microsoft-docs

name	microsoft-docs
description	**ANALYSIS SKILL** — Query official Microsoft documentation to understand concepts, find tutorials, and learn how services work. WHEN: "Microsoft Learn", "Azure docs", "quickstart guide", "limits and quotas", "WAF reference", "architecture pattern docs". DO NOT USE FOR: Azure pricing; route pricing through the cost-estimate subagent.
compatibility	Works through the Azure MCP `documentation` namespace (`mcp_azure-mcp_documentation`), which proxies the Microsoft Learn MCP backend at `https://learn.microsoft.com/api/mcp`. Can also use the `mslearn` CLI as a fallback.
license	MIT
metadata	{"author":"microsoftdocs","version":"2.0","category":"documentation"}

Microsoft Docs

Search and retrieve official Microsoft documentation from learn.microsoft.com. Covers Azure, .NET, Microsoft 365, Windows, Power Platform, and all Microsoft technologies.

Prerequisites

• Azure MCP server configured in .vscode/mcp.json. Microsoft Learn tools are exposed through its documentation router.
• Outbound HTTPS to learn.microsoft.com
• Node.js ≥ 18 for the mslearn CLI fallback (via npx @microsoft/learn-cli ...)

Tools

All operations go through the Azure MCP documentation router (mcp_azure-mcp_documentation); pass the sub-command via command.

command value	Use For
microsoft_docs_search	Find documentation — concepts, guides, tutorials, configuration
microsoft_docs_fetch	Get full page content (when search excerpts aren't enough)
microsoft_code_sample_search	Find runnable code samples in official docs

// Tool: mcp_azure-mcp_documentation
{
  "intent": "find AKS private cluster guidance",
  "command": "microsoft_docs_search",
  "parameters": { "query": "AKS private cluster best practices" }
}

Rules

• Search first, fetch second — always start with microsoft_docs_search; only fetch the full page when the search excerpt is insufficient
• Be specific — include version (.NET 8, EF Core 8), task intent (quickstart, tutorial, overview, limits), and platform (Linux, Windows) where relevant
• Live docs over training data — prefer this skill over model knowledge for accuracy and freshness
• Out of scope — Azure pricing (use the cost-estimate subagent)
• Avoid loading entire docs trees — fetch single pages
• CLI fallback — when MCP server unavailable: npx @microsoft/learn-cli search "..."

Steps

1. Frame the question — service, version, intent (quickstart / config / limits / best practice)
2. microsoft_docs_search with a specific query
3. Read the excerpts — if they cover the question, stop
4. microsoft_docs_fetch on the most relevant URL only when the excerpt is cut off
5. microsoft_code_sample_search when the user wants runnable examples
6. Cite sources — include the learn.microsoft.com URL

CLI Alternative

If the Azure MCP server is unavailable:

npx @microsoft/learn-cli search "azure functions timeout"

The fetch command supports --section <heading> and --max-chars <number>.

MCP invocation	CLI equivalent
mcp_azure-mcp_documentation (command: "microsoft_docs_search")	mslearn search "..."
mcp_azure-mcp_documentation (command: "microsoft_docs_fetch")	mslearn fetch "..."