The Ultimate Agent Harness Performance Optimization Kit
Supercharge your AI coding agents with rules, skills, commands, and hooks.
Built for Claude Code � Cursor � OpenCode � Codex
Quick Start � What's Inside � Requirements � Installation � Which Agent? � FAQ
Most AI coding agents are generic. everything-claude gives your AI the context, standards, and capabilities to produce production-ready code � every time.
| Problem | Solution |
|---|---|
| AI ignores your coding standards | Rules � enforce them every session |
| Repetitive project setup | Skills � reusable, domain-specific capabilities |
| Manual workflows waste time | Commands � /build, /review, /ship, /launch |
| Generic AI for every task | Agents � specialized personas per project type |
| No quality gates | Hooks � automated checks at key moments |
1. Install
# macOS / Linux
curl -fsSL https://raw.githubusercontent.com/Elomami1976/everything-claude/main/install.sh | bash# Windows (PowerShell)
irm https://raw.githubusercontent.com/Elomami1976/everything-claude/main/install.ps1 | iex2. Open your AI assistant (Claude Code, Cursor, Windsurf...)
3. Start building
/build A Chrome extension that blocks social media during work hours
Your AI now has the rules, skills, and context to ship production code.
More commands to try
/review src/main.ts # Review code by severity
/audit # Scan for security issues
/plan Add Stripe billing # Get an implementation plan
/ship # Run pre-launch checklist
/launch # Generate Product Hunt kit
60+ files across 8 categories:
everything-claude/
+-- rules/ (13 files) Coding standards by language
+-- skills/ (9 files) Domain-specific capabilities
+-- commands/ (6 files) Slash command definitions
+-- agents/ (3 files) Specialized agent personas
+-- hooks/ (1 file) Lifecycle automation
+-- docs/ (4 files) Extended documentation
+-- templates/ (14 files) Ready-to-copy project starters
+-- comparisons/ (3 files) How this compares to other tools
| Category | Files | Covers |
|---|---|---|
common/ |
4 | Security, git workflow, documentation, core patterns |
javascript/ |
3 | Modern JS, Manifest V3 extensions, SaaS patterns |
typescript/ |
2 | Strict types, React hooks, component patterns |
python/ |
3 | Type hints, FastAPI, CLI scripts |
| Skill | What it does |
|---|---|
web-tool-builder |
Privacy-first single-HTML tools � no server, no upload |
chrome-extension-builder |
Manifest V3 + BYOK monetization |
saas-builder |
Next.js + Supabase + Stripe full stack |
api-connector |
Third-party API integration patterns |
code-review |
Severity-based review: Critical / High / Medium / Low |
security-audit |
Secrets, XSS, SQLi, CORS, CVE scanning |
seo-optimizer |
Technical SEO + GEO for AI search engines |
product-hunt-launch |
Tagline, copy, social posts, launch schedule |
monetization-planner |
Pricing tiers, BYOK models, revenue strategy |
| Command | Input | Output |
|---|---|---|
/build |
Product description | Tech stack + file structure + build steps |
/review |
File or directory | Issues by severity with code fixes |
/audit |
Current project | Security report + remediation steps |
/plan |
Feature request | Phased plan + risks + acceptance criteria |
/ship |
Current project | Pass/fail checklist across 6 categories |
/launch |
Current project | Full Product Hunt kit + social posts |
| Agent | Specialization |
|---|---|
chrome-ext-agent |
Chrome extensions with BYOK monetization |
saas-agent |
Full-stack SaaS with auth + payments |
web-tool-agent |
Privacy-first single-file browser tools |
| Hook | Trigger | Action |
|---|---|---|
PreToolUse |
Before file write | Block hardcoded secrets |
PostToolUse |
After file write | Auto-lint + format |
SessionStart |
Session begins | Detect project type, load context |
Stop |
Session ends | Generate summary, remind to commit |
| Template | Description | Time to First Run |
|---|---|---|
chrome-extension/ |
Manifest V3 popup + service worker | 5 minutes |
web-tool/ |
Privacy-first single-file tool | 2 minutes |
saas/ |
Next.js + Supabase + Stripe | 30 minutes |
See templates/README.md for usage.
| File | Topic |
|---|---|
| vs-cursor-native-rules.md | everything-claude vs Cursor .cursorrules |
| vs-github-copilot-instructions.md | everything-claude vs Copilot instructions |
| vs-chatgpt-prompts.md | everything-claude vs ChatGPT custom instructions |
| Requirement | Details |
|---|---|
| AI Tool | Claude Code, Cursor, Windsurf, OpenCode, Codex, or any assistant that reads markdown context |
| Git | For installation via clone |
| Node.js | v18+ � optional, only needed if you want hooks to auto-run linters |
| OS | macOS, Linux, Windows (WSL or PowerShell) |
No Node.js? You can still use all rules, skills, commands, and agents. Node.js is only needed for hook-triggered formatters.
# macOS / Linux
curl -fsSL https://raw.githubusercontent.com/Elomami1976/everything-claude/main/install.sh | bash# Windows
irm https://raw.githubusercontent.com/Elomami1976/everything-claude/main/install.ps1 | iexInstalls to ~/.claude/ and sets up rules, skills, commands, and agents globally.
git clone https://github.com/Elomami1976/everything-claude.git
cd everything-claude
./install.sh # macOS/Linux
.\install.ps1 # Windowscd my-project
git clone https://github.com/Elomami1976/everything-claude.git .claudeThen reference files in your prompts:
Follow the rules in .claude/rules/typescript/react.md
Use the skill at .claude/skills/saas-builder.md
cp -r rules/ ~/.claude/rules/
cp -r skills/ ~/.claude/skills/
cp -r commands/ ~/.claude/commands/
cp -r agents/ ~/.claude/agents/cat ~/.claude/rules/common/core.mdWhat are you building?
�
+-- A website or web app?
� +-- Needs server / database / auth? ------? saas-agent
� +-- Runs 100% client-side / offline? -----? web-tool-agent
�
+-- A browser extension? ----------------------? chrome-ext-agent
web-tool-agent |
chrome-ext-agent |
saas-agent |
|
|---|---|---|---|
| Server | No | No | Yes |
| Database | localStorage | chrome.storage | Supabase |
| Auth | None | Optional | Supabase Auth |
| Payments | None | Stripe one-time | Stripe subscriptions |
| Distribute via | URL / download | Chrome Web Store | Domain + hosting |
| Time to ship | Hours | Days | Weeks |
| What you are building | Use |
|---|---|
| PDF merger in the browser | web-tool-agent |
| Extension that summarizes articles with AI | chrome-ext-agent |
| Habit tracking app with user accounts | saas-agent |
| Color palette generator | web-tool-agent |
| Tab manager with cross-device sync | chrome-ext-agent |
| Invoice generator SaaS | saas-agent |
| Doc | Description |
|---|---|
| Installation Guide | All installation methods + troubleshooting |
| Skills Reference | Complete skill docs + examples |
| Commands Reference | Command usage, output formats, examples |
| Hooks Reference | Hook config + writing custom hooks |
| CONTRIBUTING.md | How to contribute rules, skills, and templates |
| CHANGELOG.md | Release history |
| ROADMAP.md | Planned features and direction |
Does this work with Cursor?
Yes. Copy rules/ to ~/.cursor/rules/, or generate a .cursorrules file:
cat rules/**/*.md > .cursorrulesDoes this work with GitHub Copilot / VS Code?
Yes. Copy CLAUDE.md to .github/copilot-instructions.md in your project.
What is the difference between a skill and an agent?
Skills are capabilities you call on demand:
Use the code-review skill to review auth.ts
Agents are full personas you activate for a whole session:
@saas-agent build a user dashboard
Skills are tools. Agents are specialists that carry multiple skills and behaviors.
Can I use only some parts?
Yes � everything is modular. Use only the rules you want, skip agents, ignore hooks. Each file works independently.
Do hooks run automatically?
Yes, in Claude Code. The hooks/hooks.json file is read by Claude Code's hook system. For other AI tools, hooks are advisory and can be manually invoked.
How do I add my own rules?
Create a .md file in the appropriate rules/ folder and reference it in your prompts or CLAUDE.md.
Will this slow down my AI assistant?
No. Rules and skills are only loaded when referenced. Hooks run in the background and do not affect response time.
Is this free?
Yes, 100% free and MIT licensed. Use it commercially, fork it, modify it.
Contributions are welcome. See CONTRIBUTING.md for the full guide.
Good first contributions:
- New language rules (Go, Rust, Swift, Kotlin)
- Framework skills (Vue, Django, Rails)
- New templates (React Native, Django, FastAPI)
git checkout -b feat/go-rules
git commit -m "feat: add Go language rules"
git push origin feat/go-rulesMIT � see LICENSE for details.
Built for indie developers who ship fast.
Report Bug �
Request Feature �
Discussions
If this helps you ship faster, give it a ?