refection1918/openai-mcp-server

OpenAI MCP Server - GPT-5 to Claude CLI Orchestration

This MCP server enables GPT-5 to act as a primary orchestrator/supervisor, directing Claude CLI as a secondary executor through standardized messaging protocols.

Architecture

GPT-5 (Primary/Orchestrator) 
    ↓ Strategic decisions & guidance
OpenAI MCP Server 
    ↓ Task execution commands
Claude CLI (Secondary/Executor)
    ↓ Results & status
OpenAI MCP Server
    ↓ Analysis & next steps
GPT-5 (Primary/Orchestrator)

Features

🤖 Dual AI Orchestration

• GPT-5: Strategic planning, quality control, decision-making
• Claude CLI: Task execution, detailed analysis, implementation

🛠 Core Tools

1. orchestrate_claude_task - Send tasks to Claude CLI with GPT-5 oversight
2. query_gpt5_guidance - Get strategic guidance from GPT-5
3. get_task_status - Monitor Claude CLI task execution
4. cancel_task - Cancel running tasks
5. list_active_tasks - View all active tasks

📊 Task Management

• Concurrent task limiting (configurable)
• Task timeout handling
• Real-time status tracking
• Error recovery with GPT-5 guidance

🔄 Intelligent Workflow

• Pre-execution guidance from GPT-5
• Post-execution analysis and recommendations
• Multi-step workflow coordination
• Quality control and optimization suggestions

Installation

cd /home/lkw/openai-mcp-server
npm install
npm run build

Configuration

Environment Variables

# OpenAI Configuration
export OPENAI_API_KEY="your-openai-api-key"
export OPENAI_MODEL="gpt-4"  # Will upgrade to gpt-5 when available

# Claude CLI Configuration  
export CLAUDE_COMMAND="claude"  # Path to Claude CLI

# Optional: Override defaults
export MAX_CONCURRENT_TASKS=3
export DEFAULT_TIMEOUT=300000
export LOG_LEVEL="info"

MCP Client Configuration

Add to your MCP client configuration:

{
  "mcpServers": {
    "openai-orchestrator": {
      "command": "node",
      "args": ["/home/lkw/openai-mcp-server/dist/index.js"],
      "env": {
        "OPENAI_API_KEY": "your-api-key"
      }
    }
  }
}

Usage Examples

1. Basic Task Orchestration

// GPT-5 directs Claude CLI to analyze a codebase
await mcp.call("orchestrate_claude_task", {
  task_description: "Analyze Python codebase for security vulnerabilities",
  claude_command: "claude analyze --security /path/to/codebase",
  context: {
    priority: "high",
    timeout: 600000
  }
});

2. Strategic Guidance

// Get GPT-5 guidance on complex decisions
await mcp.call("query_gpt5_guidance", {
  query: "Should we prioritize performance optimization or security fixes first?",
  context: {
    active_tasks: ["security-scan", "perf-analysis"],
    claude_status: "busy",
    previous_results: [...]
  }
});

3. Workflow Coordination

// Multi-step workflow with GPT-5 coordination
const analysisResult = await mcp.call("orchestrate_claude_task", {
  task_description: "Initial codebase analysis",
  claude_command: "claude analyze /project"
});

const guidance = await mcp.call("query_gpt5_guidance", {
  query: "Based on analysis results, what should be the next priority?",
  context: { analysis_result: analysisResult }
});

// Execute next step based on GPT-5 guidance
const nextTask = await mcp.call("orchestrate_claude_task", {
  task_description: guidance.recommended_action,
  claude_command: guidance.claude_command
});

Key Benefits

🎯 Strategic + Tactical Excellence

• GPT-5 provides high-level strategy and quality oversight
• Claude CLI handles detailed execution and technical implementation
• Best of both AI systems combined

⚡ Efficient Resource Management

• Concurrent task limiting prevents resource exhaustion
• Timeout handling ensures tasks don't hang
• Graceful error recovery with AI guidance

🔍 Enhanced Quality Control

• Pre-execution validation by GPT-5
• Post-execution analysis and optimization
• Continuous improvement suggestions

🔄 Seamless Integration

• Standard MCP protocol compatibility
• Easy integration with existing workflows
• Configurable for different use cases

Workflow Patterns

Pattern 1: Supervised Execution

GPT-5: "Analyze this codebase for security issues"
↓
MCP Server: Validates request, gets pre-execution guidance
↓
Claude CLI: Executes detailed security analysis
↓
MCP Server: Gets post-execution analysis from GPT-5
↓
Result: Comprehensive analysis + strategic recommendations

Pattern 2: Iterative Optimization

GPT-5: Initial task planning
↓
Claude CLI: Execute first iteration
↓
GPT-5: Analyze results, suggest improvements
↓
Claude CLI: Execute refined version
↓
Continue until GPT-5 approves quality

Pattern 3: Complex Project Management

GPT-5: Break down large project into phases
↓
Claude CLI: Execute Phase 1 tasks
↓
GPT-5: Review progress, adapt plan
↓
Claude CLI: Execute Phase 2 with adjustments
↓
Repeat with continuous optimization

Development

# Development mode with auto-rebuild
npm run dev

# Run tests
npm test

# Build for production
npm run build

# Start production server
npm start

Security Considerations

• OpenAI API key security
• Task execution sandboxing
• Input validation and sanitization
• Audit logging for task execution
• Resource usage monitoring

Limitations

• Requires valid OpenAI API key with GPT-4/GPT-5 access
• Claude CLI must be available and properly configured
• Network connectivity required for OpenAI API calls
• Task execution limited by Claude CLI capabilities

Future Enhancements

• Support for multiple Claude CLI instances
• Advanced workflow templates
• Real-time collaboration features
• Integration with more AI models
• Enhanced monitoring and analytics

Architecture Philosophy: This server implements a "Strategic AI + Tactical AI" pattern where GPT-5 handles high-level orchestration and decision-making while Claude CLI provides detailed execution and analysis capabilities. The combination leverages the strengths of both systems for superior outcomes.