as "designed for humans and agents" reached the Hacker News front page April 14, 2026. Plain, built by the Dropseed team behind PullApprove, is a Django fork that prioritizes explicit, typed, and predictable code—qualities the creators argue benefit both human programmers and AI coding systems.
Agent-First Architecture Includes Rules System and Searchable Documentation
Plain implements several features specifically for AI agent consumption. The framework includes a rules system where project guardrails are stored in agent-specific directories like .claude/rules/, providing context and constraints to AI coding tools. Documentation is command-line accessible with search capabilities designed for LLM consumption.
Built-in "skills" provide slash-command workflows including /plain-install for setup, /plain-upgrade for updates, /plain-optimize for performance, and /plain-bug for debugging. The framework uses explicit type annotations throughout, with models using decorators like @postgres.register_model with full type hints.
Bootstrap Projects by Piping URLs Directly to Claude or Other AI Agents
Developers can initialize new Plain projects by piping a URL directly to AI coding agents: mkdir my-app && cd my-app && claude "$(curl -sSf https://plainframework.com/start.md)". The initialization process works with Claude, Codex, Amp, OpenCode, and other AI coding agents.
"What's good for humans is good for agents," the team states in documentation, positioning the framework's explicit design as beneficial for both constituencies.
Opinionated Stack Includes 30 First-Party Packages Across Development Lifecycle
Plain enforces Python 3.13+, PostgreSQL exclusively, Jinja2 templates, htmx and Tailwind CSS for frontend, uv for package management, ruff for linting/formatting, ty for type checking, oxc for JavaScript linting, esbuild for bundling, and pytest for testing.
The framework ships 30 first-party packages covering foundation (core framework, PostgreSQL ORM, auth, sessions), backend (REST APIs, background jobs, email, caching), frontend (dynamic components, styling, static pages), development (local server, testing, debugging), production (admin interfaces, request tracing, feature flags, security scanning), and authentication (password login, OAuth, magic links).
Command-line tools include plain dev for development server with auto-reload and HTTPS, plain fix for unified code formatting across Python, CSS, and JavaScript, plain check for validation, plain test for pytest integration, and plain docs for searchable framework documentation.
GitHub Repository Shows Active Development with 731 Stars
The project's GitHub repository shows 36,030 commits, 15 forks, and 731 stars, with a codebase that is 92.2% Python, 5.1% HTML, and 1.7% JavaScript. Released under BSD-3 license, the framework differentiates itself through its entire architecture—from type annotations to documentation formats to built-in agent tooling—optimized for both human and AI consumption.
The Hacker News post received 61 points with 23 comments, with discussion focusing on whether the explicit agent-oriented design represents meaningful evolution or marketing positioning.
Key Takeaways
- Plain is a Django fork explicitly designed for both human developers and AI coding agents, emphasizing explicit, typed, and predictable code
- The framework includes agent-specific features like rules systems, command-line accessible documentation, and slash-command workflows
- Developers can bootstrap new projects by piping initialization URLs directly to Claude, Codex, or other AI agents
- Plain ships 30 first-party packages covering the full development lifecycle from local development to production deployment
- The GitHub repository has 731 stars and 36,030 commits, released under BSD-3 license with Python 3.13+ and PostgreSQL as core requirements