WeSight's Obsidian Publishing Assistant Now Pushes Notes to Nine Platforms, Makes Image-Text Cards, and Syncs the Vault Over GitHub

Hello everyone, I'm Cang He.

Do you remember the earlier article about using WorkBuddy/Codex + Obsidian to build a self-growing personal knowledge base?

In it I shared the WeSight plugin I made, and at the time many friends said it was really useful.

It has 32k downloads now, confetti time ❀

Seeing so many people use something I made makes me happy 😊

I rely on it for my own daily writing, and I recently gave it a major upgrade 🆙

GitHub address: https://github.com/freestylefly/wesight-obsidian

Previously, after finishing an article, publishing to several platforms meant copying it one by one and re-uploading images.

After all that writing, the work still wasn't done.

Posting to one platform was fine, but posting to several meant repeating the same work again and again, which was genuinely annoying.

So I built the WeSight Publishing Assistant. Now I upload directly from Obsidian, pick the platforms, and it syncs over.

Connect the browser extension, click 'Multi-Platform' in Obsidian, and the article is sent to the publishing center.

Back in the publishing center, the imported article shows title, summary, cover, and body preview. If something looks wrong, switch to Markdown and edit it.

It currently supports short posts and long articles, and works with Xiaohongshu, Jike, Zhihu, CSDN, Juejin, Bilibili, Toutiao, Weibo, and X.

For example, I selected CSDN and Juejin, clicked sync, and the content was automatically sent to both. Watching it post by itself, I no longer have to keep switching between tabs. So comfortable.

To install: download the WeSight Publishing Assistant ZIP package and unzip it.

Open the browser's extension management page, enable Developer mode, and load the folder you just unzipped.

Click the plugin in the browser and you will see the login status of each platform.

But posting to Xiaohongshu, Weibo, and X is another story.

Long articles cannot just be moved over directly; the copy needs rewriting and images need matching. Rewriting everything for every platform would be exhausting.

So I built a quick convert-to-image-text feature.

You can pick multiple platforms among Xiaohongshu, Weibo, Jike, and X, or select just one.

The copy can be generated in one batch and then adjusted individually, or generated separately according to each platform's style.

In this example, I converted one article into copy for three platforms, each following that platform's style.

Embedded content 'multi-platform copy generation.mp4' was not included in the shared post.

After the copy is generated, I can add images, review it, and push it to drafts. The whole flow happens inside Obsidian.

Super smooth!

Embedded content 'Xiaohongshu copy.mp4' was not included in the shared post.

Image styles include text layout, real-photo photography, magazine collage, illustration narrative, infographic, and screenshot enhancement; pick whichever you like.

A configured style can be saved as a template and reused next time. That is a relief; once I have tuned a style, I do not want to start over every time.

WeSight also analyzes the content automatically and recommends an image style. Click to view the reasoning behind the recommendation.

The generated image-text copy and images can also be synced to the corresponding platforms through the Publishing Assistant.

For example, I asked it to convert an article into an X image-text post. After generating the copy and selecting images, I clicked 'One-Click Publish to Platform'.

The text and selected images upload directly to the publishing center, and the tags are already written for me. Click 'Start Sync'.

When sync finishes, it jumps to X, and there I see the copy and images arrived. From article to image-text post to platform sync, I do not have to do it again by hand.

Tip: non-members can currently sync only one platform. The price of a cup of milk tea unlocks the Creator membership and multi-platform distribution 🤪

The WeChat Official Account version can be previewed directly in Obsidian. Click 'Official Account' and the layout appears on the right.

In publishing settings, AI can generate the title, summary, and images, and there is a one-click option to get the same setup as Cang He.

After adding a publish link, analytics are available: reads, likes, reposts, and comments.

Now I do not have to open every platform to check whether anyone read the post. Very lazy-user friendly.

Previously, sharing an Obsidian note with friends meant copying it to Feishu, setting permissions, and generating a share link. All that just to show a friend something. So I turned the whole flow into a single button.

With WeSight Cloud, notes can be shared publicly on the internet, and friends can comment.

The browsing experience is close to reading it on the Official Account platform.

One-click upload to Feishu documents is supported too. Install the Feishu CLI and complete authorization first.

You can choose the save location when uploading, and if the article gets updated later, the changes can sync to the same Feishu document.

Open the link to view the document; text, layout, and images are all uploaded together.

After publishing, I kept thinking about how to archive the vault and how to sync it across computers. I tried iCloud and OneDrive before, but with many devices, one machine would edit while another had not caught up yet. To continue writing, I had to check whether sync had finished. That killed the mood.

So I added GitHub sync. The vault can now be bound to a GitHub repository.

It automatically pulls repository updates, can automatically push modified files, and supports a configurable sync interval.

For cached content you do not want on GitHub, put the path in the 'Auto-Commit Exclusion Paths' list and WeSight skips it.

Recent commits and pulls appear in the sync history.

If there are uncommitted local changes, it warns me, so I do not pull and lose work.

That is the thing I fear most mid-writing: it was right there, now it is gone? Rewriting it would probably lose the thought.

So uncommitted changes need a reminder.

For manual sync, click the avatar, then 'Pull Now' or 'Commit and Push'.

This also syncs between Windows and Mac.

I have set it up on a Mac mini too, so content I accumulate during the day is synced over, and when I go out I do not need to remote in.

Sometimes I am deep into writing and the quota runs out. Switching models means repeating everything, and at that point I want to yell at the AI.

So this upgrade also covers models: it supports three engines, Claude Code, Codex, and OpenCode, and Claude Code supports both local and custom configuration.

Most importantly, switching models does not affect the context!

That means when WeSight is being used for writing and one model runs out of quota, another can step in immediately. No need to re-explain the whole conversation after switching.

In custom configuration, you can connect your own model services, such as TokenDance, DeepSeek, Kimi, Qwen, GLM, and others.

Model configurations can be imported and exported, so DIY, here we go.

OpenCode also supports multiple model configurations.

Honestly, when it comes to creation, we should embrace AI, pull people out of repetitive work, and focus on creation itself.

But we also cannot trust AI blindly. AI-generated content sometimes does not match my own ideas, so I keep control.

Previously it was one hour of writing and two hours of copying, basically a digital sharecropper.

Now the copying is entirely handed to tools, and the saved time is mine to enjoy. Truly wonderful.

AI carries the bricks; you deliver the wow 😎😎

Oh, and the 1024 Guancha Conference is coming soon. WeSight will be exhibiting there, and I have saved up a lot of new features to show.

Eat often, meet often; come by the booth and chat 😎

Which feature in this upgrade stands out to you? Talk about it in the comments 😎