Skip to content

Everything Claude Code (ECC)

What it is

Everything Claude Code (ECC) is a comprehensive performance optimization system designed for AI agent harnesses, primarily Claude Code. It is not just a configuration pack but a complete ecosystem of specialized agents, skills, hooks, and rules evolved from intensive daily production use.

What problem it solves

It bridges the gap between a raw AI CLI and a production-ready autonomous engineering environment. ECC addresses context window management, security risks, memory persistence across sessions, and language-specific coding standards through automated enforcement and optimized prompt engineering. It is specifically tuned to maximize the reasoning efficiency of frontier models like claude-4-8-opus-20260528 and GPT-5.5 using MCP 3.0.

Where it fits in the stack

Category: AI Assistants & Knowledge / Developer Tooling

Typical use cases

  • Autonomous Engineering: Leveraging specialized subagents (Architect, Planner, TDD Guide) for complex feature implementation.
  • Security Auditing: Using the integrated AgentShield to scan configurations for secrets and injection risks.
  • Continuous Learning: Automatically extracting patterns from development sessions into reusable "instincts" and skills.
  • Multi-Agent Orchestration: Managing complex workflows across multiple services using PM2 and collaborative planning commands.

Strengths

  • Comprehensive Ecosystem: Offers 48+ agents and 182+ skills covering 10+ programming languages.
  • Cross-Platform & Harness: Supports Claude Code, Cursor, OpenCode, Codex, and Antigravity with unified rules.
  • Security-First: Includes AgentShield for adversarial reasoning and secret detection.
  • Optimized Performance: Pre-configured token optimization settings (e.g., MAX_THINKING_TOKENS) to reduce costs while maintaining high-quality reasoning.

Limitations

  • Configuration Overhead: Requires manual rule installation as Claude Code plugins cannot distribute files to the filesystem.
  • Context Usage: Large numbers of enabled MCP servers or skills can rapidly consume the LLM's context window.
  • Maintenance: Keeping 180+ skills updated requires active community participation.

Key Components

  • Subagents (48+): Specialized personas like typescript-reviewer, sql-auditor, and build-error-resolver for delegation.
  • Skills (182+): Domain-specific workflow definitions ranging from frontend-slides to market-research.
  • Hooks Runtime: Trigger-based automations that fire on tool events (e.g., auto-formatting after an edit or secret detection before prompt submission).
  • Rules (34+): Standardized, language-specific guidelines (TypeScript, Python, Go, Swift, PHP) that ensure consistency across the codebase.

AgentShield — Security Auditor

A dedicated auditor (v2.0 as of June 2026) that scans Claude Code configurations (.claude/ directory) for vulnerabilities. It uses adversarial reasoning (Red Team/Blue Team agents) to evaluate protection layers and synthesize risk assessments.

Skill Creator

Analyzes local Git history to automatically generate SKILL.md files and instinct collections, allowing the agent to "learn" from the existing codebase patterns.

When to use it

  • When building complex production applications with AI agents.
  • When requiring consistent coding standards and hooks across a distributed engineering team.
  • When seeking an "all-in-one" performance system for AI agent harnesses.

When not to use it

  • For trivial, single-file scripts where a standard prompt is sufficient.
  • If you prefer a completely "vanilla" AI experience without automated hooks or specialized subagents.

Getting started

Installation (Plugin)

# Add the marketplace
/plugin marketplace add https://github.com/affaan-m/everything-claude-code

# Install the plugin
/plugin install everything-claude-code@everything-claude-code

Installation (Manual)

git clone https://github.com/affaan-m/everything-claude-code.git
cd everything-claude-code
# Deploy agents to your local configuration
cp agents/*.md ~/.claude/agents/

CLI examples

Audit Configurations with AgentShield

/plugin run ecc:agentshield --path .claude/

Create a New Skill from History

/plugin run ecc:skill-creator --since "3 days ago" --name "new-feature-pattern"

List Active Subagents

/plugin run ecc:list-agents

API examples

Configuring Subagent Delegation (JSON)

ECC allows defining delegation rules in a central agents.json file. 

{
  "delegation_rules": {
    "security_audit": "ecc:agentshield",
    "ui_review": "ecc:frontend-reviewer",
    "backend_refactor": "ecc:architect"
  }
}

Custom Hook Trigger (JavaScript)

// .claude/hooks/post-edit.js
module.exports = async ({ file, content }) => {
  if (file.endsWith('.ts')) {
    await run('npm run lint -- --fix ' + file);
    console.log(`ECC: Linted ${file}`);
  }
};

Sources / references

Contribution Metadata

  • Last reviewed: 2026-06-28
  • Confidence: high