adopt

What problem does it solve? Onboarding an AI coding workflow onto a project that already has thousands of lines of working code is painful: greenfield templates assume an empty app, and hand-writing project plans, build plans, and coding standards from scratch is slow and error-prone. This Skill bootstraps the Blueprint workflow onto an existing repository by reading the real code and asking only for what the code cannot reveal. ## Core Features & Use Cases - Read-only codebase survey: Detects stack, frameworks, package manager, dev/build/test/lint commands, conventions, testing reality, and existing CI from manifests and source files. - Intent interview: Asks three to five targeted questions about purpose, users, and roadmap, marking uncertain inferences with > TODO (confirm) instead of guessing. - Owned file generation: Writes blueprint/project-plan.md, blueprint/build-plan.md (shipped features pre-checked), blueprint/context/coding-standards.md matched to actual conventions, and the Commands section of AGENTS.md. - Safety gates: Refuses to overwrite filled-in plans, protects existing READMEs, never runs scaffolders, and asks before untracking already-committed Blueprint files. - Use Case: You have a production Next.js app and want AI-assisted feature development. Run /adopt to generate accurate plans and standards from the real code, then run /overview to enter the normal feature loop. ## Quick Start Run /adopt in your existing repository to survey the codebase, answer a few intent questions, and generate the project plan, build plan, and coding standards.