AI Champion — Command Cheatsheet

LVL 0 Foundation: AI Resources Repository

Fork or clone ai-kit as your personal AI tools repository

git clone https://github.com/urbanisierung/ai-kit ~/github.com/urbanisierung/ai-kit
# Or fork it first on GitHub, then clone your fork

Copy `.env.example` to `.env` and fill in your API keys

cp .env.example .env
# Edit .env — fill in:
# ANTHROPIC_API_KEY=sk-ant-...
# BRAVE_API_KEY=BSA_...
# XAI_API_KEY=xai-...

Run `bash tools/setup.sh` to wire all configs in one shot

bash tools/setup.sh

# What it does:
# → Links claude/CLAUDE.md.global  →  ~/.claude/CLAUDE.md
# → Links claude/skills/           →  ~/.claude/skills/

Push to a private remote repo and verify the setup reproduces from scratch

# On a second machine (or after wiping ~/.claude):
git clone [email protected]:yourname/ai-kit.git
cp ai-kit/.env.example ai-kit/.env  # fill in real keys
bash ai-kit/tools/setup.sh

LVL 1 Context Engineering

After each correction session, add one rule to your context file

# Example: you just corrected the agent for over-wrapping in try/catch

# 1. Ask the agent to encode the rule:
"Add a rule to CLAUDE.md: never wrap internal function calls in try/catch.
Only validate at system boundaries (user input, external APIs)."

LVL 2 Compounding Engineering

Install compound-engineering Claude

/plugin marketplace add EveryInc/compound-engineering-plugin
/plugin install compound-engineering

Install compound-engineering Gemini

bunx @every-env/compound-plugin install compound-engineering --to gemini

Install compound-engineering Copilot

bunx @every-env/compound-plugin install compound-engineering --to copilot

LVL 3 Skills and Extensions

Install skill-creator from skills.sh (meta-skill for writing skills) Claude
```
/plugin install skill-creator
```
Install Superpowers and run on a real task Claude
```
/plugin install superpowers@claude-plugins-official
```
Run harness audit from Everything Claude Code Claude
```
/harness-audit
```

LVL 4 MCP Servers

Add Brave Search MCP (get key at api.search.brave.com)

# Claude Code
claude mcp add brave-search -e BRAVE_API_KEY=BSA_YOUR_KEY -- npx -y @modelcontextprotocol/server-brave-search

# Gemini CLI — add to ~/.gemini/settings.json
# { "mcpServers": { "brave-search": { "command": "npx", "args": ["-y", "@modelcontextprotocol/server-brave-search"], "env": { "BRAVE_API_KEY": "BSA_..." } } } }

Add DeepWiki MCP — no API key required

{ "mcpServers": { "deepwiki": { "command": "npx", "args": ["-y", "@deepwiki/mcp"] } } }

Add Grok MCP for X/Twitter search (sign up at console.x.ai)

# Get the MCP from github.com/merterbak/Grok-MCP
# Add XAI_API_KEY to your .env

Add mem0 MCP for cross-session memory

# Cloud (easiest — free tier available)
npx mcp-add --name mem0-mcp --type http --url "https://mcp.mem0.ai/mcp"
claude mcp add mem0-mcp --scope global

# Add to .env:

LVL 5 Prompting Discipline

Use Plan Mode to iterate before any code is written Claude
```
# Press Shift+Tab in Claude Code to enter Plan Mode
```

LVL 6 Token Management and Observability

Install claude-hud for live context bar and subagent status Claude

/plugin marketplace add jarrodwatts/claude-hud
/plugin install claude-hud
/claude-hud:setup

Configure claude-hud to show agent and todo lines Claude
```
/claude-hud:configure
```

Install RTK (Rust Token Killer) — 60–90% token savings on shell output

# macOS
brew install rtk

# Linux
curl -fsSL https://raw.githubusercontent.com/rtk-ai/rtk/refs/heads/master/install.sh | sh

Activate RTK hook for Claude Code Claude
```
rtk init -g
```
Activate RTK hook for Gemini CLI Gemini
```
rtk init -g --gemini
```
Activate RTK hook for Copilot / Codex Copilot
```
rtk init -g --codex
```
Check RTK token savings after your next session
```
rtk gain
```

Install last30days skill Claude+Gemini

# Claude Code
/plugin marketplace add mvanhorn/last30days-skill
/plugin install last30days@last30days-skill

# Gemini CLI

Run last30days on a current topic and review the Polymarket section Claude+Gemini

/last30days <your topic>
/last30days <your topic> --quick   # faster, less depth
/last30days topic1 vs topic2       # comparative mode

LVL 7 Hooks and Persistent Memory

Add a PostToolUse hook that runs your linter after every file write Claude

// In claude/settings.json
{
  "hooks": {
    "PostToolUse": [
      { "command": "pnpm lint 2>&1 | tail -20" }

Install Rudel for session analytics Claude

npm install -g rudel
rudel login
rudel enable

Install pi-self-learning for automatic session memory capture Claude
```
npm install -g @pi-labs/cli
pi install npm:pi-self-learning
```
Review a month of captured learnings Claude
```
/learning-month
```

Set up mem0 for cross-session memory

# Cloud path
# 1. Sign up at app.mem0.ai, get API key
# 2. Add to .env: MEM0_API_KEY=m0-...
# 3. Add MCP:
npx mcp-add --name mem0-mcp --type http --url "https://mcp.mem0.ai/mcp"

Install jai — lightweight copy-on-write sandbox for agents (Linux) Claude
```
# Arch (AUR)
yay -S jai

# From source
git clone https://github.com/stanford-scs/jai.git
```
Run Claude inside the jai sandbox when testing new skills or hooks Claude
```
jai claude
```
Use strict jai mode before delegating long autonomous tasks Claude
```
jai --mode strict claude
```

LVL 8 Remote Setup

Install Node.js 22 and OpenCode on the server

# Install nvm
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.40.1/install.sh | bash
nvm install 22 && nvm use 22

# Install opencode

Start a persistent tmux session

tmux new-session -s main
# Detach: Ctrl+B D
# Reconnect: tmux attach -t main

Set up a systemd service for headless OpenCode

[Unit]
Description=OpenCode AI Agent
After=network.target

[Service]

Install Tailscale for private SSH access from any device

curl -fsSL https://tailscale.com/install.sh | sh
tailscale up

LVL 9 Parallel Agents

Enable experimental agent teams in Claude Code settings Claude
```
{ "CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS": true }
```

Deploy DeerFlow via Docker

git clone https://github.com/bytedance/deer-flow && cd deer-flow
make config    # generates config.yaml, fill in your model provider
docker compose up

Install Hermes Agent

curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash
source ~/.bashrc
hermes setup