<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>MCP on Bogdan Novosad — Staff Software Engineer, AI Developer Tools</title><link>https://sacrorum.com/tags/mcp/</link><description>Recent content in MCP on Bogdan Novosad — Staff Software Engineer, AI Developer Tools</description><generator>Hugo</generator><language>en-us</language><copyright>Copyright 2024 Bogdan Novosad</copyright><lastBuildDate>Wed, 18 Feb 2026 18:45:12 +0100</lastBuildDate><atom:link href="https://sacrorum.com/tags/mcp/index.xml" rel="self" type="application/rss+xml"/><item><title>Better than LSP for LLM coding agents</title><link>https://sacrorum.com/posts/better-than-lsp-for-llms/</link><pubDate>Wed, 18 Feb 2026 09:00:00 +0100</pubDate><guid>https://sacrorum.com/posts/better-than-lsp-for-llms/</guid><description>LSP was built to power IDE UX. Glider MCP is built to power LLM workflows—fewer calls, fewer tokens, less state pain, and better results on large C#/.NET codebases.</description></item><item><title>It’s genuinely satisfying when an LLM finds your MCP server helpful (9/10)</title><link>https://sacrorum.com/posts/glider-mcp-helpful-9-of-10/</link><pubDate>Thu, 05 Feb 2026 00:00:00 +0000</pubDate><guid>https://sacrorum.com/posts/glider-mcp-helpful-9-of-10/</guid><description>A practical comparison of LLM workflows in a large C# codebase: plain text search vs Roslyn-powered MCP tools (symbols, types, references, callers) — and why autosync changes the ergonomics.</description></item><item><title>Glider MCP for C# code: semantic tooling that saves tool calls</title><link>https://sacrorum.com/posts/glider-mcp/</link><pubDate>Fri, 16 Jan 2026 00:00:00 +0000</pubDate><guid>https://sacrorum.com/posts/glider-mcp/</guid><description>A Roslyn-powered MCP server for semantic C# analysis: fewer tool calls, less context bloat, and answers grep cannot produce.</description></item><item><title>How to Setup MCP with Claude App</title><link>https://sacrorum.com/posts/how-to-setup-mcp-with-claude-app/</link><pubDate>Mon, 26 May 2025 00:00:00 +0000</pubDate><guid>https://sacrorum.com/posts/how-to-setup-mcp-with-claude-app/</guid><description>&lt;h2 id="what-is-mcp"&gt;What is MCP?&lt;/h2&gt;
&lt;p&gt;MCP is an open protocol that standardizes how applications provide context to LLMs. Think of MCP like a USB-C port for AI applications. Just as USB-C provides a standardized way to connect your devices to various peripherals and accessories, MCP provides a standardized way to connect AI models to different data sources and tools.&lt;/p&gt;
&lt;h3 id="core-architecture"&gt;Core Architecture&lt;/h3&gt;
&lt;p&gt;&lt;strong&gt;MCP Server&lt;/strong&gt;: The component that exposes resources, tools, and prompts to the AI assistant. Servers can provide access to databases, APIs, file systems, or other external services.&lt;/p&gt;</description></item></channel></rss>