Claude Code Project Configuration Showcase

Most software engineers are seriously sleeping on how good LLM agents are right now, especially something like Claude Code.

Once you've got Claude Code set up, you can point it at your codebase, have it learn your conventions, pull in best practices, and refine everything until it's basically operating like a super-powered teammate. The real unlock is building a solid set of reusable "skills" plus a few "agents" for the stuff you do all the time.

What This Looks Like in Practice

Custom UI Library? We have a skill that explains exactly how to use it. Same for how we write tests, how we structure GraphQL, and basically how we want everything done in our repo. So when Claude generates code, it already matches our patterns and standards out of the box.

Automated Quality Gates? We use hooks to auto-format code, run tests when test files change, type-check TypeScript, and even block edits on the main branch. Claude Code also created a bunch of ESLint automation, including custom rules and lint checks that catch issues before they hit review.

Deep Code Review? We have a code review agent that Claude runs after changes are made. It follows a detailed checklist covering TypeScript strict mode, error handling, loading states, mutation patterns, and more. When a PR goes up, we have a GitHub Action that does a full PR review automatically.

Scheduled Maintenance? We've got GitHub workflow agents that run on a schedule:

• Monthly docs sync - Reads commits from the last month and makes sure docs are still aligned
• Weekly code quality - Reviews random directories and auto-fixes issues
• Biweekly dependency audit - Safe dependency updates with test verification

Intelligent Skill Suggestions? We built a skill evaluation system that analyzes every prompt and automatically suggests which skills Claude should activate based on keywords, file paths, and intent patterns.

A ton of maintenance and quality work is just... automated. It runs ridiculously smoothly.

JIRA/Linear Integration? We connect Claude Code to our ticket system via MCP servers. Now Claude can read the ticket, understand the requirements, implement the feature, update the ticket status, and even create new tickets if it finds bugs along the way. The /ticket command handles the entire workflow—from reading acceptance criteria to linking the PR back to the ticket.

We even use Claude Code for ticket triage. It reads the ticket, digs into the codebase, and leaves a comment with what it thinks should be done. So when an engineer picks it up, they're basically starting halfway through already.

There is so much low-hanging fruit here that it honestly blows my mind people aren't all over it.

---

Table of Contents

• Directory Structure
• Quick Start
• Configuration Reference
  • CLAUDE.md - Project Memory
  • settings.json - Hooks & Environment
  • MCP Servers - External Integrations
  • LSP Servers - Real-Time Code Intelligence
  • Skill Evaluation Hooks
  • Skills - Domain Knowledge
  • Agents - Specialized Assistants
  • Commands - Slash Commands
• GitHub Actions Workflows
• Best Practices
• Examples in This Repository

---

Directory Structure

your-project/
├── CLAUDE.md                      # Project memory (alternative location)
├── .mcp.json                      # MCP server configuration (JIRA, GitHub, etc.)
├── .claude/
│   ├── settings.json              # Hooks, environment, permissions
│   ├── settings.local.json        # Personal overrides (gitignored)
│   ├── settings.md                # Human-readable hook documentation
│   ├── .gitignore                 # Ignore local/personal files
│   │
│   ├── agents/                    # Custom AI agents
│   │   └── code-reviewer.md       # Proactive code review agent
│   │
│   ├── commands/                  # Slash commands (/command-name)
│   │   ├── onboard.md             # Deep task exploration
│   │   ├── pr-review.md           # PR review workflow
│   │   └── ...
│   │
│   ├── hooks/                     # Hook scripts
│   │   ├── skill-eval.sh          # Skill matching on prompt submit
│   │   ├── skill-eval.js          # Node.js skill matching engine
│   │   └── skill-rules.json       # Pattern matching configuration
│   │
│   ├── skills/                    # Domain knowledge documents
│   │   ├── README.md              # Skills overview
│   │   ├── testing-patterns/
│   │   │   └── SKILL.md
│   │   ├── graphql-schema/
│   │   │   └── SKILL.md
│   │   └── ...
│   │
│   └── rules/                     # Modular instructions (optional)
│       ├── code-style.md
│       └── security.md
│
└── .github/
    └── workflows/
        ├── pr-claude-code-review.yml           # Auto PR review
        ├── scheduled-claude-code-docs-sync.yml # Monthly docs sync
        ├── scheduled-claude-code-quality.yml   # Weekly quality review
        └── scheduled-claude-code-dependency-audit.yml

---

Quick Start

1. Create the .claude directory

mkdir -p .claude/{agents,commands,hooks,skills}

2. Add a CLAUDE.md file

Create CLAUDE.md in your project root with your project's key information. See CLAUDE.md for a complete example.

# Project Name

## Quick Facts
- **Stack**: React, TypeScript, Node.js
- **Test Command**: `npm run test`
- **Lint Command**: `npm run lint`

## Key Directories
- `src/components/` - React components
- `src/api/` - API layer
- `tests/` - Test files

## Code Style
- TypeScript strict mode
- Prefer interfaces over types
- No `any` - use `unknown`

3. Add settings.json with hooks

Create .claude/settings.json. See settings.json for a full example with auto-formatting, testing, and more.

{
  "hooks": {
    "PreToolUse": [
      {
        "matcher": "Edit|Write",
        "hooks": [
          {
            "type": "command",
            "command": "[ \"$(git branch --show-current)\" != \"main\" ] || { echo '{\"block\": true, \"message\": \"Cannot edit on main branch\"}' >&2; exit 2; }",
            "timeout": 5
          }
        ]
      }
    ]
  }
}

4. Add your first skill

Create .claude/skills/testing-patterns/SKILL.md. See testing-patterns/SKILL.md for a comprehensive example.

---
name: testing-patterns
description: Jest testing patterns for this project. Use when writing tests, creating mocks, or following TDD workflow.
---

# Testing Patterns

## Test Structure
- Use `describe` blocks for grouping
- Use `it` for individual tests
- Follow AAA pattern: Arrange, Act, Assert

## Mocking
- Use factory functions: `getMockUser(overrides)`
- Mock external dependencies, not internal modules

Tip: The description field is critical—Claude uses it to decide when to apply the skill. Include keywords users would naturally mention.

---

Configuration Reference

CLAUDE.md - Project Memory

CLAUDE.md is Claude's persistent memory that loads automatically at session start.

Locations (in order of precedence):

1. .claude/CLAUDE.md (project, in .claude folder)
2. ./CLAUDE.md (project root)
3. ~/.claude/CLAUDE.md (user-level, all projects)

What to include:

• Project stack and architecture overview
• Key commands (test, build, lint, deploy)
• Code style guidelines
• Important directories and their purposes
• Critical rules and constraints

📄 Example: CLAUDE.md

---

settings.json - Hooks & Environment

The main configuration file for hooks, environment variables, and permissions.

Location: .claude/settings.json

📄 Example: settings.json | Human-readable docs

Hook Events

| Event | When It Fires | Use Case | |-------|---------------|----------| | PreToolUse | Before tool execution | Block edits on main, validate commands | | PostToolUse | After tool completes | Auto-format, run tests, lint | | UserPromptSubmit | User submits prompt | Add context, suggest skills | | Stop | Agent finishes | Decide if Claude should continue |

Hook Response Format

{
  "block": true,           // Block the action (PreToolUse only)
  "message": "Reason",     // Message to show user
  "feedback": "Info",      // Non-blocking feedback
  "suppressOutput": true,  // Hide command output
  "continue": false        // Whether to continue
}

Exit Codes

• 0 - Success
• 2 - Blocking error (PreToolUse only, blocks the tool)
• Other - Non-blocking error

---

MCP Servers - External Integrations

MCP (Model Context Protocol) servers let Claude Code connect to external tools like JIRA, GitHub, Slack, databases, and more. This is how you enable workflows like "read a ticket, implement it, and update the ticket status."

Location: .mcp.json (project root, committed to git for team sharing)

📄 Example: .mcp.json

How MCP Works

┌──