Materials for the 90-minute deep introduction to Claude Code: how an AI agent actually works, how to brief one, and how to structure a project so the agent gets better every week instead of starting from zero.
claude-code-101/ ← this repo: the course session (frozen, canonical)
├── slides/claude-code-101.md ← the marp deck (speaker notes inside)
├── 00-agent-loop/ ← a complete agent in ~120 lines of Python
├── exercises/HOMEWORK.md ← week 2 homework (4 exercises + stretch)
└── cheatsheet.md ← one-page reference for during class
xolo-consulting ← separate TEMPLATE repo: your own copy of the org
The demo organization lives in its own template repository —
github.com/profrodai/xolo-consulting — because you will own it and grow it
for the next 15 weeks. Click "Use this template" there to create your
copy (don't fork; you want a clean history of your own). Weekly course
additions arrive on the upstream template as tags (week-03, week-04, …)
for you to diff and pull in.
- Install Claude Code and log in — see docs.claude.com → Claude Code →
Quickstart. Verify with
claude --version. - Python 3.10+ available (
python3 --version). - Create your repo from the xolo-consulting template, clone it, then:
cd xolo-consulting && python generate_data.py
- Optional (for
00-agent-loop): an Anthropic API key.
npm i -g @marp-team/marp-cli
marp --html slides/claude-code-101.md -o slides/claude-code-101.html # or --pdf
# the --html flag is required — the deck uses inline-styled layoutsEvery week the xolo-consulting/ repo grows: more skills, more subagents,
more hooks, tighter controlling. By week 17 it should run its recurring
operations with the human only at the signature points. The repo is the
organization; the .claude/ directory is its constitution, org chart, and
SOP binder.