What problem does it solve? Starting a new product project with an AI assistant often means scattered context, premature tech-stack decisions, and no durable record of product direction. This Skill establishes a consistent product context spine (PRODUCT.md, PRODUCT-STATE.md, DESIGN.md, PRODUCT-RULES.md, TODO.md) so every later step in the PM workflow builds on recorded facts instead of re-asked questions. ## Core Features & Use Cases - Four-way project classification: Detects whether a target directory is a brand-new project, a documents-only folder, an existing codebase, or an already-onboarded PMAI project, and applies the correct path for each. - Context spine creation: Runs init-project.sh to generate root product documents, docs/ structure, host configuration (AGENTS.md, CLAUDE.md, hooks), and an initial git commit without deciding project type, framework, or scaffolding. - Intake manifest and baseline verification: For existing material, captures SHA-256 hashes of pre-existing files before any writes, then verifies six product-baseline questions against that evidence before allowing a direct jump to design. - Use Case: A solo PM points the assistant at a folder of PRDs and research notes; the Skill fixes an intake manifest, builds the product spine, verifies the documents form an equivalent baseline with PM confirmation, and routes to /pmai-design or /pmai-proposal accordingly. ## Quick Start Initialize this directory as a new PMAI product project and set up the product context files for my app idea.
