claude-code-glm-suite/MASTER-PROMPT.md

🚀 Claude Code & GLM Suite

Master Integration Prompt

Complete installation guide with ALL sources, explanations, and examples

---

📑 Table of Contents

1. Before You Begin
2. How to Use
3. Copy-Paste Master Prompt
4. Real-Life Examples
5. Agent Reference
6. Configuration Files
7. Quick Reference
8. Troubleshooting

---

⚠️ Before You Begin

Choose Your API Provider

Option A: Z.AI / GLM Coding Plan (90% Cheaper) ⭐

Recommended for cost savings

# Install Z.AI Coding Helper
npm install -g @z_ai/coding-helper

# Run the setup wizard
npx @z_ai/coding-helper init

# Start Claude Code
claude

📖 Documentation: https://docs.z.ai/devpack/tool/claude

---

Option B: Anthropic Claude (Official)

# Get API key from: https://console.anthropic.com/

# Configure Claude Code
mkdir -p ~/.claude
cat > ~/.claude/settings.json << 'EOF'
{
  "env": {
    "ANTHROPIC_AUTH_TOKEN": "sk-ant-your-key-here",
    "ANTHROPIC_BASE_URL": "https://api.anthropic.com"
  }
}
EOF

# Install Claude Code
npm install -g @anthropic-ai/claude-code

# Start
claude

---

📖 How to Use the Master Prompt

Quick Start

1. Configure your API (choose Z.AI or Anthropic above)
2. Start Claude Code: claude
3. Copy the prompt below (one continuous block)
4. Paste into Claude Code
5. Follow the step-by-step installation

---

📋 Copy the Master Prompt

✂️ COPY FROM HERE (One Continuous Block)

═══════════════════════════════════════════════════════════════════════════════
                    CLAUDE CODE & GLM SUITE - MASTER PROMPT
═══════════════════════════════════════════════════════════════════════════════

I want to install the complete Claude Code & GLM Customization Suite with ALL integrations including agents, MCP tools, code analysis, and automation.

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

📦 OVERVIEW: What We're Installing

This suite includes 7 major components from open-source projects:

1. contains-studio/agents → 38 specialized AI agents with auto-triggering
2. @z_ai/mcp-server → 8 vision and analysis tools
3. @z_ai/coding-helper → Web search, GitHub integration
4. llm-tldr → Token-efficient code analysis (95% reduction)
5. ui-ux-pro-max → Professional UI/UX design guidance
6. @iannuttall/ralph → Autonomous agent looping (optional)
7. sync-agents.sh → Automated agent updates

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

📥 STEP 1: Install Contains Studio Agents (38 agents)

Source: https://github.com/contains-studio/agents

Clone and install:
git clone https://github.com/contains-studio/agents.git /tmp/contains-studio-agents
cp -r /tmp/contains-studio-agents/agents/* ~/.claude/agents/

What you get:
• 37 specialized agents across 8 departments
• 7 PROACTIVELY auto-triggering coordinators
• Automatic multi-agent coordination

Departments:
├── Engineering (7): AI Engineer, Backend, DevOps, Frontend, Mobile, Prototyper, Tester
├── Marketing (7): TikTok, Growth Hacker, Content, Instagram, Reddit, Twitter, ASO
├── Design (6): Brand, UI Designer, UX Research, Visual Storyteller, Whimsy, UI/UX Pro Max
├── Product (3): Feedback, Sprint Prioritizer, Trend Researcher
├── Project Management (3): Experiment Tracker, Project Shipper, Studio Producer
├── Studio Operations (5): Analytics, Finance, Infrastructure, Legal, Support
├── Testing (5): API Tester, Performance, Test Results, Tool Evaluator, Workflow
└── Bonus (3): Studio Coach, Agent Updater, Joker

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

📥 STEP 1.5: Install UI/UX Pro Max Skill

Source: https://github.rommark.dev/admin/claude-code-glm-suite

Install the skill:
mkdir -p ~/.claude/skills
cd claude-code-glm-suite
cp -r skills/ui-ux-pro-max ~/.claude/skills/
chmod +x ~/.claude/skills/ui-ux-pro-max/scripts/*.py

What you get:
• /ui-ux-pro-max slash command for design guidance
• Search 50+ design patterns (minimalism, glassmorphism, etc.)
• Color schemes, typography pairs, component examples
• WCAG 2.1 accessibility compliance

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

📥 STEP 1.75: Install Agent Sync Script

Source: https://github.rommark.dev/admin/claude-code-glm-suite

Download and install:
wget -O ~/.claude/sync-agents.sh https://raw.githubusercontent.com/github.rommark.dev/admin/claude-code-glm-suite/main/sync-agents.sh
chmod +x ~/.claude/sync-agents.sh

What you get:
• Automatic agent updates from official repository
• Timestamped backups before updating
• Preserves your customizations
• Optional git push to your Gitea repository

Usage:
~/.claude/sync-agents.sh

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

📥 STEP 2: Install MCP Vision & Analysis Tools

Source: https://github.com/zai-ai/mcp-server

Install:
npm install -g @z_ai/mcp-server

Verify:
npx @z_ai/mcp-server --help

What you get (8 tools):
• analyze_image → Understand any image
• analyze_video → Analyze video content
• ui_to_artifact → Convert UI screenshots to code
• extract_text → OCR text extraction
• diagnose_error → Understand errors from screenshots
• ui_diff_check → Compare two UI screenshots
• analyze_data_viz → Extract insights from charts
• understand_diagram → Understand technical diagrams

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

📥 STEP 3: Install MCP Web & GitHub Tools

Source: https://github.com/fcakyon/claude-codex-settings (inspired by)

Install:
npm install -g @z_ai/coding-helper

Verify:
npx @z_ai/coding-helper --help

What you get (3 tools):
• web-search-prime → AI-optimized web search
• web-reader → Convert web pages to markdown
• zread → Read and analyze GitHub repositories

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

📥 STEP 4: Install TLDR (Token-Efficient Code Analysis)

Source: https://github.com/parcadei/llm-tldr

Install:
pip install llm-tldr
tldr warm .

What you get:
• 95% token reduction for large codebases
• 5-layer analysis (AST, Call Graph, CFG, DFG, PDG)
• 18 MCP tools for code analysis
• Semantic search by behavior
• Impact analysis for safe refactoring

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

⚙️ STEP 5: Configure MCP Servers

Create MCP configuration:
mkdir -p ~/.claude
cat > ~/.claude/claude_desktop_config.json << 'EOF'
{
  "mcpServers": {
    "tldr": {
      "command": "tldr-mcp",
      "args": ["--project", "."]
    },
    "zai-vision": {
      "command": "npx",
      "args": ["@z_ai/mcp-server"]
    },
    "zai-web": {
      "command": "npx",
      "args": ["@z_ai/coding-helper"]
    }
  }
}
EOF

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

⚙️ STEP 6: Configure Ralph CLI (Optional - Advanced)

Source: https://github.com/iannuttall/ralph

⚠️ SKIP THIS STEP if you want simple setup
✅ COMPLETE THIS STEP for autonomous infinite looping

Install Ralph:
npm install -g @iannuttall/ralph

Create auto-trigger hook:
mkdir -p ~/.claude/hooks
cat > ~/.claude/hooks/ralph-auto-trigger.sh << 'HOOK_SCRIPT'
#!/bin/bash
# Ralph Auto-Trigger Hook
# Modes: "always" | "agents" (default) | "off"
# Background: Runs non-blocking, logs to ~/.claude/ralph-output.log

CLAUDE_DIR="$HOME/.claude"
RALPH_STATE_FILE="$CLAUDE_DIR/ralph-loop.local.md"
RALPH_PID_FILE="$CLAUDE_DIR/ralph.pid"
RALPH_LOCK_FILE="$CLAUDE_DIR/ralph.lock"
RALPH_LOG_FILE="$CLAUDE_DIR/ralph-output.log"

HOOK_INPUT=$(cat)
USER_PROMPT=$(echo "$HOOK_INPUT" | jq -r '.prompt // empty' 2>/dev/null || echo "${1:-}")

RALPH_AUTO_MODE="${RALPH_AUTO_MODE:-agents}"
RALPH_MAX_ITERATIONS="${RALPH_MAX_ITERATIONS:-50}"

if [[ "$RALPH_AUTO_MODE" == "off" ]]; then
    exit 0
fi

if [[ -f "$RALPH_LOCK_FILE" ]]; then
    LOCK_PID=$(cat "$RALPH_LOCK_FILE" 2>/dev/null || echo "")
    if [[ -n "$LOCK_PID" ]] && kill -0 "$LOCK_PID" 2>/dev/null; then
        exit 0
    else
        rm -f "$RALPH_LOCK_FILE" "$RALPH_PID_FILE"
    fi
fi

should_trigger=false

if [[ "$RALPH_AUTO_MODE" == "always" ]]; then
    should_trigger=true
elif [[ "$RALPH_AUTO_MODE" == "agents" ]]; then
    if echo "$USER_PROMPT" | grep -qiE "build|create|implement|develop|fix|add|refactor|optimize|write|generate|delegate|autonomous|agent|engineer|developer|designer|strategist"; then
        should_trigger=true
    fi
fi

if [[ "$should_trigger" == true ]]; then
    mkdir -p "$CLAUDE_DIR"

    cat > "$RALPH_STATE_FILE" << STATE
# Ralph Loop State - Auto-Triggered
# Generated: $(date -u +"%Y-%m-%d %H:%M:%S UTC")

**User Request:**
$USER_PROMPT

**Mode:** $RALPH_AUTO_MODE
**Max Iterations:** $RALPH_MAX_ITERATIONS
**Timestamp:** $(date -Iseconds)

## Usage

Monitor Ralph:
  tail -f ~/.claude/ralph-output.log

Stop Ralph:
  kill $(cat ~/.claude/ralph.pid)
  rm ~/.claude/ralph.lock

STATE

    if command -v ralph &> /dev/null; then
        touch "$RALPH_LOG_FILE"
        echo "[$(date -u +"%Y-%m-%d %H:%M:%S UTC")] Starting Ralph..." >> "$RALPH_LOG_FILE"
        echo "Mode: $RALPH_AUTO_MODE" >> "$RALPH_LOG_FILE"
        echo "---" >> "$RALPH_LOG_FILE"

        nohup ralph build "$RALPH_MAX_ITERATIONS" >> "$RALPH_LOG_FILE" 2>&1 &
        RALPH_PID=$!

        echo "$RALPH_PID" > "$RALPH_PID_FILE"
        echo "$RALPH_PID" > "$RALPH_LOCK_FILE"

        echo "🔄 Ralph auto-started (PID: $RALPH_PID)" >&2
        echo "   Monitor: tail -f ~/.claude/ralph-output.log" >&2
    else
        echo "⚠️  Ralph CLI not installed" >&2
        echo "   Install: npm install -g @iannuttall/ralph" >&2
    fi
fi

exit 0
HOOK_SCRIPT

chmod +x ~/.claude/hooks/ralph-auto-trigger.sh

Configure hooks:
cat > ~/.claude/hooks.json << 'EOF'
{
  "description": "Ralph auto-trigger hook",
  "hooks": {
    "UserPromptSubmit": [
      {
        "hooks": [
          {
            "type": "command",
            "command": "/home/uroma/.claude/hooks/ralph-auto-trigger.sh",
            "timeout": 5
          }
        ]
      }
    ]
  }
}
EOF

Note: Replace "/home/uroma" with your actual home directory

Control Modes:
export RALPH_AUTO_MODE="agents"  # Default
export RALPH_AUTO_MODE="always"  # Everything
export RALPH_AUTO_MODE="off"     # Disabled
export RALPH_MAX_ITERATIONS="100"

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

✅ STEP 7: Verify Installation

Run checks:
ls -la ~/.claude/agents/
find ~/.claude/agents -name "*.md" | wc -l

npx @z_ai/mcp-server --help
npx @z_ai/coding-helper --help
tldr --help

tldr status .
ralph --version 2>/dev/null || echo "Ralph optional"

ls -la ~/.claude/hooks/ralph-auto-trigger.sh 2>/dev/null || echo "Ralph hook optional"

Expected results:
✅ 39+ agent files (38 agents + README)
✅ All 8 departments present
✅ MCP tools accessible
✅ TLDR ready
✅ Settings configured
⚠️ Ralph (optional)

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

📊 SUMMARY: What You Now Have

┌─────────────────────────────────────────────────────────────┐
│  COMPONENTS INSTALLED                                       │
├─────────────────────────────────────────────────────────────┤
│  • 40+ AI Agents (8 departments)                           │
│  • 15+ MCP Tools (vision, web, code analysis)             │
│  • PROACTIVELY Auto-Triggering (7 coordinators)           │
│  • Ralph CLI (optional autonomous looping)                │
│  • Agent Sync Script (keep everything updated)            │
└─────────────────────────────────────────────────────────────┘

Benefits:
• Token Efficiency: 95% reduction via TLDR
• Code Understanding: 5-layer analysis
• Visual Capabilities: Screenshot-to-code
• Automation: Auto-tests, auto-UI, auto-coordination
• Current Info: Real-time web search

Please execute this complete integration step by step and confirm when each component is installed.

═══════════════════════════════════════════════════════════════════════════════

✂️ COPY ENDS HERE

---

💡 Real-Life Examples

Note: These are illustrative examples showing the conceptual difference between auto-triggering and manual workflows.

Example 1: Adding OAuth

❌ Without Auto-Trigger

You: Add OAuth to my app
Claude: [Writes code, no tests]
You: [Manually write tests later]

✅ With Auto-Trigger

You: Add OAuth to my app
Claude: [Writes code]
[test-writer-fixer PROACTIVELY triggers]
→ Tests written automatically

---

Example 2: Creating a Spinner

❌ Without Auto-Trigger

You: Create a loading spinner
Claude: [Basic spinner]
You: [Manually request improvements]

✅ With Auto-Trigger

You: Create a loading spinner
Claude: [Creates spinner]
[whimsy-injector PROACTIVELY triggers]
→ Adds bounce animation + messages

---

Example 3: Launching a Feature

❌ Without Auto-Trigger

You: We're launching next week
Claude: [Unclear what to prepare]
You: [Manual launch checklist]

✅ With Auto-Trigger

You: We're launching next week
[project-shipper PROACTIVELY triggers]
→ Complete launch plan created

---

📚 Agent Departments

Engineering (7)

• ai-engineer → ML/LLM integration
• backend-architect → API design
• devops-automator → CI/CD
• frontend-developer → React/Vue/Angular
• mobile-app-builder → iOS/Android
• rapid-prototyper → Quick MVPs
• test-writer-fixer → Auto tests ⚡

Marketing (7)

• tiktok-strategist → Viral TikTok
• growth-hacker → Growth loops
• content-creator → Multi-platform
• instagram-curator → Visual strategy
• reddit-builder → Community
• twitter-engager → Twitter growth
• app-store-optimizer → ASO

Design (6)

• brand-guardian → Brand consistency
• ui-designer → Interface design
• ux-researcher → User research
• visual-storyteller → Visual narratives
• whimsy-injector → Delight ⚡
• ui-ux-pro-max → Professional UI/UX ⚡

Product (3)

• feedback-synthesizer → User feedback
• sprint-prioritizer → Feature priority
• trend-researcher → Market trends

Project Management (3)

• experiment-tracker → A/B tests ⚡
• project-shipper → Launch coordination ⚡
• studio-producer → Team coordination ⚡

Studio Operations (5)

• analytics-reporter → Data insights
• finance-tracker → Financial tracking
• infrastructure-maintainer → System maintenance
• legal-compliance-checker → Legal compliance
• support-responder → Customer support

Testing (5)

• api-tester → API testing
• performance-benchmarker → Performance
• test-results-analyzer → Test analysis
• tool-evaluator → Tool assessment
• workflow-optimizer → Process improvement

Bonus (3)

• studio-coach → Agent coordination ⚡
• agent-updater → Agent updates
• joker → Humor injection

⚡ = PROACTIVELY auto-triggers

---

⚙️ Configuration Files

~/.claude/settings.json

Main Claude Code configuration:

{
  "env": {
    "ANTHROPIC_AUTH_TOKEN": "your-api-key",
    "ANTHROPIC_BASE_URL": "https://api.anthropic.com",
    "API_TIMEOUT_MS": "3000000"
  }
}

For Z.AI / GLM:

{
  "env": {
    "ANTHROPIC_AUTH_TOKEN": "your-zai-key",
    "ANTHROPIC_BASE_URL": "https://api.z.ai/api/anthropic",
    "ANTHROPIC_DEFAULT_HAIKU_MODEL": "glm-4.5-air",
    "ANTHROPIC_DEFAULT_SONNET_MODEL": "glm-4.7",
    "ANTHROPIC_DEFAULT_OPUS_MODEL": "glm-4.7"
  }
}

~/.claude/claude_desktop_config.json

MCP server configuration:

{
  "mcpServers": {
    "tldr": {
      "command": "tldr-mcp",
      "args": ["--project", "."]
    },
    "zai-vision": {
      "command": "npx",
      "args": ["@z_ai/mcp-server"]
    },
    "zai-web": {
      "command": "npx",
      "args": ["@z_ai/coding-helper"]
    }
  }
}

---

🎯 Quick Reference

Common Commands

# Sync agents (get latest updates)
~/.claude/sync-agents.sh

# Check TLDR status
tldr status .

# Verify Claude Code
claude --version

# Monitor Ralph (if installed)
tail -f ~/.claude/ralph-output.log

# Stop Ralph
kill $(cat ~/.claude/ralph.pid)

Environment Variables

# Auto-push to Gitea after sync
export GITEA_REPO_URL="https://github.rommark.dev/admin/claude-code-glm-suite.git"

# Ralph control
export RALPH_AUTO_MODE="agents"    # Options: always, agents, off
export RALPH_MAX_ITERATIONS="50"   # Max loop iterations

---

🔧 Troubleshooting

Agents not triggering?

• Check ~/.claude/agents/ directory exists
• Verify agent files are present
• Restart Claude Code

MCP tools missing?

• Check ~/.claude/claude_desktop_config.json
• Verify tools are installed: npm list -g
• Restart Claude Code

TLDR not working?

• Run tldr warm . in project directory
• Check Python is installed
• Verify tldr-mcp is in PATH

Ralph not starting?

• Check hook is executable: ls -la ~/.claude/hooks/
• Verify mode: echo $RALPH_AUTO_MODE
• Check log: cat ~/.claude/ralph-output.log

---

📖 Resources

Official Documentation

• Claude Code: https://docs.anthropic.com/en/docs/claude-code
• Z.AI / GLM: https://docs.z.ai/devpack/tool/claude
• contains-studio/agents: https://github.com/contains-studio/agents
• @z_ai/mcp-server: https://github.com/zai-ai/mcp-server
• llm-tldr: https://github.com/parcadei/llm-tldr

Repository

• Gitea: https://github.rommark.dev/admin/claude-code-glm-suite

---

Last Updated: 2025-01-16 Version: 3.0 Maintained By: rommark.dev

---

For updates, check the repository regularly and run ~/.claude/sync-agents.sh