claude-code-blueprint

A battle-tested reference architecture for Claude Code power users. Not a plugin to install — a blueprint to learn from and adapt.

This repository documents a production Claude Code setup built over numerous sessions of real development work. Every agent, skill, hook, and rule exists because a real incident taught us it was needed.

This is NOT a generic starter kit. It's a reference architecture showing how a power user configures Claude Code for maximum productivity, with the reasoning behind every decision.

New to Claude Code? Start with GETTING-STARTED.md — a beginner-friendly walkthrough covering the CLI, MCP servers, plugins, and your first 30 minutes.

| Component | Count | Purpose | |-----------|-------|---------| | **Agents** | 11 | Specialized subagents with model tiering (opus/sonnet/haiku) | | **Skills** | 17 | Natural-language-triggered workflows (no slash commands needed) | | **Hooks** | 10 | Deterministic lifecycle automation (10 hook events) | | **Rules** | 5 | Path-scoped behavioral constraints | | **Memory System** | Dual | Auto-memory + external git-backed persistence | | **CLAUDE.md** | Template | Battle-tested behavioral rules | | **Settings** | Template | Full hook + permission configuration |

1. Hooks for enforcement, CLAUDE.md for guidance — Hooks fire 100% of the time. CLAUDE.md instructions are followed ~80%. If something MUST happen, make it a hook.

1. Agent-scoped knowledge, not global bloat — Design principles live in the frontend agent, not in every session's context. Security patterns live in the security-reviewer, not in CLAUDE.md.

1. Context is currency — Every token loaded into context is a token not available for your code. Keep MEMORY.md under 100 lines. Extract to topic files. Use path-scoped rules so