Internal developer course · Invite only

Prompting is a diff.

A hands-on course that teaches your team to turn vague requests into precise, reliable instructions for Claude — across the app, the API, and Claude Code. Every module ends with a real exercise that Claude itself grades, with feedback and a coach you can chat to.

Enter the course → Registration requires an invite code from your admin.
prompt.txt−1 +1
summarize this feedback
+Summarize this customer feedback into 3 themes. For each: a one-line label, the count of mentions, and one representative quote. Return it as a markdown table.
why Same intent — the second version names the shape, the granularity, and the format, so the model has nothing left to guess.

What makes this different

Reading about prompting only gets you so far. This course makes you do it, then tells you how you did.

// practise

Claude grades your work

Each module ends with an exercise. You write a real prompt; Claude scores it against a rubric and gives specific, actionable feedback.

// coach

Ask follow-up questions

Not sure why you lost points? Open a short chat with Claude about the exercise and work it through until it clicks.

// resume

Pick up where you left off

Your progress, attempts, and feedback are saved. Log out, lose your connection, come back next week — it’s all still there.

The ten modules

From the mental model of how Claude reads a prompt, through building on the API, to baking testing into how you develop with Claude.

For your developers, whatever their level

Written for engineers who are new to LLMs, but grounded in Anthropic’s current (2026) prompting guidance — so nobody picks up habits that stopped working two model generations ago.

Literal instruction following Few-shot & XML structure Adaptive thinking & effort Structured Outputs (no prefill) Tool use & Claude Code Tests-first development
Sign in to begin →