everything-claude-code

Language: English | 繁體中文

Everything Claude Code

English | 简体中文

The complete collection of Claude Code configs from an Anthropic hackathon winner.

Production-ready agents, skills, hooks, commands, rules, and MCP configurations evolved over 10+ months of intensive daily use building real products.

---

Why Everything Claude Code?

Problems Solved

Pain Point	Our Solution
Context window exhaustion	Strategic compaction, memory persistence hooks - sessions last hours, not minutes
Repetitive work	30+ pre-configured commands and skills for one-click workflows
Inconsistent AI responses	Rules system ensures consistent code style, security compliance, test coverage
High token costs	Subagent architecture + model selection strategy reduces costs by 50-70%
Cross-session amnesia	Continuous learning auto-extracts patterns, applies them in future sessions
Tedious manual setup	One-click plugin install for complete configuration

Who Is This For?

• Daily Claude Code users: Save setup time, focus on coding
• Team tech leads: Standardize AI-assisted development workflows
• Efficiency-focused engineers: Boost AI coding productivity 2-5x
• Multi-language developers: Supports TypeScript, Go, C++, Qt, Python, and more

Efficiency Gains

Based on real-world usage:

Metric	Improvement
Effective coding time per session	3-5x longer
Repetitive task time	80% reduction
Code review coverage	From manual 30% to AI+manual 90%
Token consumption	50% less with mgrep + subagents

---

5-Minute Quick Start

Step 1: Install Plugin (30 seconds)

# In Claude Code
/plugin marketplace add affaan-m/everything-claude-code
/plugin install everything-claude-code@everything-claude-code

Step 2: Install Rules (1 minute)

# Clone the repo
git clone https://github.com/affaan-m/everything-claude-code.git

# Install rules to user directory (global effect)
cp -r everything-claude-code/rules/* ~/.claude/rules/

Step 3: Verify Installation (30 seconds)

In Claude Code, type:

/tdd          # Should see TDD workflow start
/code-review  # Should see code review start

Step 4: Start Using (3 minutes to familiarize)

Top 5 Most Used Commands:

Command	Purpose	When to Use
/plan	Plan feature implementation	Before starting new features
/tdd	Test-driven development	Writing new features or fixing bugs
/code-review	Code review	Before commits
/build-fix	Fix build errors	When compilation fails
/refactor-clean	Clean dead code	After refactoring

---

Maximizing Efficiency

1. Delegate with Subagents

Don't make the main Claude do everything. Delegate to specialized subagents:

# Scenario: Implementing a new feature

1. Use planner agent to plan implementation steps
2. Use tdd-guide agent to write tests and code
3. Use code-reviewer agent to review code
4. Use security-reviewer agent to check security issues

Result: Each agent focuses on single responsibility, more accurate results, less context consumed.

2. Manage Context Window

Golden Rule: 200k context = precious resource, don't waste it

Strategy	Action
Disable unused MCPs	/mcp to check, keep <10 enabled
Use mgrep	50% token savings vs grep
Strategic compaction	Manual /compact after exploration
Modular code	Keep files 200-400 lines, max 800

3. Build Continuous Learning Loop

# Before session ends
/learn  # Extract patterns learned this session

# View learned instincts
/instinct-status

# Evolve instincts into formal skills
/evolve

Result: Claude learns from your coding habits, auto-applies next time.

4. Parallel Workflows

Scenario: Need to develop multiple features simultaneously

# Create Git worktrees
git worktree add ../feature-auth feature-auth
git worktree add ../feature-cache feature-cache

# Each worktree runs independent Claude instance
cd ../feature-auth && claude
cd ../feature-cache && claude

Result: Conflict-free parallel development, doubled efficiency.