home / mcp / claude code mcp enhanced mcp server
Enhanced Claude Code MCP server with orchestration capabilities, reliability improvements, and self-contained execution patterns
Configuration
View docs{
"mcpServers": {
"grahama1970-claude-code-mcp-enhanced": {
"command": "npx",
"args": [
"github:grahama1970/claude-code-mcp-enhanced"
],
"env": {
"MCP_CLAUDE_DEBUG": "false",
"MCP_EXECUTION_TIMEOUT_MS": "1800000",
"MCP_HEARTBEAT_INTERVAL_MS": "15000"
}
}
}
}This MCP server lets Claude Code run in a controlled, orchestrated way with permissions bypass and robust task management. It enables you to execute complex, multi-step coding tasks, delegate subtasks, monitor progress, and retry on transient errors, all while keeping execution reliable and auditable.
You can connect an MCP client to this server to perform coding tasks, orchestration, and file operations. Use the Claude Code MCP tool to send prompts, specify a working directory, and control task flow. The system supports parent tasks and subtasks, heartbeat progress updates, and optional Roo modes to tailor behavior for coding, design, or other workflows.
Typical usage patterns include starting with a broad task description and letting the system break it into subtasks, then guiding execution through the boomerang pattern where Claude Code returns results to a parent task. You can enable auto-retries for transient issues, observe live progress through periodic heartbeats, and adjust return granularity with returnMode (summary or full).
Choose one installation method that fits your setup. Each method provides a ready-to-use MCP server configuration you can reference in your client.
Method 1: Via GitHub URL (Recommended) creates a dynamic, always-updated installation by pulling from the repository.
Method 2: Via npm Package uses the published package to install the MCP server.
Method 3: Local Installation builds and runs from a local checkout for development or testing.
{
"mcpServers": {
"claude-code-mcp-enhanced": {
"command": "npx",
"args": [
"github:grahama1970/claude-code-mcp-enhanced"
],
"env": {
"MCP_CLAUDE_DEBUG": "false",
"MCP_HEARTBEAT_INTERVAL_MS": "15000",
"MCP_EXECUTION_TIMEOUT_MS": "1800000"
}
}
}
}{
"mcpServers": {
"claude-code-mcp-enhanced": {
"command": "npx",
"args": [
"-y",
"@grahama1970/claude-code-mcp-enhanced@latest"
],
"env": {
"MCP_CLAUDE_DEBUG": "false",
"MCP_HEARTBEAT_INTERVAL_MS": "15000",
"MCP_EXECUTION_TIMEOUT_MS": "1800000"
}
}
}
}For development or testing, you can run the server from a local installation after building the project.
# Clone the repository
git clone https://github.com/grahama1970/claude-code-mcp-enhanced.git
cd claude-code-mcp-enhanced
# Install dependencies and build
npm install
npm run build
# Start from local dist
# Configure your MCP client to use the local serverExecutes a prompt using the Claude Code CLI with permissions bypass enabled and supports optional workFolder, parentTaskId, returnMode, and mode controls.
Returns health status, version, and current configuration of the Claude Code MCP server.
Converts markdown task files into Claude Code MCP-compatible JSON for execution.