Working with ClaudeFree · koha · CC BY 4.0
All lessons

Working with Claude

A practical, free course on getting trustworthy work out of Claude — not just faster work. The companion to the Village course on judgment: that one teaches which work to give an agent; this one teaches how to get reliable results, keep custody of your data, and stay accountable.

Custody, accountability and judgment — not speed. That's the line this course holds and most courses skip: keep your data where you want it, stay the one who answers for the work, and keep your own judgment sharp while the tool gets faster.

Start here → Printable tools →

This course teaches one tool — Claude (Anthropic). The durable skills transfer; the specifics are Claude’s. Depending on a single overseas vendor is itself a decision worth understanding — that’s the first custody lesson.

What you'll be able to do

  • Know what Claude is and how it works — and where relying on a single overseas vendor fits your risk.
  • Use it well — brief it like a colleague, organise real work in Projects, and fit it to your role.
  • Evaluate what it gives back — verify claims, tell confident from correct, and keep custody of your data.
  • Stay accountable — the ethics and legal responsibility for what you send, which never leave you.

These four — know, use, evaluate, and the ethics running through them — are the aspects of AI literacy set out by Ng and colleagues (2021), and echoed in the UNESCO AI competency frameworks and the EU’s DigComp. We lean on public-interest, non-US frameworks by design.

Your progress

· Saved on this device only — no account, no tracking.

Start here

PrimerStart here — the 5-minute primer

Tier 1 · Foundations

1.1Claude vs Claude Code 1.2Prompting is delegation, not magic words 1.3Projects and Artifacts — organising real work 1.4Stop Claude flattering you, caving, and making things up 1.5What to keep out of Claude

Tier 2 · Practitioner

2.1Prompting patterns for your work 2.2The verification habit — telling when Claude is wrong 2.3Standing instructions for your role 2.4Managing a conversation 2.5Which model — and when the top tier earns its keep

Tier 3 · Power

3.1Claude's agentic surface — tools and computer use 3.2Connecting Claude to your systems (MCP) 3.3Claude alongside your other tools 3.4Scheduled and automated work

Tier 4 · Frontier

4.1Claude Code — what it is, and when you would graduate to it 4.2Capability is not agency 4.3Build your own guardrails 4.4Legal and accountability — what you actually own 4.5Custody — when to move to AI you control

Tools you can keep

Five printable one-page cards — the accuracy prompt, the keep-out checklist, your own guardrails, the five-part brief, and the model-triage prompt — drawn straight from the lessons. Open the tools →

Found the course useful? Share it with someone, or show a QR code to scan.