Using Claude to automate markdown front matter updates

01 Mar, 2026

The ldstephens.net site was migrated from WordPress using the 11ty Base Blog and the WordPress import plugin. While I initially used the base blog, I've since replaced it with my own custom 11ty build.

With around 50 markdown files exported from WordPress, that I wanted to add to this site, I needed to transform their front matter into a new format compatible with the 11ty site. Instead of manually editing each file, I used Claude to automate the process.

The Problem

The exported files had front matter that looked like this:

---
    title: 12 Months with the iPad Mini
    authors:
       - name: ldstephens
         url: https://gravatar.com/ldstephensblog
    date: 2024-01-23T15:30:32.000Z
    metadata:
       categories:
          - Apple
          - iPad
       tags: []
       uuid: 11ty/import::wordpressapi-hosted::...
       type: wordpressapi-hosted
       url: http://ldstephens.net/2024/01/23/...
    ---

I needed it transformed to a clean Eleventy-compatible format:

---
    title: 12 Months with the iPad Mini
    description:
    date: 2024-01-23T15:30:32.000Z
    tags:
       - posts
    layout: layouts/post.njk
    ---
    January 23, 2024

The requirements were:

• Keep the original title and date
• Remove authors, metadata, and all other fields
• Add description (empty), tags: [posts], and layout: layouts/post.njk
• Add the date as a human-readable line at the top of the post body (e.g. January 23, 2024)
• Handle files that already had the correct front matter — those just needed the date added to the body

What Claude Did

Claude wrote a Python script that processes an entire folder of markdown files automatically. The script handles three scenarios:

1. Old format — transforms the front matter and adds the human-readable date to the body
2. Already correct format — leaves the front matter untouched and just adds the date to the body
3. Already processed — skips the file entirely, making the script safe to re-run

The script gives clear output as it runs, telling you exactly what it did to each file.

Running the Script

With the script saved to ~/scripts/ and the posts in ~/2024_ldstephensnet/, the command was simply:

python3 ~/scripts/transform_posts.py ~/2024_ldstephensnet

Takeaway

What would have been a tedious, error-prone manual task across 50+ files was solved with a short conversation and a Python script. Claude was able to understand the before/after requirements from a single example, handle edge cases like already-updated files, and deliver a script that worked correctly.