Blog Import Post Skill

Import existing Markdown files into the Hugo blog with intelligent handling of front matter, automatic image processing, and optimized token usage.

What I do

•Import Markdown files into the blog's content/posts/ directory
•
Token-optimized processing:
- •If file has complete front matter → use cp + sed commands (saves 70% tokens)
- •If file lacks front matter → use Write tool with user-provided metadata
•Automatically detect and copy images from source directory
•Update image paths to use Hugo's static path structure
•Optionally publish immediately to GitHub

When to use me

Use this skill when the user wants to:

•Add an existing Markdown file to the blog
•Import articles written elsewhere (Documents, Downloads, etc.)
•Migrate content from other blogging platforms
•Publish a markdown file they've prepared

Trigger phrases (examples)

•"导入文章 ~/Documents/kubernetes-guide.md"
•"帮我把这篇文章添加到博客 /path/to/my-article.md"
•"import this markdown file to my blog ~/article.md"
•"把 /home/user/blog-post.md 发布到博客"
•"添加文章到博客"

How I work

Step 1: Validate the file path

Check if the user provided a markdown file path:

bash

# Verify file exists
if [ ! -f "<file-path>" ]; then
  echo "错误: 文件不存在"
  exit 1
fi

# Check extension
if [[ ! "<file-path>" =~ \.md$ ]] && [[ ! "<file-path>" =~ \.markdown$ ]]; then
  echo "错误: 文件不是 Markdown 格式"
  exit 1
fi

If not provided or invalid:

•Ask: "请提供要导入的 Markdown 文件路径"
•Verify and retry

If valid:

•Show: ✓ 找到文件: <file-path>

Step 2: Smart front matter detection (Token-optimized)

IMPORTANT: Only read the first 50 lines to save tokens

Use Read tool with limit:

code

filePath: <source-file>
offset: 0
limit: 50

Check for front matter structure:

•Line 1 must be ---
•Must have another --- within first 50 lines
•Extract: title, tags, categories, author, date, draft

Determine completeness:

•Complete: Has title field (minimum requirement). Keep existing metadata and do not ask follow-ups.
•Incomplete: Missing title or no front matter at all. Ask user for metadata.

Step 3A: Import with complete front matter (Token-saving path ⚡)

When to use: File has --- ... --- structure with at least a title field

Token estimate: ~500-800 tokens (saves 70% compared to Write)

Commands:

bash

cd ~/wzh_blog/my-blog

# Extract original filename (preserve Chinese characters)
ORIGINAL_NAME=$(basename "<source-file>" .md)
ORIGINAL_NAME=$(basename "$ORIGINAL_NAME" .markdown)

# Copy file directly
cp "<source-file>" "content/posts/${ORIGINAL_NAME}.md"

# Update author to wzh001create
sed -i 's/^author:.*/author: "wzh001create"/' "content/posts/${ORIGINAL_NAME}.md"

# Add date if missing
if ! grep -q "^date:" "content/posts/${ORIGINAL_NAME}.md"; then
  CURRENT_DATE=$(date +"%Y-%m-%dT%H:%M:%S+08:00")
  sed -i "/^---$/a date: $CURRENT_DATE" "content/posts/${ORIGINAL_NAME}.md"
fi

# Add draft: false if missing
if ! grep -q "^draft:" "content/posts/${ORIGINAL_NAME}.md"; then
  sed -i "/^date:/a draft: false" "content/posts/${ORIGINAL_NAME}.md"
fi

Show progress:

code

✓ 找到文件: <source-file>
✓ 检测到完整的 front matter，保留原有元数据
  - 标题: <extracted-title>
  - 标签: <extracted-tags>
  - 分类: <extracted-categories>
✓ 文章已导入到 content/posts/<filename>.md

Then jump to Step 4 (handle images)

Step 3B: Import without front matter (Standard path)

When to use: File has no front matter OR incomplete front matter (no title)

Token estimate: ~2500-3000 tokens (normal Write operation)

Ask user for metadata:

Use Question tool or direct text:

code

需要添加文章元数据，请提供：

1. 标题 (title):
2. 标签 (tags, 逗号分隔):
3. 分类 (categories, 逗号分隔):

Create file with Write tool:

yaml

---
title: "<user-provided-title>"
date: <current-datetime-+08:00>
draft: false
tags: ["<tag1>", "<tag2>", "<tag3>"]
categories: ["<category1>", "<category2>"]
author: "wzh001create"
---

<original-file-content>

Important:

•Use original filename (preserve Chinese characters)
•If original file had incomplete front matter, skip it and only copy the body
•Generate current date in format: 2026-01-22T19:30:00+08:00

Show progress:

code

✓ 找到文件: <source-file>
✓ 已添加 front matter 元数据
✓ 文章已导入到 content/posts/<filename>.md

Step 4: Handle images automatically

Detect images in source directory:

bash

SOURCE_DIR=$(dirname "<source-file>")
ARTICLE_NAME=$(basename "<source-file>" .md | basename - .markdown)

# Check for image files
if ls "$SOURCE_DIR"/*.{png,jpg,jpeg,gif,svg,webp} 2>/dev/null | grep -q .; then
  IMAGE_COUNT=$(ls "$SOURCE_DIR"/*.{png,jpg,jpeg,gif,svg,webp} 2>/dev/null | wc -l)
  echo "检测到 $IMAGE_COUNT 张图片"
fi

If images found: Automatically copy them (as per user requirement)

bash

# Create image directory
mkdir -p "static/images/${ARTICLE_NAME}"

# Copy all images
cp "$SOURCE_DIR"/*.{png,jpg,jpeg,gif,svg,webp} "static/images/${ARTICLE_NAME}/" 2>/dev/null

# Update image paths in the markdown
sed -i "s|\!\[\([^]]*\)\](\([^/)][^)]*\))|\![\1](/images/${ARTICLE_NAME}/\2)|g" "content/posts/${ARTICLE_NAME}.md"

echo "✓ 已复制 $IMAGE_COUNT 张图片到 static/images/${ARTICLE_NAME}/"
echo "✓ 已更新图片路径引用"

If no images found:

code

ℹ️  未检测到图片文件

Step 5: Verify import success (Minimal token usage)

DO NOT re-read the entire file - just verify it exists:

bash

ls "content/posts/<article-name>.md" && echo "✓ 文件已成功创建"

Optional: If you used Step 3A (cp + sed), you can quickly verify front matter:

bash

head -15 "content/posts/<article-name>.md" | grep -E "^(title|author|date|draft):"

Step 6: Ask about publishing

Use the Question tool to ask:

code

是否立即发布到 GitHub？
- 是 - 立即提交并推送
- 否 - 稍后手动发布

Step 7: Handle publishing

If user chooses YES:

bash

cd ~/wzh_blog/my-blog
git add -A
git commit -m "feat: 添加文章《<article-title>》"
git push

Then show:

code

✅ 导入并发布成功！

✓ 已提交更改
✓ 已推送到 GitHub
🚀 GitHub Actions 正在部署...

你的文章将在几分钟内上线：
https://wzh001create.github.io/posts/<article-slug>/

📄 文件：content/posts/<article-name>.md
🖼️  图片：static/images/<article-name>/ (如果有图片)

If user chooses NO:

code

✅ 导入成功！

📄 文件位置: content/posts/<article-name>.md
🖼️  图片目录: static/images/<article-name>/ (如果有图片)

需要发布时，请运行：
  cd ~/wzh_blog/my-blog/scripts
  ./publish.sh

或者手动执行：
  cd ~/wzh_blog/my-blog
  git add -A
  git commit -m "feat: 添加文章《<article-title>》"
  git push

Important paths and files

•Blog root directory: ~/wzh_blog/my-blog/
•Posts directory: ~/wzh_blog/my-blog/content/posts/
•Images directory: ~/wzh_blog/my-blog/static/images/
•Import script (for manual use): ~/wzh_blog/my-blog/scripts/import-post.sh
•Author name: wzh001create (always use this)

Token optimization strategy

Path A: File with complete front matter

code

1. Read first 50 lines only (500 tokens)
2. Use cp + sed commands (50 tokens)
3. Handle images with bash (100 tokens)
4. Verify with ls (50 tokens)
Total: ~700 tokens ⚡ (70% savings)

Path B: File without front matter

code

1. Read first 50 lines (500 tokens)
2. Ask user for metadata (100 tokens)
3. Write entire file (2000 tokens)
4. Handle images with bash (100 tokens)
Total: ~2700 tokens (standard)

Key principle: Avoid reading or writing the full file content unless absolutely necessary.

Error handling

File not found

code

❌ 错误: 文件不存在

检查路径: <provided-path>
请提供有效的 Markdown 文件路径。

Not a markdown file

code

❌ 错误: 文件不是 Markdown 格式

文件扩展名: <extension>
请提供 .md 或 .markdown 文件。

Git push fails

code

❌ 推送失败

文章已在本地导入，但未能推送到 GitHub。
错误信息: <error-output>

请检查网络连接和 Git 认证，然后手动推送：
  cd ~/wzh_blog/my-blog
  git push

Important notes

About the import script

•The import-post.sh script is available for manual use
•This skill does NOT call the script (to avoid interactive prompts)
•The skill uses direct file operations (cp, sed, Write) for automation

File naming

•Always preserve the original filename (including Chinese characters)
•Use the filename without extension as the article slug
•Hugo will generate the URL slug automatically

Front matter detection

•Only read first 50 lines to detect front matter
•Minimum requirement for "complete": must have title field
•If title exists, preserve all other fields (tags, categories, etc.)

Image handling

•Automatically copy images when detected (no asking)
•Always check source directory for common image formats
•Update relative paths to /images/<article-name>/...

Tools to use

•Read: Only with offset/limit to minimize token usage
•Bash: For cp, sed, ls, mkdir operations
•Write: Only when creating new files from scratch
•Question: For publishing confirmation

Example interaction

User input:

code

导入文章 ~/Downloads/Git常用命令速查手册.md

Your response (Path A - has front matter):

code

正在导入文章...

✓ 找到文件: ~/Downloads/Git常用命令速查手册.md
✓ 检测到完整的 front matter，保留原有元数据
  - 标题: Git 常用命令速查手册
  - 标签: Git, 版本控制, 命令速查
  - 分类: 教程

正在复制文件和处理图片...
ℹ️  未检测到图片文件
✓ 文章已导入到 content/posts/Git常用命令速查手册.md

📤 是否立即发布到 GitHub？(y/N)

User input:

code

Your response:

code

✅ 导入并发布成功！

✓ 已提交更改
✓ 已推送到 GitHub
🚀 GitHub Actions 正在部署...

你的文章将在几分钟内上线：
https://wzh001create.github.io/posts/git常用命令速查手册/

📄 文件：content/posts/Git常用命令速查手册.md

Remember: This skill is optimized for minimal token usage while maintaining full automation. Always choose the token-saving path (cp + sed) when possible.