the-y-coding-standard

Run the_y_coding_standard governance audit. Dispatches three parallel sub-agents (security, audit, permissions) inspired by Microsoft's agent-governance-toolkit. Each runs deterministic scripts and inspects the repo. Aggregates findings into findings/YYYY-MM-DD-governance.md. Use before every release.

Scaffold the_y_coding_standard agentic engineering stack into the current repo. Creates AGENTS.md, CLAUDE.md, the canonical folder tree (.claude/, .agents/, .conductor/, .cursor/, .devcontainer/, .github/, .husky/, Docs/, plans/, findings/, progress/), .gitignore, .editorconfig, and stub configs. Idempotent - safe to re-run.

Sprint Loop step 01 - 'What should we build?' Run office hours to brainstorm and shape an idea into a problem statement, premise challenge, and explored alternatives. Triggers on 'office hours', 'brainstorm', 'shape an idea', 'new product', 'new feature idea', 'should we build', 'is this worth building', 'explore the problem', 'design direction', 'explore alternatives', 'premise challenge', 'what should we build'. Output is a design doc in plans/YYYY-MM-DD-slug.md with sections for Problem, Trigger, Alternatives explored, Recommended shape, Constraints, and Open questions.

Sprint Loop step 02 - 'How big could it be?' CEO-mode strategic and scope review of an active plan. Triggers on 'review the plan', 'scope review', 'strategic review', 'is this scope right', 'what is the blast radius', 'core review', 'CEO review', 'challenge the plan', 'poke holes', 'think bigger', 'scope down', 'lock the scope', 'what is not in scope'. Output is an updated plan file in plans/ with locked scope, NOT-in-scope section, blast radius map, and decision log.

Sprint Loop step 03 - 'Make the user happy.' Designer's-eye review of an active plan for UI and UX surfaces. Triggers on 'design review', 'UX review', 'designer review', 'review the design', 'is this UX good', 'make the user happy', 'UI critique', 'UX critique', 'rate the design', 'design dimensions', 'design rating', 'design plan review'. Output is an updated plan in plans/ with scored design dimensions, gap analysis, and concrete fixes per dimension.

MUST USE whenever the user starts a new project, scaffolds a repo, creates a new feature, module, service, or package, or creates any `.py`, `.ts`, `.tsx`, `.js`, `.jsx`, `.java`, `.cpp`, `.hpp`, `.cc`, `.h`, or `.sql` file. MUST USE when the user mentions Next.js, FastAPI, Django, Flask, React, Python, TypeScript, Node, Express, NestJS, Hono, Java, Spring, Spring Boot, C++, CMake, Postgres, Prisma, Drizzle, SQLAlchemy, App Router, Server Components, AI SDK, migrations, schema design, or indexing. MUST USE when the user says "best practices", "production ready", "structure this properly", "make this modular", "follow standards", "set up project", "scaffold", "init repo", "bootstrap", "lay this out", "organize this code", "split this file", "review this code", "audit this", "is this clean", "SOLID", "OOP", "object-oriented", "design pattern", "inheritance", "composition", "polymorphism", "encapsulation", "abstraction", "SRP", "OCP", "LSP", "ISP", "DIP", "DRY", "KISS", "YAGNI", "Law of Demeter", "dependency injection", or "should I extend or compose". MUST USE before generating boilerplate, before suggesting a folder layout, before writing the first module of any new repo, and before any refactor that touches file structure. This is Yash Singh's personal opinionated standard - apply silently while coding, do not lecture the user about the rules. Triggers on agent-governance, AGENTS.md, plans/findings/progress/skills folders, .claude, .conductor, .cursor, .devcontainer, .husky, and any request to "make this an agentic repo". If in doubt, load this skill - the cost of missing it is wrong defaults that the user will reject.