MCP Conductor

MCP Conductor is an orchestration layer that coordinates multiple MCPs to provide instant project context for AI-assisted development. It caches comprehensive project intelligence and synchronizes memory, filesystem, git, and database insights so you can start productive work in seconds rather than minutes. This guide walks you through practical usage, installation steps, and important configuration notes to get you up and running quickly.

How to use

You will run a local MCP Conductor instance and connect your Claude Desktop client to it. Once connected, you can load project context on demand, using the Memory MCP to retrieve cached intelligence and the other MCPs to enrich the context with filesystem, git, and analytics data.

How to install

Prerequisites you need before installation are Node.js and npm, Claude Desktop installed and logged in, and internet access for dependencies. You will also identify your project directory and Claude workspace so MCP Conductor can auto-detect and work with your environment.

- Node.js 18+ (download from nodejs.org)
- npm 8+ (included with Node.js)
- Claude Desktop (download and install)
- Internet access for dependency downloads
- A known project directory (e.g., ~/Projects) and Claude workspace (e.g., ~/Projects/claude) set up on your system

Optional: enable test mode for development using MCP_TEST_MODE environment variable set to true

Configuration and MCP connections

You can configure MCP Conductor connections to run locally as standard input/output (stdio) servers. The following configurations show how to start the local MCPs and expose an IPC-based interface that Claude Desktop can connect to.

{
  "mcpServers": {
    "conversation_continuity": {
      "command": "node",
      "args": ["/path/to/mcp-conductor/dist/index.js"],
      "env": {
        "MCP_CONDUCTOR_PROJECT_DIR": "/Users/YourName/Projects",
        "MCP_CONDUCTOR_WORKSPACE": "/Users/YourName/Projects/claude"
      }
    },
    "memory": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-memory"]
    },
    "filesystem": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-filesystem", "/Users/YourName/Projects"]
    }
  }
}

Usage patterns with the MCP client

• Start a daily development session by loading project intelligence from Memory MCP to initialize instant context for your current project.

• Load context for bug investigations or feature development by targeting the appropriate project intelligence cache and enabling workflow rules for compliance and review.

Additional sections

Environment variables you may set to tailor behavior include MCP_CONDUCTOR_PROJECT_DIR, MCP_CONDUCTOR_WORKSPACE, and optional LOG_LEVEL for logging verbosity. The project directory is auto-detected when possible, and Claude workspace is auto-detected or provided via configuration.

Security notes: when enabling local stdio servers, ensure your local machine is secured and access to the Claude Desktop application is restricted to authorized users. Use strong access controls and consider containerizing or sandboxing to limit exposure of local MCP processes.

Troubleshooting tips: verify Node.js and npm versions meet the minimums, confirm Claude Desktop is logged in, and ensure the MCP Conductor process has access to your project directories for read/write operations.

Available tools

create_project_intelligence_cache

Create the project intelligence cache to accelerate future sessions.

Load ProjectIntelligence_MyProject

Load instant context for MyProject from Memory MCP for immediate productive work.

Load ProjectIntelligence_TeamProject

Load team project context to coordinate work like handoffs and task ownership.

Load ProjectIntelligence_ProductionApp

Load production context for urgent production issue investigation.