A collection of practical recipes for typed functional data analysis in Morel — Julian Hyde's Standard ML interpreter with relational extensions.

This repo is not a tutorial, not a reference manual, not a language spec. It is a cookbook: 20 self-contained pages that each answer one concrete question a working data person might have. "How do I group and aggregate?" "How do I handle missing values?" "How do I define a metric once and reuse it?" Each recipe is short, runnable in the browser, and cross-linked to the handful of related ones.

Working data people — analysts, engineers, quants, actuaries — who already know their way around SQL or pandas and want to see what types, pattern matching, and relational composition in a real programming language can do for data work. No prior ML / Haskell / OCaml experience required. No theory about monads.

Planning and scaffolding documents at the top level, recipes in recipes/, the Astro site (once it exists) in site/.

• AGENTS.md — instructions for Claude Code or any agent working on this repo. Read this first.
• ROADMAP.md — phased plan from empty repo to launch.
• RECIPES.md — the 20-recipe outline, in order, with per-recipe notes.
• CONVENTIONS.md — recipe page template and editorial conventions. Follow these.
• ARCHITECTURE.md — tech stack, the Morel-in-browser plan, hosting.
• OPEN_QUESTIONS.md — things to verify about Morel 0.8 before committing to specific recipe designs.
• recipes/ — recipe pages as they get written.

Pre-v1. Planning and scaffolding only. The site does not yet exist.

TBD. See ARCHITECTURE.md for the planned stack.

Pick an unwritten recipe from RECIPES.md, follow the template in CONVENTIONS.md, open a PR. Small and opinionated beats big and bloated — a cookbook page that tries to be a tutorial is a bad cookbook page.

Apache 2.0 — same as Morel itself.