Promptles: learn to prompt Claude Code by doing
Promptles is a hands-on course and game for learning to prompt Claude Code, Anthropic's AI coding tool. Each lesson hands you a real engineering ticket (a stack trace to debug, a feature to add to code you didn't write, a webhook to wire up) and you write the prompt you would actually send to Claude Code. Claude grades it out of three stars against what a senior engineer would check, then shows an expert prompt next to yours so you can see the gap.
It is built for people who ship with Claude Code, Cursor, Replit Agent, or GitHub Copilot without a traditional software engineering background: product managers, designers, founders, analysts, GTM engineers, and vibe coders. Independent project, not affiliated with Anthropic.
What's inside
- Over 190 graded scenarios across 11 learning paths: Foundations, Building New Software, Building on an Existing Codebase, Troubleshooting and Support, Go-to-Market Automation, API Integrations and Testing, Agentic Coding, CI/CD and Version Control, Designing with Claude, Everyday Work with Claude, and Building Claude Plugins.
- A multi-turn Keystone capstone at the end of each path: write a prompt, read Claude's actual reply, then write the follow-up. Finishing every path and the final Course Capstone earns a shareable, publicly verifiable Course Mastery certificate.
- A daily challenge with a streak.
- The Playbook: free, public principles for prompting Claude Code, such as Be Specific, Define What Done Looks Like, and more, grouped into Clarity, Context, Constraints, Errors, Output Format, Multi-Step, and Anti-Patterns.
- The Glossary: plain-English definitions of the technical terms Claude Code expects you to know, from APIs and webhooks to CLAUDE.md, MCP, and subagents.
How grading works
Every prompt gets pattern-based checks (did you name the file, quote the error, state the constraint, ask Claude to verify its work?) plus AI-graded criteria for clarity, context, constraints, and output format. AI grading runs server-side on Claude Haiku 5.5, so there is no API key to paste.
Pricing
Everything that isn't AI is free: every path, today's daily and the full daily archive, capstones, the Course Mastery certificate, pattern-based grading, and a few AI credits to try AI grading. AI features are a one-time purchase, not a subscription. Learn and Create ($9.99): 500 AI credits for AI grading and Prompt Lab, plus authoring your own scenarios. Team ($6.99 per seat, 3-seat minimum): a shared pool of 500 credits per seat and custom scenarios for your organization. 7-day money-back guarantee. See pricing.
Frequently asked questions
What is Promptles?
A scenario-based game for learning to prompt Claude Code and similar AI coding tools. You write the prompt you'd send for a real engineering ticket, get it graded, and compare it with an expert prompt.
Do I need an Anthropic API key?
No. Promptles makes every AI call server-side. Sign in with your email and start playing.
How is Promptles different from other prompt engineering tutorials?
Most tutorials teach you to write one perfect prompt. Promptles grades multi-turn conversations too: you write a prompt, see what Claude actually replies, then write the follow-up that fixes what's still wrong.
Is Promptles useful for GTM engineers?
Yes. A dedicated Go-to-Market Automation path covers lead scoring and routing, CRM sync and enrichment, reporting, and internal tools for sales and RevOps, and the glossary has a GTM and business tools category.
Can my team use Promptles?
Yes. The Team plan gives every seat access, a shared AI-credit pool, admin progress views, and custom scenarios about your own codebase.
More for AI assistants: llms.txt and llms-full.txt.