<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Chace Medeiros on Medium]]></title>
        <description><![CDATA[Stories by Chace Medeiros on Medium]]></description>
        <link>https://medium.com/@chace.medeiros_?source=rss-6fce387959e2------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*2TbLi1oGHgvQeuc4RQivgw.jpeg</url>
            <title>Stories by Chace Medeiros on Medium</title>
            <link>https://medium.com/@chace.medeiros_?source=rss-6fce387959e2------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 01:35:39 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@chace.medeiros_/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Sliding Mainframe into the Context Window: Connect your LLM with Endevor using MCP]]></title>
            <link>https://medium.com/modern-mainframe/sliding-mainframe-into-the-context-window-connect-your-llm-with-endevor-using-mcp-cea6dc48ef78?source=rss-6fce387959e2------2</link>
            <guid isPermaLink="false">https://medium.com/p/cea6dc48ef78</guid>
            <category><![CDATA[mainframe]]></category>
            <category><![CDATA[endevor]]></category>
            <category><![CDATA[mcps]]></category>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[llm]]></category>
            <dc:creator><![CDATA[Chace Medeiros]]></dc:creator>
            <pubDate>Tue, 17 Feb 2026 12:50:15 GMT</pubDate>
            <atom:updated>2026-11-12T15:43:48.178Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*V4vOougq_1oIlthVJsGctg.png" /></figure><p>Large Language Models (LLMs) are incredibly powerful, yet they are traditionally limited by a knowledge cutoff and a lack of access to your private, evolving data. When discussing a complex Endevor environment, the difference between a standard LLM interaction and one powered by a <strong>Model Context Protocol (MCP)</strong> server is the difference between general speculation and actionable intelligence.</p><p>Without an MCP server the LLM operates in isolation. It relies solely on its static training data and local context, lacking visibility and automation within your specific Endevor inventory structure, element history, metadata, or processor configurations.</p><p>When staring at 10,000 lines of COBOL in VS Code, even the latest and greatest LLM cannot help the new mainframe developer assigned to fix the latest bug in the program. Outside of the code within the element, the LLM is clueless about the element’s purpose in the larger context of an inventory.</p><h4>🚩 It does not know how an element is used.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*q8fYxWx2YC5qBL1nRso_sQ.png" /></figure><h4>🚩 It does not know about change history metadata.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*eybhrBuQXMyalE76mJWtww.png" /></figure><h4>🚩 It definitely does not know how an element is built.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mrsS8mwCy-I9G6q17QpH3Q.png" /></figure><blockquote><strong>Does the LLM work with </strong><a href="https://marketplace.visualstudio.com/items?itemName=broadcomMFD.explorer-for-endevor"><strong>Explorer for Endevor</strong></a><strong>?</strong></blockquote><p>Modern Mainframe developers rely on the Explorer for Endevor (E4E) Visual Studio Code extension from Code4z to navigate familiar codebases with precision. However, the extension was never designed to interface with an LLM. This disconnect makes it difficult to feed Endevor information into the model’s context. Additionally, the LLM cannot directly use E4E for resource retrieval, meaning the user always has to be in the loop to relay this context. This creates an efficiency bottleneck around information access.</p><h4><strong>The Challenges</strong></h4><p><strong>⚠️ Contextual Blindness:</strong> The model cannot read your specific stages, systems, or subsystems, leading to generic advice that may not apply to your environment.</p><p><strong>⚠️ Manual Data Entry:</strong> You are forced to copy-paste logs, SCL (Software Control Language) snippets, print listings, or inventory reports into the chat, which is time consuming and risks hitting context window limits.</p><p><strong>⚠️ Inaccuracy &amp; Hallucination:</strong> Lacking real-time data, the LLM may guess at element locations or dependency impacts, potentially leading to errors in your change management process.</p><h3>Introducing Endevor MCP</h3><p>We are excited to introduce <strong><em>Endevor MCP</em></strong>, a read-only bridge that connects your LLM of choice directly to your Endevor inventory.</p><p><strong><em>Endevor MCP </em></strong>is a VS Code extension or a local npm package containing an MCP server compatible with any local MCP client that supports MCP tools and streamable HTTP. It transforms the LLM from a general assistant into a <strong>context-aware subject matter expert</strong> that can query your mainframe data on demand. Think of <strong><em>Endevor MCP </em></strong>as a toolbox containing special tools for interacting with Endevor Web Services.</p><p><strong>The Benefits:</strong></p><p><strong>✅ Real-Time Retrieval:</strong> The LLM can proactively retrieve the status of a package, an environment mapping, or the history of an element using standardized tools.</p><p><strong>✅ Reduced Overhead:</strong> Instead of flooding the chat with data or not having enough of it, the MCP server provides only the specific resources needed for the current task. Keeping the conversation and the context window size efficient.</p><p><strong>✅ Precision Tooling:</strong> The LLM can troubleshoot build failures and inspect packages with direct access to the actual error logs and environment configurations.</p><p><em>Endevor MCP</em> is a core component of <a href="https://www.broadcom.com/products/mainframe/application-development-testing-devops/code4z-intelligence">Code4z Intelligence</a>, our AI-driven application understanding and impact analysis solution.</p><h3>Upgrade Your Daily Flow</h3><h4>Understand code and analyze impact</h4><p>Mainframe Applications are a web of dependencies. Before, finding the impact of a change involved manual searches. Now the LLM uses the <strong><em>get element usage</em></strong> or <strong><em>get element dependencies </em></strong>tools to traverse up and down the dependency tree. As a result, the developer gets a top-level map of the environment in one window:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hxq8Bft7Ce58eklb86y59A.png" /></figure><h4>Track and coordinate changes</h4><p>Mainframe code isn’t just code; it’s a history of business decisions. By exposing <strong><em>get element changes, </em></strong>the LLM can explain why and how a piece of code exists:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6TY3aZ3DERuI695U4UbQdA.png" /></figure><h4>Understand Endevor Builds</h4><p>Processors and component lists shouldn’t be a black box. By exposing <strong><em>get processor group, </em></strong>the LLM is enabled to retrieve a list of processor groups for a given element. If you are unsure of how a given element is built, you can ask the LLM to summarize the build process. It’s like having a DevOps engineer sitting right next to you:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6RchuP3lMsOOk6i8vclkWA.png" /></figure><h3>How it Works</h3><p>We designed this with the modern developer experience in mind. No complex green screen navigation is required.</p><ol><li><strong>Install:</strong> Developers who use GitHub Copilot install the <em>Endevor MCP</em> extension in their VS Code environment alongside Code4z , including Explorer for Endevor, and any other preferred extensions. All other developers install the local npm package.</li><li><strong>Configure &amp; Connect via EWS:</strong> Developers connect to Endevor via existing Endevor Web Services (EWS) host URLs and credentials (or a locally provisioned Zowe profile).</li><li><strong>The Secure Bridge:</strong> The MCP server allows the MCP client to fetch context directly without ever having to share EWS credentials with the LLM.</li></ol><h3>Ready to give your LLM a Mainframe education?</h3><h4>Connect with the <em>Endevor MCP team </em>to access our pre-release MCP Server available on the DevX and DevOps Validation Community:</h4><ul><li>Ekaterina Tumanova — ekaterina.tumanova@broadcom.com</li><li>Chace Medieros— chace.medeiros@broadcom.com</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cea6dc48ef78" width="1" height="1" alt=""><hr><p><a href="https://medium.com/modern-mainframe/sliding-mainframe-into-the-context-window-connect-your-llm-with-endevor-using-mcp-cea6dc48ef78">Sliding Mainframe into the Context Window: Connect your LLM with Endevor using MCP</a> was originally published in <a href="https://medium.com/modern-mainframe">Modern Mainframe</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>