I Built a Tool to Automate My WordPress Publishing in a Weekend. It Saved Me 400 Hours.

Week 10 of the rebuild-in-public. The single most useful thing I've built so far: a tool to automate WordPress publishing. Including every bug I hit, why you don't need to be technical to copy this, and what to do instead if you're starting a blog from scratch today. In this blog post I'm going to walk you through how I learned to automate WordPress publishing for myself, in a single weekend, with no developer background and roughly $25 in API tokens. It is the small, ugly, weekend-built piece of code that quietly saved my Sundays, my newsletter, and probably my will to keep blogging. I publish four long-form blog posts a week. Each one was taking me between one and three hours to upload, format, and make SEO-friendly inside WordPress. So I'd write the blog (the bit I love), and then spend the rest of the night fighting WordPress to get it live (the bit I'd come to dread). Four blogs a week. One to three hours each. So do the maths. That's somewhere between 200 and 600 hours a year on the publishing side alone, and most of those hours sit at half ten on a Sunday night when you're already running on coffee and the desire to be in bed. I built a tool to automate WordPress publishing for me. It took 1.5 days. It saves me roughly 400 hours a year. Most of what you'll read about AI in 2026 is the LinkedIn version. Demos. Cinematic agent loops. Someone's seventh post about a meeting transcript summariser they used once. That is not where the value is. The value is the small, ugly, weirdly specific automation that fits your business and removes a recurring tax from your week. This is one of those. And the thing I most want you to take away before you read another word, I am not a developer. I do not write Python for a living. I write Python the way someone with passable French orders dinner in Paris. Slowly, with a lot of pointing, and the occasional small humiliation. If I can build this in a weekend, you can. There is a version of this for your business. Find it. Build it. Get your evenings back.

TL;DR I publish four long-form blog posts a week. Uploading and SEO-ing each one was taking me between one and three hours of WordPress wrestling. So I built a Python script with Claude Code that drafts each post into WordPress in 30 seconds, with every Yoast SEO field populated, every image placed, and every internal link preserved. Built in 1.5 days. Cost me $25 in API tokens. Saves roughly 400 hours a year. You don't need to be technical to do this. I'm living proof.

First, the recap

This is Week 10 of my rebuild-in-public series. The short version for anyone who's just arrived:

• Week 1: I de-indexed 1,300 pages from my website. Several people questioned my sanity. Briefly, so did I.
• Week 2: my open rate crashed to 11%. The entire email industry changed its rules and forgot to tell anyone. Well, technically they did. It went to spam.
• Week 3: five unglamorous technical SEO tasks that sit on everyone's to-do list until the damage becomes visible.
• Week 4: the content rescue operation across 1,219 pages Google had visited and quietly rejected.
• Week 5: open rate from 11% to 70%, and the case that the list you already have is worth more than the one you're chasing.
• Week 6: sirens and a husband with broken ribs. Nothing went to plan. I sorted my inbox instead and wrote the guide on doing the same.
• Week 7: the income stream that's saved my business three times. Sponsored posts. The thing nobody talks about openly.
• Week 8: I let Claude loose on my own website. Found things I didn't want to find. Published them anyway.
• Week 9: the newsletter design experiment. Open rate held. Click rate imploded. I broke my own testing rules in public on purpose and learned more than I would have done correctly.

Right. So, Week 10. The week I finally sat down to automate WordPress publishing once and for all, because if I didn't, I was going to stop blogging. And I am not exaggerating that sentence.

Why I almost quit blogging in March (and built a tool to automate WordPress publishing)

So I publish four long-form blog posts a week. That's a lot. By any reasonable metric, it is more than is sensible. (My therapist agrees. So does my husband. So do most people who know what I do for a living. I am, however, an entrepreneur, which means I am incapable of doing one thing at a time without immediately wanting to do four.) But the writing isn't the hard part. I love the writing. I've been writing one form or another of marketing content for over twenty years and I'm faster at it than I have any right to be. The hard part was everything between finishing the article and the article going live. First, open WordPress. Then paste the body. Reformat the bits that came through wrong. Generate at least eight inline images and for some of my round-up posts (the 50 best tools kind, the 30 free business tools kind) have up to fifty inline images. Yes. Fifty. One per tool, one per screenshot. And I wasn't copy-pasting them in from anywhere. I was generating each image, dragging it into WordPress one at a time, typing the alt text, resizing, checking the alignment, then moving on to the next one. Fifty times. In a single post. Five hours, on the long ones. Sometimes more. For one article. That I'd already finished writing. I’d have to drag each one in, one at a time, into the right spot. Set the featured image. Then open the Yoast box. Type the SEO title. Watch Yoast turn red because the title is two characters too long. Trim it. Now watch it turn amber. Type the meta description. Watch it turn red again, because Yoast measures in pixels and a capital W is a fat letter. Type the slug. Set the focus keyphrase. Add the alt text. Add the internal links. Set the category. Re-check the keyphrase is in the first paragraph. Re-check the slug. Re-check the meta. Hit save. Then realise you forgot the featured image. Add it. Hit save. Refresh. Now the first inline image is centre-aligned and four sizes too big. Resize it. Save. Refresh. Still the second inline image has the wrong alt text. Fix it. Save. One to three hours. Per regular blog. Four blogs a week. If you've ever started a blog and quietly let it die after eight weeks, this is why. So it's not the writing. The writing's fine. It's the 47 little manual steps between writing and live, and the way they accumulate into a quiet little Sunday-night despair that nothing about your earlier strategy session prepared you for. And so, last week, I sat in front of WordPress at half ten on a Sunday night and seriously considered stopping the newsletter. Three Sundays running. The article was good. The audience was warm. My open rate was holding at 70% on a 15,000-subscriber list, which is a number most marketers would chew their own arm off for. But I had nothing left in the tank to upload it. My husband, who has watched me do this for twenty years, said something like 'you're being dramatic again.' He was right. I was being dramatic. But the dramatic had a point. The point was this. The work I love (writing) was being slowly buried under the work I don't (uploading), and the work I don't was the only thing standing between the work I love and any of it mattering. Which, when you accumulate it over four blogs a week and a year of weeks, is a quiet way to kill a business. So I finally sat down with Claude Code on a Friday morning and decided to fix it. By Sunday evening the script was running. By Monday lunchtime I'd cleared three months of backlogged drafts in a single afternoon. Here's what I built and how. Including every bug I hit. Including the ones that briefly cost me my dignity.

What does it mean to automate WordPress publishing

To automate WordPress publishing is to remove the manual upload, image-placing, and SEO-field-filling work that sits between a finished blog post and a live page on your site. Instead of clicking through the WordPress dashboard for an hour, you run a script that talks directly to your site's REST API and creates the post for you, with images, metadata, and SEO fields already in place. But it is not the same as scheduling. It's not Buffer for blog posts. It's not a content calendar tool. It's the unglamorous middle layer between your finished draft and your published page, and it's the bit almost nobody automates because the existing options are bloated, expensive, or fragile. When I say script, I mean a small Python file that lives on my Mac. It reads a Word document. It pulls the SEO fields from a tagged footer block at the bottom of the doc. It uploads the images. It calls the WordPress REST API. It creates a draft post. That's it. No SaaS subscription. No Zapier loop. No agency. Roughly 600 lines of Python across seven files. Built in 1.5 days with Claude Code as my coding partner. It will work the same way in five years as it does today, because it depends on three things that aren't going anywhere. Python. The WordPress REST API. A .env file with my credentials in it. If you've ever felt the urge to automate WordPress publishing and immediately spiralled into 'do I need a VA, a plugin, or a Make loop with three sub-zaps,' the answer is none of the above. The answer is a small custom tool that fits the way you work. Which used to be the kind of thing only developers could build, and now isn't, and that's the actual story of AI in 2026.

If you're starting a blog today, build it on Lovable. If you've already got WordPress, read on.

Quick honest detour, because someone's going to ask. If I were starting my blog from scratch today, in 2026, I would not build it on WordPress. I'd build it on Lovable, or one of the other AI-native website builders that have arrived in the last year. I'd skip the entire WordPress-Yoast-Mammoth-REST-API-redirect-bug experience. I'd let the AI handle the publishing layer from day one, the way the AI was designed to. But I haven't been building from scratch. I've been writing on WordPress for over a decade. My site has thousands of published posts, a domain authority of 49, and an SEO history I am not about to throw away to migrate to a new platform. So Lovable isn't the move for me. Building a small custom automation layer on top of what I've already got is. So if you're trying to work out which side of the line you're on, here's the rough test.

• Have you been blogging for over a year? Yes? Stay where you are. Build the layer.
• Do you have over 100 published posts? Yes? Stay where you are. Build the layer.
• Is there real organic traffic coming in? Yes? Stay where you are. Build the layer.
• No to all three? Honestly? Start fresh on Lovable or another AI-native builder. Save yourself the migration pain. Skip the publishing layer problem entirely.

So the rest of this article is for everyone in the first group. Established WordPress site. Existing SEO history. Hundreds or thousands of posts. The publishing tax is real and it's costing you hours a week. But the fix isn't to migrate. The fix is to automate the bit that hurts. Right. Onwards.

How my script to automate WordPress publishing works in 8 steps

So the whole flow takes about 30 seconds from terminal command to draft created in WordPress. Here's exactly what happens in between, in the order it happens.

1. It reads the Word document, including paragraph styles, headings, lists, and inline images.
2. It finds the SEO footer at the bottom, marked with a clear divider that says 'DELIVERY FOOTER, DELETE BEFORE PUBLISHING.'
3. It strips that footer (paragraphs and tables, both) so it never reaches the published post.
4. It converts the body to clean HTML using Mammoth, a converter that preserves headings, lists, formatting, and every single hyperlink.
5. It extracts every inline image from the docx and uploads each one to my WordPress media library, then uploads the featured image separately.
6. It runs an 11-point Yoast green pre-flight check. If anything fails, the script aborts loudly and tells me exactly what to fix. No silent uploads of broken posts.
7. It auto-fixes the three things that bite people the most: meta descriptions over the pixel limit, missing image alt text, and leaked footer tables.
8. It creates the post as a Draft, with all Yoast SEO fields populated, all images placed at the right anchors, and all hyperlinks intact. The status is hardcoded to 'draft.' The script literally cannot publish a post live without my hand on the WordPress button.

After that, I open the draft in WordPress, eyeball it, set the category, hit publish. Two minutes of human time, mostly the eyeballing.