
A working issue tracker with an unrunnable repository around it: no AGENTS.md, task commands that do not match the docs, a clean clone that will not install, and the server port written into the commands instead of read from the environment.
A working issue tracker whose conventions live only in people's heads: strict type checking switched off on both halves, no formatter wired into the lint command, an unpinned toolchain, and an AGENTS.md Conventions section full of rules that no lint, formatter or compiler setting enforces.
A working issue tracker whose configuration is scattered: the database URL is a literal in the source, the only copy of the credentials sits in a tracked .env, module after module reads the environment for itself, and a missing variable surfaces as a cryptic crash minutes after boot.
Build, from an empty repository, the starter any coding agent can clone and be productive in: a server and a React + Vite client behind one build command, an AGENTS.md whose commands are real, one validated configuration per half, conventions enforced by strict typing and a linter, and the Tracker issue tracker on top of it.
Build full-stack apps the way AI coding agents need them: real AGENTS.md commands, conventions enforced by tooling, clean config. MERN, Django, Spring Boot, .NET.
This path is about preparing a codebase so an AI coding agent can work in it without guessing. You should already be able to build and run a web app in one of the stacks it covers: MERN, Django, Spring Boot or .NET. No experience with coding agents is needed. The first module ends with a full-stack starter that an agent can clone, install and run by following its AGENTS.md.
1 module: The Agent-Ready Repository.
4 exercises, all easy.
Challenges run on Django, .Net, Mern, React and Springboot.
No. Every challenge runs in a browser-based IDE with everything already set up, and your code is graded automatically against a test suite.