Loading video player...
In this video, Alistair demonstrates how to exert precise control over your workflow by calling MCP (Model Context Protocol) tools directly. Unlike Tutorial 3, where an LLM Agent autonomously decided which tool to use, this method allows you to explicitly define when and how a specific tool is executed. What You’ll Learn: • Direct MCP Configuration: How to set up a "Service Task" as an MCP Tool to bypass the LLM and execute a specific function (in this case, read_wiki_contents from Deep Wiki). • Response Variables: Configuring the workflow to store the tool's output into a specific variable (mcp_tool_result) for downstream use. • Processing Data: Using a Python "Script Task" to analyze the data returned by the tool • Dynamic Variable Substitution: A key lesson on syntax: ◦ Using double curly braces (e.g., {{formData.get('repository')}}) to inject data into text and JSON tool arguments. ◦ Using standard variable references within Python scripts. • Refining the Workflow: Moving from a hardcoded repository to a dynamic User Form that allows users to input any repository for analysis. --- Tutorial Resources --- First script: response = "The documentation is " + str(len(mcpToolResult)) + " characters long for scipy/scipy" Second script: response = "The documentation is " + str(len(mcpToolResult)) + " characters long for the " + formData.get('repository') + " repository" Next Up: In the next tutorial, we will cover how to implement subprocesses within a single BPMN diagram to organize complex workflows. Links: Github Repository: https://github.com/Dedoctive/DedoctiveDeveloperEdition Website link: https://dedoctive.ai/ #DedoctiveAI #DeveloperEdition #BPMN #Subprocesses #WorkflowAutomation