# Orkes Conductor MCP > Orkes Conductor connects your agent directly to complex workflow engines, giving you full visibility into microservice processes. It lets you list entire workflow definitions, track running instances across services, and search through historical execution data. Stop opening dozens of dashboards; ask your AI client everything about your system's operational state in one go. ## Overview - **Category:** loved-by-devs - **Price:** Free - **Tags:** orchestration, microservices, workflow-engine, task-scheduling, execution-monitoring ## Description You're dealing with systems where a single user action triggers a dozen backend steps—payment processing, inventory updates, notifications. Tracking if that whole chain worked is usually a nightmare of clicking through different monitoring UIs. This MCP lets your AI client bypass the dashboards entirely. It connects directly to your orchestration layer, giving you immediate read access to workflow definitions, currently running instances, and the granular history of any task execution. Need to know why an order failed last week? You can search across all historical runs using powerful queries. Want to see if a process is stuck right now? You list active workflows by name and inspect their current state. If you're building complex agentic applications, connecting this MCP via Vinkius gives your agent the single source of truth it needs to debug or audit multi-step business logic without needing dedicated API calls for every piece of data. ## Tools ### list_workflow_defs Lists all registered overarching workflow definitions available in your system. ### get_workflow_def Retrieves the full definition schema for a specific named workflow. ### list_task_defs Lists all individual task definitions that can be used within your workflows. ### list_running Provides a list of currently active workflow instances, allowing you to monitor what's executing right now. ### get_execution Fetches detailed state information for one specific workflow execution run. ### search_workflows Performs an advanced search across all historical workflow executions using filters like status or ID. ## Prompt Examples **Prompt:** ``` Show me all registered workflow definitions. ``` **Response:** ``` Found 12 workflow definitions. 1) 'order-processing' v3 — 8 tasks, including payment validation and shipping triggers. 2) 'user-onboarding' v2 — 5 tasks with email verification fork. 3) 'data-pipeline-daily' v1 — 4 tasks for ETL orchestration. Want me to inspect the graph of any of these? ``` **Prompt:** ``` Are there any failed workflows in the last 24 hours? ``` **Response:** ``` Found 3 failed executions in the last 24 hours. 1) order-processing (ID: abc-123) — failed at 'payment-validation' task at 14:23 UTC. Error: 'Gateway timeout'. 2) user-onboarding (ID: def-456) — failed at 'send-welcome-email' at 16:45. Error: 'SMTP connection refused'. 3) data-pipeline-daily (ID: ghi-789) — timed out at 'extract-transform' at 03:12. Want me to get the full execution trace for any? ``` **Prompt:** ``` How many instances of the order-processing workflow are currently running? ``` **Response:** ``` 8 instances of 'order-processing' currently running. 5 are at the 'payment-validation' stage, 2 at 'shipping-trigger', and 1 at 'inventory-check'. Average execution time so far: 4.2 seconds. No instances are stuck or timed out. Want me to inspect one? ``` ## Capabilities ### View all defined workflows List and inspect every registered workflow definition, including their versions and underlying task structures. ### Find active process runs Get a list of currently running workflow instances by filtering them based on the name they belong to. ### Deeply inspect execution failures Retrieve detailed state information for any specific workflow run, showing task-by-task history and error messages. ### Search historical runs quickly Perform broad searches across all past workflow executions using customizable query filters like status or correlation ID. ## Use Cases ### The Order Failure Mystery A user asks their agent, 'Why did order #XYZ fail last Tuesday?' The agent runs search_workflows, filters by ID and date, and uses get_execution to pinpoint that the payment task timed out at 14:23 UTC. The fix is immediate. ### Debugging a New Feature A developer wants to know how the new 'premium user' path works. They ask the agent to list_workflow_defs and get_workflow_def for that specific workflow, inspecting the branching logic without touching the staging environment. ### Daily Operations Check The ops team needs to know if any critical ETL pipelines are running. They use list_running, filtering by 'data-pipeline', and instantly see 3 active instances currently processing data. ### Understanding Dependencies A new architect joins the project and asks, 'What processes rely on user onboarding?' The agent runs list_task_defs and inspects related workflow definitions to map out dependencies for them. ## Benefits - Stop guessing where a process broke. You use get_execution to pull deep-dive trace histories for any run, telling you exactly which task failed and why. - Audit complex systems faster than ever. The search_workflows tool lets you query months of historical data using filters like 'failed' or specific correlation IDs. - Understand your architecture without reading documentation. By listing all registered workflow definitions, architects can quickly map out the entire system flow. - Monitor real-time operations with minimal effort. list_running gives you a quick snapshot of every active instance for rapid operational checks. - Validate new services before deployment. You can inspect task definitions to ensure your microservice outputs match what the workflow expects. ## How It Works The bottom line is you get a conversational window into the guts of your distributed system, without needing any terminal commands or UI clicks. 1. Subscribe to the Orkes Conductor MCP and provide your required Access Key, Secret, and Base URL. 2. Your AI agent uses these credentials to authenticate against your orchestration cluster. 3. The agent then executes specific commands—like searching or listing definitions—and returns structured data directly to your chat interface. ## Frequently Asked Questions **How does Orkes Conductor MCP help me debug failed workflows?** It provides deep state details via get_execution. You don't just see 'failed'; you see the exact task that threw an exception and the error message, letting you know exactly what needs fixing. **Can Orkes Conductor MCP track running services in real time?** Yes. By using list_running, your agent pulls a live count of active instances for any specific workflow name. This is crucial for monitoring capacity and immediate status checks. **Does Orkes Conductor MCP help me map out my system architecture?** Absolutely. You can list_workflow_defs to see every major process, and get_workflow_def to inspect the task-by-task graphs, helping architects understand dependencies. **What if I need to check history from last month? Does Orkes Conductor MCP support that?** Yes. The search_workflows tool allows you to perform powerful searches across all historical executions using filters like status, type, or a correlation ID. **Is this just for viewing data, or can it trigger workflows?** This MCP is read-only. It's designed solely for monitoring and auditing; it lets your agent query definitions and states but cannot initiate new workflow runs itself.