MCPInstall

everything-claude-code-harness

Agent harness performance system for Claude Code and other AI coding agents — skills, instincts, memory, hooks, commands, and security scanning

Installs
1.8K
Last 24h
—
Trending rank
—
Security
Not audited

Install

Works with every agent the skills CLI supports.

npx skills add https://github.com/reason-machines/trending-skills --skill everything-claude-code-harness
  1. 01

    Install

    Run the command in your project. The CLI asks which agents to install for.

  2. 02

    Describe the task

    Ask your agent for something this skill covers. It loads everything-claude-code-harness automatically when relevant.

  3. 03

    Review the result

    The agent follows the SKILL.md steps, scripts and references bundled with the skill.

SKILL.md

The instructions your agent reads when the skill is triggered.

Everything Claude Code (ECC) — Agent Harness Performance System

Skill by ara.so — Daily 2026 Skills collection.

Everything Claude Code (ECC) is a production-ready performance optimization system for AI agent harnesses. It provides specialized subagents, reusable skills, custom slash commands, memory-persisting hooks, security scanning, and language-specific rules — all evolved from 10+ months of daily real-world use. Works across Claude Code, Cursor, Codex, OpenCode, and Antigravity.


Installation

Option 1: Plugin Marketplace (Recommended)

# Inside Claude Code, run:
/plugin marketplace add affaan-m/everything-claude-code
/plugin install everything-claude-code@everything-claude-code

Option 2: Manual Clone

git clone https://github.com/affaan-m/everything-claude-code.git
cd everything-claude-code

# Install rules for your language stack
./install.sh typescript
# Multiple languages:
./install.sh typescript python golang swift
# Target a specific IDE:
./install.sh --target cursor typescript

Install Rules (Always Required)

Claude Code plugins cannot auto-distribute rules — install them manually via ./install.sh or copy from rules/ into your project's .claude/rules/ directory.


Directory Structure

everything-claude-code/
├── .claude-plugin/         # Plugin and marketplace manifests
│   ├── plugin.json
│   └── marketplace.json
├── agents/                 # Specialized subagents (planner, architect, etc.)
├── commands/               # Slash commands (/plan, /security-scan, etc.)
├── skills/                 # Reusable skill modules
├── hooks/                  # Lifecycle hooks (SessionStart, Stop, PostEdit, etc.)
├── rules/
│   ├── common/             # Language-agnostic rules
│   ├── typescript/
│   ├── python/
│   ├── golang/
│   └── swift/
├── scripts/                # Setup and utility scripts
└── install.sh              # Interactive installer

Key Commands

After installation, use the namespaced form (plugin install) or short form (manual install):

# Planning & architecture
/everything-claude-code:plan "Add OAuth2 login flow"
/everything-claude-code:architect "Design a multi-tenant SaaS system"

# Research-first development
/everything-claude-code:research "Best approach for rate limiting in Node.js"

# Security
/everything-claude-code:security-scan
/everything-claude-code:harness-audit

# Agent loops and orchestration
/everything-claude-code:loop-start
/everything-claude-code:loop-status
/everything-claude-code:quality-gate
/everything-claude-code:model-route

# Multi-agent workflows
/everything-claude-code:multi-plan
/everything-claude-code:multi-execute
/everything-claude-code:multi-backend
/everything-claude-code:multi-frontend

# Session and memory
/everything-claude-code:sessions
/everything-claude-code:instinct-import

# PM2 orchestration
/everything-claude-code:pm2

# Package manager setup
/everything-claude-code:setup-pm

With manual install, drop the everything-claude-code: prefix: /plan, /sessions, etc.


Hook Runtime Controls

ECC hooks fire at agent lifecycle events. Control strictness at runtime without editing files:

# Set hook strictness profile
export ECC_HOOK_PROFILE=minimal    # Least intrusive
export ECC_HOOK_PROFILE=standard   # Default
export ECC_HOOK_PROFILE=strict     # Maximum enforcement

# Disable specific hooks by ID (comma-separated)
export ECC_DISABLED_HOOKS="pre:bash:tmux-reminder,post:edit:typecheck"

Hook events covered: SessionStart, Stop, PostEdit, PreBash, PostBash, and more.


Package Manager Detection

ECC auto-detects your package manager with this priority chain:

  1. CLAUDE_PACKAGE_MANAGER environment variable
  2. .claude/package-manager.json (project-level)
  3. package.json → packageManager field
  4. Lock file detection (package-lock.json, yarn.lock, pnpm-lock.yaml, bun.lockb)
  5. ~/.claude/package-manager.json (global)
  6. First available manager as fallback
# Set via environment
export CLAUDE_PACKAGE_MANAGER=pnpm

# Set globally
node scripts/setup-package-manager.js --global pnpm

# Set per-project
node scripts/setup-package-manager.js --project bun

# Detect current setting
node scripts/setup-package-manager.js --detect

Skills System

Skills are markdown modules the agent loads to gain domain expertise. Install individually or in bulk.

Using a Skill

# Reference a skill explicitly in your prompt
"Use the search-first skill to find the right caching approach before implementing"

# Or trigger via slash command
/everything-claude-code:research "content hashing strategies for API responses"

Notable Built-in Skills

SkillPurpose
search-firstResearch before coding — avoids hallucinated APIs
cost-aware-llm-pipelineOptimizes token spend across model calls
content-hash-cache-patternCache invalidation via content hashing
skill-stocktakeAudits which skills are loaded and active
frontend-slidesZero-dependency HTML presentation builder
configure-eccGuided interactive ECC setup wizard
swift-actor-persistenceSwift concurrency + persistence patterns
regex-vs-llm-structured-textDecides when to use regex vs LLM parsing

Writing a Custom Skill

Create skills/my-skill.md:

---
name: my-skill
description: What this skill does
triggers:
  - "phrase that activates this skill"
---

# My Skill

## When to Use
...

## Pattern
\`\`\`typescript
// concrete example
\`\`\`

## Rules
- Rule one
- Rule two

Instincts System (Continuous Learning)

Instincts are session-extracted patterns saved for reuse. They carry confidence scores and evolve over time.

Export an Instinct

/everything-claude-code:instinct-import

Instinct File Format

---
name: prefer-zod-for-validation
confidence: 0.92
extracted_from: session-2026-02-14
---

# Action
Always use Zod for runtime schema validation in TypeScript projects.

# Evidence
Caught 3 runtime type errors that TypeScript alone missed during session.

# Examples
\`\`\`typescript
import { z } from 'zod'

const UserSchema = z.object({
  id: z.string().uuid(),
  email: z.string().email(),
  role: z.enum(['admin', 'user'])
})

type User = z.infer<typeof UserSchema>
\`\`\`

Rules Architecture

Rules enforce coding standards per language. Install only what your stack needs.

# TypeScript + Python
./install.sh typescript python

# Check what's installed
ls .claude/rules/

Rule Directory Layout

rules/
├── common/         # Applies to all languages
│   ├── research-first.md
│   ├── security-baseline.md
│   └── verification-loops.md
├── typescript/
│   ├── no-any.md
│   ├── zod-validation.md
│   └── strict-mode.md
├── python/
│   ├── type-hints.md
│   └── django-patterns.md
└── golang/
    └── error-wrapping.md

Agents (Subagent Delegation)

Agents are specialized personas the orchestrator delegates to:

# In your prompt, reference an agent explicitly
"Delegate architecture decisions to the architect agent"
"Use the planner agent to break this feature into tasks"

Available agents include: planner, architect, researcher, verifier, security-auditor, and more. Each lives in agents/<name>.md with its own system prompt, tools list, and constraints.


AgentShield Security Scanning

Run security scans directly from Claude Code:

/everything-claude-code:security-scan

This invokes the AgentShield scanner (1282 tests, 102 rules) against your codebase and surfaces:

  • Hardcoded secrets
  • Injection vulnerabilities
  • Insecure dependencies
  • Agent prompt injection patterns

Memory Persistence Hooks

ECC hooks automatically save and restore session context:

// hooks/session-start.js — loads prior context on new session
const fs = require('fs')
const path = require('path')

const memoryPath = path.join(process.env.HOME, '.claude', 'session-memory.json')

if (fs.existsSync(memoryPath)) {
  const memory = JSON.parse(fs.readFileSync(memoryPath, 'utf8'))
  console.log('Restored session context:', memory.summary)
}
// hooks/stop.js — saves session summary on exit
const summary = {
  timestamp: new Date().toISOString(),
  summary: process.env.ECC_SESSION_SUMMARY || '',
  skills_used: (process.env.ECC_SKILLS_USED || '').split(',')
}

fs.writeFileSync(memoryPath, JSON.stringify(summary, null, 2))

Cross-Platform Support

PlatformSupport
Claude CodeFull (agents, commands, skills, hooks, rules)
CursorFull (via --target cursor installer flag)
OpenCodeFull (plugin system, 20+ hook event types, 3 native tools)
Codex CLIFull (codex.md generated via /codex-setup)
Codex AppFull (AGENTS.md-based)
AntigravityFull (via --target antigravity installer flag)

Common Patterns

Research-First Development

"Before implementing the payment webhook handler, use the search-first skill to 
verify current Stripe webhook verification best practices."

Token Optimization

# Route to cheaper model for simple tasks
/everything-claude-code:model-route "Write a unit test for this pure function"

# Use background processes for long analysis
/everything-claude-code:harness-audit

Parallelization with Git Worktrees

# Create isolated worktrees for parallel agent tasks
git worktree add ../feature-auth -b feature/auth
git worktree add ../feature-payments -b feature/payments

# Each Claude Code session operates in its own worktree
# Merge when both complete

Verification Loop

/everything-claude-code:loop-start    # Begin tracked loop
# ... agent does work ...
/everything-claude-code:loop-status   # Check progress
/everything-claude-code:quality-gate  # Enforce pass criteria before merge

Troubleshooting

Plugin commands not found after install

/plugin list everything-claude-code@everything-claude-code
# If empty, re-run: /plugin install everything-claude-code@everything-claude-code

Rules not applied

# Rules require manual install — plugin system cannot distribute them
cd everything-claude-code && ./install.sh typescript
# Verify:
ls ~/.claude/rules/   # or .claude/rules/ in project root

Hooks not firing

# Check profile setting
echo $ECC_HOOK_PROFILE
# Check disabled list
echo $ECC_DISABLED_HOOKS
# Reset to defaults
unset ECC_HOOK_PROFILE
unset ECC_DISABLED_HOOKS

Instinct import drops content Ensure you're on v1.4.1+. Earlier versions had a bug where parse_instinct_file() silently dropped Action/Evidence/Examples sections. Pull latest and re-run.

Wrong package manager used

node scripts/setup-package-manager.js --detect
export CLAUDE_PACKAGE_MANAGER=pnpm   # Override explicitly

Resources

FAQ

What does the everything-claude-code-harness skill do?

Agent harness performance system for Claude Code and other AI coding agents — skills, instincts, memory, hooks, commands, and security scanning

How do I install everything-claude-code-harness?

Run npx skills add https://github.com/reason-machines/trending-skills --skill everything-claude-code-harness in your project. The skills CLI detects the agents you use and copies the skill into the right folder. For Claude Code you can also place the skill folder at ~/.claude/skills/everything-claude-code-harness/ (personal) or .claude/skills/everything-claude-code-harness/ (project).

Which AI agents can use everything-claude-code-harness?

Any agent that supports the open Agent Skills format (a folder with a SKILL.md file), including Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, Windsurf, Cline, OpenCode and more.

Is everything-claude-code-harness safe to install?

No partner security audit is available yet. Skills can run commands and scripts, so read the SKILL.md and any bundled files before installing.

Who maintains everything-claude-code-harness?

It is published by reason-machines in reason-machines/trending-skills. It has 1,831 recorded installs.

More from reason-machines/trending-skills

View all →

openclaw-control-center

reason-machines/trending-skills

Local-first, security-first control center for OpenClaw agents — visibility dashboard with readonly defaults, token attribution, collaboration tracing, and safe write operations.

Data & Databases4.4K installs

inkos-multi-agent-novel-writing

reason-machines/trending-skills

Multi-agent CLI system for autonomous novel writing, auditing, and revision with human review gates

Docs & Writing2.8K installs

claude-hud-statusline

reason-machines/trending-skills

A Claude Code plugin that displays a real-time HUD showing context usage, active tools, running agents, and todo progress in your terminal statusline.

understand-anything-knowledge-graph

reason-machines/trending-skills

Turn any codebase into an interactive knowledge graph using Claude Code skills — explore, search, and ask questions about any project visually.

agency-agents-ai-specialists

reason-machines/trending-skills

A collection of specialized AI agent personalities for Claude Code, Cursor, Aider, Windsurf, and other AI coding tools — covering engineering, design, marketing, sales, and more.

AI & Agents2.1K installs

agent-browser-automation

reason-machines/trending-skills

Headless browser automation CLI for AI agents using native Rust binary with Chrome DevTools Protocol

Popular code quality & workflow skills

View category →

grill-me

mattpocock/skills

A relentless interview to sharpen a plan or design.

Scan a codebase for deepening opportunities, present them as a visual HTML report, then grill through whichever one you pick.

handoff

mattpocock/skills

Compact the current conversation into a handoff document for another agent to pick up.

domain-modeling

mattpocock/skills

Build and sharpen a project's domain model. Use when discussing codebase terminology, writing or editing a GLOSSARY.md, or recording or editing an ADR.

codebase-design

mattpocock/skills

Shared vocabulary for designing deep modules. Use when the user wants to design or improve a module's interface, find deepening opportunities, decide where a seam goes, make code more testable or AI-navigable, or when another skill needs the deep-module vocabulary.

diagnosing-bugs

mattpocock/skills

Diagnosis loop for hard bugs and performance regressions. Use when the user says "diagnose"/"debug this", or reports something broken/throwing/failing/slow.