Free, self-paced courses arranged in tracks: AI fluency for non-technical users, product training for everyday use and deeper material for developers working with the API. A certificate follows on completion, and registration requires only an email.

The reference for everything from basic usage to API configuration, Claude Code and the prompt-engineering guide. The place to check when something needs to be precise.

Anthropic’s own notes on agentic coding: configuring the setup, writing CLAUDE.md files, planning before implementation and verifying the output. The most useful official account of how the tool is meant to be used.

Five practical courses, best taken in order: API fundamentals, an interactive prompt-engineering tutorial, real-world prompting, prompt evaluation and tool use. Each is a Jupyter notebook to run locally.

Reusable code samples for more advanced work: PDF processing, embeddings, citations and the like. Most useful once the foundational courses are behind you.

A browsable set of example prompts for writing, coding, analysis and business tasks. Each is annotated, so the design choices behind it are visible and easy to adapt.

A thorough account of AI-assisted academic work across slides, papers and analysis. It rests on specialised review agents, an adversarial loop of critic and fixer, quality gates and a replication-first approach to code, and ships as a template others can fork. Its ecosystem section doubles as a map of the wider field.

Tools, templates and a working philosophy for coding, teaching and presentations with Claude Code. It is the source of Referee 2, an independent audit and replication protocol; Blindspot, a structured review of one’s own output; and the Rhetoric of Decks, an approach to Beamer slides worth looking at.

A guide to artificial intelligence for professionals who do not write code. It covers executive-assistant workflows, inbox triage, proposal drafting, project dashboards and a self-improving configuration, alongside a library of skills that install with a single command. The name is a joke; the tools are not.

Research workflows organised around the paper, with a large cast of specialised agents (worker and critic pairs, referees, a data engineer and a verifier), a simulated blind peer-review pipeline and compliance with American Economic Association replication standards across the research lifecycle.

A continually updated list of Claude Code skills, slash commands and agents drawn from across the community. A good place to see what others have already built before starting your own.

A collection of tips, from the basic to the advanced: custom status lines, trimming the system prompt, editor integration and assorted tricks for managing context. Quick wins for everyday use.