Skill 详情
microsoft-exam-docs
Obtains study material, but only for Microsoft certification exams.
使用前先检查
自动化审核只检查相关性,不代表安全审查或推荐。使用前请阅读来源中的说明。
SKILL.md
这段内容是审核时保存的快照。外部来源才是完整且最新的版本。
---
name: microsoft-exam-docs
description: Downloads Microsoft Learn study material for a Microsoft certification exam code such as AB-620, PL-400, AI-102, AZ-104, or MB-910. Use when the user asks to download offline study material for a Microsoft exam.
---
# Microsoft Exam Docs Downloader
Use this skill when the user gives a Microsoft exam code and wants offline Markdown study material from Microsoft Learn.
## Microsoft Learn exam structure
Microsoft certification exams on Learn follow a consistent hierarchy:
1. **Study guide** (`/credentials/certifications/resources/study-guides/{exam}`) — exam metadata, audience profile, **skills measured** (domains with weight percentages and detailed task bullets), study resource links, and change log.
2. **Learning paths** (`/training/paths/...`) — curated collections aligned to exam domains. Each path has an overview and ordered modules.
3. **Modules** (`/training/modules/...`) — multi-unit lessons on a single topic. Module overview pages are indexes only.
4. **Units** (`/training/modules/{module}/{n-unit-slug}`) — individual lesson pages with the actual teaching content.
This skill downloads the study guide metadata and the full unit-level lesson text from every discovered learning path.
## What it does
The helper script:
1. Normalizes an exam code, for example `AB-620` or `PL-400`.
2. Fetches the official Microsoft Learn study guide and certification page for exam metadata (passing score, duration, level, languages, etc.).
3. Parses the **Skills measured** section (including dated headings like `Skills measured as of ...`) with percentages.
4. Discovers official Microsoft Learn learning paths for each exam domain via Microsoft Learn Search.
5. Downloads every module and unit under each learning path — full lesson text, not just index pages.
6. Strips YAML frontmatter and metadata so files contain clean, readable content.
7. Downloads referenced Learn image assets and rewrites Markdown image links to local files under `media/`.
8. Writes the offline study material:
- `SUMMARY.md` — compact exam overview: details, useful links, skills measured (domain names + weights), learning path index with download status.
- `CONTENT.md` — all learning paths, modules, and units concatenated with headings and source links.
- `media/` — downloaded images referenced from `CONTENT.md`, scoped by module slug to avoid filename collisions.
- `retry-failed.sh` (only if downloads failed) — re-runs the download to recover failed items.
The script uses only Python standard library.
## Usage
From the project directory where the user wants the folder created:
```bash
python3 ~/.agents/skills/personal/microsoft-exam-docs/scripts/download_exam_docs.py AB-620
```
Specify output directory:
```bash
python3 ~/.agents/skills/personal/microsoft-exam-docs/scripts/download_exam_docs.py PL-400 --out microsoft-learn-pl-400
```
Control search breadth for discovering learning paths:
```bash
python3 ~/.agents/skills/personal/microsoft-exam-docs/scripts/download_exam_docs.py PL-400 --training-search 8
```
Provide learning path URLs manually (skips auto-discovery):
```bash
python3 ~/.agents/skills/personal/microsoft-exam-docs/scripts/download_exam_docs.py PL-400 --paths \
https://learn.microsoft.com/en-us/training/paths/power-platform-developer
```
Media download controls:
```bash
# Print one line per media asset by default; lower timeout for slow networks
python3 ~/.agents/skills/personal/microsoft-exam-docs/scripts/download_exam_docs.py PL-200 --asset-timeout 10 --asset-retries 1
# Text-only fallback: skip image downloads and leave original relative image links as-is
python3 ~/.agents/skills/personal/microsoft-exam-docs/scripts/download_exam_docs.py PL-200 --no-media
```
## Output structure
```
microsoft-learn-pl-400/
├── SUMMARY.md # Exam metadata + skills measured + learning path index
├── CONTENT.md # All learning path lesson content
├── media/ # Referenced images在 GitHub 阅读完整来源 (打开外部页面)