what is professional-agentic-product-engineering fr?

This repository is a field guide called Professional Agentic Product Engineering, or PAPE, that teaches people how to work with AI coding agents like Claude Code at a professional level. Its core argument is that skilled agentic engineering is not about writing clever prompts, it is about building the system of habits, checks, and tools around the model. As tasks get harder, the guide says, the prompt itself gets shorter while the surrounding system grows. The guide organizes this skill into eight tiers, going from simple to advanced. Tier one covers writing prompts an agent can actually act on. Tier two covers planning and breaking work into smaller pieces before building. Tier three covers giving the agent the right background information and tools. Tier four, which the author calls the heart of the method, is about making the agent prove that a task is genuinely finished rather than just claiming it is done. Tier five covers saving progress with git and connecting automated tests and continuous integration. Tier six covers running multiple agents at the same time. Tier seven covers managing a whole group of agents together. Tier eight covers putting agents into real production systems. Readers are encouraged to start wherever matches their current skill level rather than working through every tier in order. The material is offered in three connected forms. The main guide is a markdown file meant to be read directly or fed to an agent. A website version presents the same content one tier per page. An Obsidian vault version splits the guide into linked pages so it can be explored like a network of ideas rather than read top to bottom. Beyond the written guide, the repository includes two interactive pieces. One is a tutor mode: when opened in Claude Code and given a greeting, it estimates the user's current skill level, with consent looks at the person's past local prompts to understand their habits, and then teaches lessons tailored to that person's own coding style. The other is an ambient coaching plugin that watches a user's normal work quietly and only speaks up when it notices a specific bad habit, such as deleting a failing test instead of investigating why it failed, then points to the relevant tier in the guide. The project targets engineers, technical founders, product managers, senior leaders, and non-technical professionals who want hands-on experience directing coding agents rather than just reading slides about them.