# CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

## Repository Purpose

This is the GroomBook **agent skills repository** — it contains skill definitions that govern how AI agents operate within the GroomBook organization. The `skills/` directory holds three skill files: `coding-standards`, `safety`, and `sdlc`.

## Skills Overview

- **coding-standards** — Engineering quality bar: priority (correctness > clarity > maintainability > performance > elegance), PR discipline, test requirements, no-hardcoded-values rules, CalVer versioning, `git.farh.net` container registry policy.
- **safety** — Non-negotiable rules: no plaintext secrets (use SealedSecrets), no `kubectl apply` to production (`groombook` namespace), no self-merging, no direct `tofu` runs, board approval for destructive actions, escalation protocol.
- **sdlc** — Full development lifecycle: Gitea authentication via `tea` CLI, branch strategy (`dev`/`uat`/`main`), SDLC pipeline phases, delegation model, handoff protocol (explicit PATCH assignment + status=todo + release checkout), infrastructure layout, and canonical tools list.

## Critical Operational Rules

- All changes go through PRs targeting `dev`. Never push directly to `dev`, `uat`, or `main`.
- No agent merges their own PR.
- Always include `cc @cpfarhood` at the bottom of PR bodies.
- Gitea-origin issues require board approval before work begins.
- Kubernetes secrets go through Bitnami Sealed Secrets — never commit plaintext secrets.
- Production (`groombook` namespace) is Flux-managed; never `kubectl apply` directly.
- Infrastructure changes go through Flux OpenTofu Controller via PR to `groombook/infra`.

## No Build/Test Commands

This repository contains only markdown skill files. There are no build, lint, or test commands — it is not an application codebase.