Announcing the Conductor Model Context Protocol (MCP) Server

We are excited to launch the Conductor MCP server. This marks a significant advancement in our capabilities, providing seamless AI integration for your workflow orchestration.

Introducing the Conductor MCP server

If you’re new to MCP, start here: What is MCP?.

The Conductor MCP server is designed to be fully compatible with both Conductor OSS and Orkes Conductor. This means you get the same powerful features and performance regardless of your Conductor implementation. The Conductor MCP server empowers AI agents to create, run, and review workflows, making agentic systems more powerful and making Conductor:

“The Execution Arm of AI”

Key capabilities

With the Conductor MCP server, AI agents can access essential workflow operations, such as:

• Creating and executing workflows: Let AI do all of the heavy lifting of designing, publishing, and executing new workflows to automate complex tasks.
• Retrieving executions: Access detailed information and historical data for any workflow or task execution.
• Gathering task and workflow metadata: Collect comprehensive workflow and task-level metadata.

These capabilities let your AI agents dig into execution details and interact directly with your workflows, giving them real, hands-on control in production environments.

What your AI Agents can do with the Conductor MCP server

The Conductor MCP server offers numerous benefits, including:

• AI integration: Give your agents direct access to Conductor’s API surface, enabling them to create, execute, and manage workflows autonomously.
• Improved analysis: Use AI agents to help you create, understand, and troubleshoot your workflows and their executions.
• Seamless compatibility: Works flawlessly with both Conductor OSS and Orkes Conductor.
• Extensibility: Add new APIs by simply defining the endpoint path and a short description.

Conductor MCP server in action

Here’s what real agent-driven automation looks like using Claude and the Conductor MCP server:

1. A user asks Claude to create a workflow that sends an email with current stock picks.
2. Claude evaluates the available tools (exposed via MCP) and selects conductor to create a workflow.
3. Claude generates the workflow definition, explains the logic, and submits it to Conductor.
4. The workflow instantly appears in Orkes Conductor — exactly as created, without manual edits.

ℹ️ Tip: The quality and completeness of the workflow an agent generates depends entirely on how you phrase the prompt. A well-crafted prompt can lead to a fully executable workflow, while a vague one may leave gaps for you to fill in.

Getting started

Ready to give your AI agents real-world execution power?

Clone the repo and wire up the Conductor MCP server to your agent framework, such as Claude or Cursor.