microsoft-docs

作者: microsoft

查询官方Microsoft文档,以查找Azure、.NET、Agent Framework、Aspire、VS Code、GitHub等平台的概念、教程和代码示例。

npx skills add https://github.com/microsoft/ai-agents-for-beginners --skill microsoft-docs

Microsoft Docs

Research skill for the Microsoft technology ecosystem. Covers learn.microsoft.com and documentation that lives outside it (VS Code, GitHub, Aspire, Agent Framework repos).


Default: Microsoft Learn MCP

Use these tools for everything on learn.microsoft.com — Azure, .NET, M365, Power Platform, Agent Framework, Semantic Kernel, Windows, and more. This is the primary tool for the vast majority of Microsoft documentation queries.

ToolPurpose
microsoft_docs_searchSearch learn.microsoft.com — concepts, guides, tutorials, configuration
microsoft_code_sample_searchFind working code snippets from Learn docs. Pass language (python, csharp, etc.) for best results
microsoft_docs_fetchGet full page content from a specific URL (when search excerpts aren't enough)

Use microsoft_docs_fetch after search when you need complete tutorials, all config options, or when search excerpts are truncated.


Exceptions: When to Use Other Tools

The following categories live outside learn.microsoft.com. Use the specified tool instead.

.NET Aspire — Use Aspire MCP Server (preferred) or Context7

Aspire docs live on aspire.dev, not Learn. The best tool depends on your Aspire CLI version:

CLI 13.2+ (recommended) — The Aspire MCP server includes built-in docs search tools:

MCP ToolDescription
list_docsLists all available documentation from aspire.dev
search_docsWeighted lexical search across aspire.dev content
get_docRetrieves a specific document by slug

These ship in Aspire CLI 13.2 (PR #14028). To update: aspire update --self --channel daily. Ref: https://davidpine.dev/posts/aspire-docs-mcp-tools/

CLI 13.1 — The MCP server provides integration lookup (list_integrations, get_integration_docs) but not docs search. Fall back to Context7:

Library IDUse for
/microsoft/aspire.devPrimary — guides, integrations, CLI reference, deployment
/dotnet/aspireRuntime source — API internals, implementation details
/communitytoolkit/aspireCommunity integrations — Go, Java, Node.js, Ollama

VS Code — Use Context7

VS Code docs live on code.visualstudio.com, not Learn.

Library IDUse for
/websites/code_visualstudioUser docs — settings, features, debugging, remote dev
/websites/code_visualstudio_apiExtension API — webviews, TreeViews, commands, contribution points

GitHub — Use Context7

GitHub docs live on docs.github.com and cli.github.com.

Library IDUse for
/websites/github_enActions, API, repos, security, admin, Copilot
/websites/cli_githubGitHub CLI (gh) commands and flags

Agent Framework — Use Learn MCP + Context7

Agent Framework tutorials are on learn.microsoft.com (use microsoft_docs_search), but the GitHub repo has API-level detail that is often ahead of published docs — particularly DevUI REST API reference, CLI options, and .NET integration.

Library IDUse for
/websites/learn_microsoft_en-us_agent-frameworkTutorials — DevUI guides, tracing, workflow orchestration
/microsoft/agent-frameworkAPI detail — DevUI REST endpoints, CLI flags, auth, .NET AddDevUI/MapDevUI

DevUI tip: Query the Learn website source for how-to guides, then the repo source for API-level specifics (endpoint schemas, proxy config, auth tokens).


Context7 Setup

For any Context7 query, resolve the library ID first (one-time per session):

  1. Call mcp_context7_resolve-library-id with the technology name
  2. Call mcp_context7_query-docs with the returned library ID and a specific query

Writing Effective Queries

Be specific — include version, intent, and language:

# ❌ Too broad
"Azure Functions"
"agent framework"

# ✅ Specific
"Azure Functions Python v2 programming model"
"Cosmos DB partition key design best practices"
"GitHub Actions workflow_dispatch inputs matrix strategy"
"Aspire AddUvicornApp Python FastAPI integration"
"DevUI serve agents tracing OpenTelemetry directory discovery"
"Agent Framework workflow conditional edges branching handoff"

Include context:

  • Version when relevant (.NET 8, Aspire 13, VS Code 1.96)
  • Task intent (quickstart, tutorial, overview, limits, API reference)
  • Language for polyglot docs (Python, TypeScript, C#)

来自 microsoft 的更多技能

oss-growth
microsoft
OSS增长黑客角色
agent-framework-azure-ai-py
microsoft
使用Microsoft Agent Framework Python SDK(agent-framework-azure-ai)构建Azure AI Foundry代理。在创建使用AzureAIAgentsProvider的持久化代理、使用托管工具(代码解释器、文件搜索、网络搜索)、集成MCP服务器、管理对话线程或实现流式响应时使用。涵盖函数工具、结构化输出和多工具代理。
development
airunway-aks-setup
microsoft
在AKS上设置AI Runway——从裸集群到运行模型。涵盖集群验证、控制器安装、GPU评估、提供商设置和首次部署。适用场景:“设置AI Runway”、“接入AKS集群”、“安装AI Runway”、“airunway设置”、“将模型部署到AKS”、“在AKS上进行GPU推理”、“在AKS上配置KAITO”、“在AKS上运行LLM”、“在AKS上使用vLLM”、“在AKS上设置模型服务”、“AI Runway控制器”。
devops
appinsights-instrumentation
microsoft
使用Azure Application Insights对Web应用进行插桩的指南。提供遥测模式、SDK设置和配置参考。适用场景:如何对应用进行插桩、App Insights SDK、遥测模式、什么是App Insights、Application Insights指南、插桩示例、APM最佳实践。
devops
applicationinsights-web-ts
microsoft
使用Application Insights JavaScript SDK(@microsoft/applicationinsights-web)为浏览器/Web应用添加检测。用于真实用户监控(RUM)——页面视图、点击、AJAX/fetch依赖项、异常、自定义事件,以及与后端OpenTelemetry追踪关联的浏览器端GenAI代理追踪。涵盖SDK加载器脚本和npm设置、框架扩展(React、React Native、Angular)、点击分析、遥测初始化器,以及从浏览器发出的代理/工具/模型跨度所遵循的OTel GenAI语义约定。
devops
azure-ai-anomalydetector-java
microsoft
使用适用于 Java 的 Azure AI 异常检测器 SDK 构建异常检测应用程序。在实现单变量/多变量异常检测、时间序列分析或 AI 驱动的监控时使用。
development
azure-ai-language-conversations-py
microsoft
使用azure-ai-language-conversations Python SDK实现对话语言理解(CLU)。当使用ConversationAnalysisClient分析对话意图和实体、构建NLP功能或将语言理解集成到应用程序中时使用。
development
azure-ai-ml-py
microsoft
Azure Machine Learning SDK v2 for Python。用于机器学习工作区、作业、模型、数据集、计算资源和管道。 触发词:“azure-ai-ml”、“MLClient”、“工作区”、“模型注册表”、“训练作业”、“数据集”。
development