MCP server for Netflix Conductor Workflow Engine - enables AI assistants to interact with Conductor workflows for troubleshooting, creation, and management.
This Model Context Protocol (MCP) server provides a comprehensive interface to Netflix Conductor, allowing AI assistants like Claude to help developers:
- Troubleshoot workflows: Get detailed status, view execution history, analyze failed tasks
- Manage workflow executions: Start, pause, resume, terminate, restart, and retry workflows
- Work with workflow definitions: Create, update, and query workflow definitions
- Manage task definitions: Create, update, and query task definitions
- Search and query: Advanced search capabilities across workflow executions
npm install -g conductor-mcpOr install locally:
npm install conductor-mcpThe MCP server connects to your Conductor instance using environment variables:
CONDUCTOR_SERVER_URL: Base URL of your Conductor server (default:http://localhost:8080)CONDUCTOR_API_PATH: API path prefix (default:/api)
Add this to your Claude Desktop configuration file:
MacOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%/Claude/claude_desktop_config.json
{
"mcpServers": {
"conductor": {
"command": "conductor-mcp",
"env": {
"CONDUCTOR_SERVER_URL": "http://localhost:8080",
"CONDUCTOR_API_PATH": "/api"
}
}
}
}For a remote Conductor server:
{
"mcpServers": {
"conductor": {
"command": "conductor-mcp",
"env": {
"CONDUCTOR_SERVER_URL": "https://conductor.yourcompany.com",
"CONDUCTOR_API_PATH": "/api"
}
}
}
}List workflow executions with optional filters.
Parameters:
workflowName(optional): Filter by workflow name/typestatus(optional): Filter by status (RUNNING, COMPLETED, FAILED, TIMED_OUT, TERMINATED, PAUSED)startTime(optional): Filter workflows started after this time (epoch milliseconds)endTime(optional): Filter workflows started before this time (epoch milliseconds)freeText(optional): Free text search
Example:
List all failed workflows in the last 24 hours
Get detailed status of a specific workflow execution.
Parameters:
workflowId(required): The workflow execution IDincludeTaskDetails(optional): Include detailed task information (default: true)
Example:
Get status of workflow abc-123-def
Start a new workflow execution.
Parameters:
workflowName(required): Name of the workflow to startversion(optional): Version of the workflow (defaults to latest)input(optional): Input parameters as JSON objectcorrelationId(optional): Correlation ID for trackingpriority(optional): Priority 0-99 (default: 0)
Example:
Start a new order_processing workflow with input {"orderId": "12345", "customerId": "C789"}
Pause a running workflow.
Parameters:
workflowId(required): The workflow execution ID
Resume a paused workflow.
Parameters:
workflowId(required): The workflow execution ID
Terminate a workflow execution.
Parameters:
workflowId(required): The workflow execution IDreason(optional): Reason for termination
Restart a workflow from the beginning.
Parameters:
workflowId(required): The workflow execution IDuseLatestDefinition(optional): Use latest workflow definition (default: false)
Retry a failed workflow from the last failed task.
Parameters:
workflowId(required): The workflow execution IDresumeSubworkflowTasks(optional): Resume subworkflow tasks (default: false)
Advanced search for workflow executions.
Parameters:
query(required): Query string (e.g., 'workflowType=MyWorkflow AND status=FAILED')start(optional): Start index for pagination (default: 0)size(optional): Number of results (default: 100)sort(optional): Sort field and order (e.g., 'startTime:DESC')
Example:
Search for all failed payment workflows in the last week
Get a workflow definition by name and version.
Parameters:
workflowName(required): Name of the workflowversion(optional): Version (defaults to latest)
List all registered workflow definitions.
Parameters:
access(optional): Filter by access type (READ or CREATE)tagKey(optional): Filter by tag keytagValue(optional): Filter by tag value
Create or update a workflow definition.
Parameters:
definition(required): Complete workflow definition as JSON objectoverwrite(optional): Overwrite existing definition (default: true)
Get details of a specific task execution.
Parameters:
taskId(required): The task execution ID
Get execution logs for a specific task.
Parameters:
taskId(required): The task execution ID
Update the status of a task execution.
Parameters:
taskId(required): The task execution IDworkflowInstanceId(required): The workflow instance IDstatus(required): New status (IN_PROGRESS, FAILED, FAILED_WITH_TERMINAL_ERROR, COMPLETED)output(optional): Task output datalogs(optional): Task execution logs
Get a task definition by name.
Parameters:
taskName(required): Name of the task
List all registered task definitions.
Parameters:
access(optional): Filter by access type (READ or CREATE)
Create or update a task definition.
Parameters:
definition(required): Complete task definition as JSON object
Get event handlers.
Parameters:
event(optional): Filter by event nameactiveOnly(optional): Return only active handlers (default: true)
Here are some common scenarios you can ask Claude to help with:
"Show me all failed workflows in the last hour"
"Why did workflow abc-123 fail?"
"What tasks are currently running in workflow xyz-789?"
"Show me the execution history of the payment_processing workflow"
"Start a new order_processing workflow with orderId 12345"
"Pause workflow abc-123"
"Retry the failed workflow xyz-789"
"Terminate all stuck workflows"
"Show me the definition of the payment_processing workflow"
"Create a new workflow definition for customer onboarding"
"List all available task definitions"
"What workflows are registered in the system?"
git clone https://github.com/opensensor/conductor-mcp.git
cd conductor-mcp
npm install
npm run buildnpm run build
CONDUCTOR_SERVER_URL=http://localhost:8080 node build/index.jsnpm run watchThis MCP server uses:
- @modelcontextprotocol/sdk: MCP protocol implementation
- axios: HTTP client for Conductor API calls
- TypeScript: Type-safe development
The server runs on stdio transport, making it compatible with Claude Desktop and other MCP clients.
This server is compatible with Netflix Conductor v3.x API. It has been tested with:
- Conductor v3.15.0
- Conductor v3.13.0
Contributions are welcome! Please feel free to submit a Pull Request.
Apache-2.0
For issues and questions:
- GitHub Issues: https://github.com/opensensor/conductor-mcp/issues
- Conductor Documentation: https://conductor.netflix.com/
Built with the Model Context Protocol SDK and inspired by the Netflix Conductor workflow orchestration platform.