microsoft-skill-creator

Microsoft Skill Creator

Create hybrid skills for Microsoft technologies that store essential knowledge locally while enabling dynamic Learn MCP lookups for deeper details.

About Skills

Skills are modular packages that extend agent capabilities with specialized knowledge and workflows. A skill transforms a general-purpose agent into a specialized one for a specific domain.

Skill Structure

skill-name/
├── SKILL.md (required)     # Frontmatter (name, description) + instructions
├── references/             # Documentation loaded into context as needed
├── sample_codes/           # Working code examples
└── assets/                 # Files used in output (templates, etc.)

Key Principles

• Frontmatter is critical: name and description determine when the skill triggers—be clear and comprehensive
• Concise is key: Only include what agents don't already know; context window is shared
• No duplication: Information lives in SKILL.md OR reference files, not both

Learn MCP Tools

CLI Alternative

If the Learn MCP server is not available, use the mslearn CLI from a terminal or shell (for example, Bash, PowerShell, or cmd) instead:

Version note: The pinned version below (@1.2.3) has been reviewed and recorded in CHANGELOG.md. Before upgrading, re-run the security review and update all occurrences in this file and in references/skill-templates.md atomically.

# Run directly (no install needed)
npx @microsoft/learn-cli@1.2.3 search "semantic kernel overview"

# Or install globally, then run
npm install -g --ignore-scripts @microsoft/learn-cli@1.2.3
mslearn search "semantic kernel overview"

Generated skills should include this same CLI fallback table so agents can use either path.

Creation Process

Step 1: Investigate the Topic

Build deep understanding using Learn MCP tools in three phases:

Phase 1 - Scope Discovery:

microsoft_docs_search(query="{technology} overview what is")
microsoft_docs_search(query="{technology} concepts architecture")
microsoft_docs_search(query="{technology} getting started tutorial")

Phase 2 - Core Content:

URL safety rule (enforced before every fetch): Only pass a URL to microsoft_docs_fetch if its scheme is https and its host is exactly learn.microsoft.com. Discard any URL from search results that does not satisfy both conditions and do not fetch it.

microsoft_docs_fetch(url="...")  # Fetch only validated learn.microsoft.com pages from Phase 1
microsoft_code_sample_search(query="{technology}", language="{lang}")

Phase 3 - Depth:

microsoft_docs_search(query="{technology} best practices")
microsoft_docs_search(query="{technology} troubleshooting errors")

Investigation Checklist

After investigating, verify:

• Can explain what the technology does in one paragraph
• Identified 3-5 key concepts
• Have working code for basic usage
• Know the most common API patterns
• Have search queries for deeper topics

Step 2: Clarify with User

Present findings and ask:

1. "I found these key areas: [list]. Which are most important?"
2. "What tasks will agents primarily perform with this skill?"
3. "Which programming language should code samples prioritize?"

Skill-name safety rule (enforced before creating any files): The value supplied for {skill-name} must match the pattern ^[a-zA-Z0-9][a-zA-Z0-9_-]{0,63}$. It must not contain /, \, ., or any whitespace. If the user-provided name does not match, ask the user to provide a valid name before proceeding. Never construct a file path from an unsanitized input.

Step 3: Generate the Skill

Use the appropriate template from skill-templates.md:

Generated Skill Structure

{skill-name}/
├── SKILL.md                    # Core knowledge + Learn MCP guidance
├── references/                 # Detailed local documentation (if needed)
└── sample_codes/               # Working code examples
    ├── getting-started/
    └── common-patterns/

Step 4: Balance Local vs Dynamic Content

Store locally when:

• Foundational (needed for any task)
• Frequently accessed
• Stable (won't change)
• Hard to find via search

Keep dynamic when:

• Exhaustive reference (too large)
• Version-specific
• Situational (specific tasks only)
• Well-indexed (easy to search)

Content Guidelines

Step 5: Validate

1. Review: Is local content sufficient for common tasks?
2. Test: Do suggested search queries return useful results?
3. Verify: Do code samples run without errors?

Common Investigation Patterns

For SDKs/Libraries

"{name} overview" → purpose, architecture
"{name} getting started quickstart" → setup steps
"{name} API reference" → core classes/methods
"{name} samples examples" → code patterns
"{name} best practices performance" → optimization

For Azure Services

"{service} overview features" → capabilities
"{service} quickstart {language}" → setup code
"{service} REST API reference" → endpoints
"{service} SDK {language}" → client library
"{service} pricing limits quotas" → constraints

For Frameworks/Platforms

"{framework} architecture concepts" → mental model
"{framework} project structure" → conventions
"{framework} tutorial walkthrough" → end-to-end flow
"{framework} configuration options" → customization

Example: Creating a "Semantic Kernel" Skill

Investigation

microsoft_docs_search(query="semantic kernel overview")
microsoft_docs_search(query="semantic kernel plugins functions")
microsoft_code_sample_search(query="semantic kernel", language="csharp")
microsoft_docs_fetch(url="https://learn.microsoft.com/semantic-kernel/overview/")

Generated Skill

semantic-kernel/
├── SKILL.md
└── sample_codes/
    ├── getting-started/
    │   └── hello-kernel.cs
    └── common-patterns/
        ├── chat-completion.cs
        └── function-calling.cs

Generated SKILL.md

---
name: semantic-kernel
description: Build AI agents with Microsoft Semantic Kernel. Use for LLM-powered apps with plugins, planners, and memory in .NET or Python.
---

# Semantic Kernel

Orchestration SDK for integrating LLMs into applications with plugins, planners, and memory.

## Key Concepts

- **Kernel**: Central orchestrator managing AI services and plugins
- **Plugins**: Collections of functions the AI can call
- **Planner**: Sequences plugin functions to achieve goals
- **Memory**: Vector store integration for RAG patterns

## Quick Start

See [getting-started/hello-kernel.cs](sample_codes/getting-started/hello-kernel.cs)

## Learn More

| Topic | How to Find |
|-------|-------------|
| Plugin development | `microsoft_docs_search(query="semantic kernel plugins custom functions")` |
| Planners | `microsoft_docs_search(query="semantic kernel planner")` |
| Memory | `microsoft_docs_fetch(url="https://learn.microsoft.com/en-us/semantic-kernel/frameworks/agent/agent-memory")` |

## CLI Alternative

If the Learn MCP server is not available, use the `mslearn` CLI instead:

| MCP Tool | CLI Command |
|----------|-------------|
| `microsoft_docs_search(query: "...")` | `mslearn search "..."` |
| `microsoft_code_sample_search(query: "...", language: "...")` | `mslearn code-search "..." --language ...` |
| `microsoft_docs_fetch(url: "...")` | `mslearn fetch "..."` |

Run directly with `npx @microsoft/learn-cli@1.2.3 <command>` or install globally with `npm install -g --ignore-scripts @microsoft/learn-cli@1.2.3`.