Claude Code Workshop — AI Coding Agents for Researchers

A 60-minute hands-on workshop for researchers at the Utrecht School of Economics. No programming experience required.

Live slide deck: https://uufinance.github.io/agents/

What this workshop covers

Claude Code is an AI coding agent that runs in your terminal, reads your files, writes and executes code, and iterates — without you driving every step. This workshop shows researchers how to use it for:

• Research workflows — referee reports, R&R revision plans, conference slide decks, cross-language code replication (Stata → Python + R)
• Teaching & admin — building academic homepages from a CV, transcribing student meetings, creating interactive exam and competition tools
• Concepts — how AI agents work (ReAct loop, tools, memory), plan mode, permission levels, context windows, usage limits

Repo contents

agents/
├── index.html              # Full slide deck (single HTML file, no build step)
├── assets/
│   └── compustat_sample.csv  # Synthetic Compustat extract for live demos
└── .claude/
    └── skills/
        └── transcribe-meeting.md  # Local skill used in Demo 6

View locally

npx serve .
# or
python3 -m http.server 8000

Open http://localhost:8000.

Demo data

assets/compustat_sample.csv — 725 synthetic firm-year observations from 25 European firms (1995–2023). Columns match a real WRDS/Compustat annual pull: gvkey, conm, fyear, ni, at, rect, invt, dpc, oancf. Used in the live demos; participants can download it directly from the workshop site.