My last open-source skill, /no-ai-slop, clearly hit a nerve with 4K GitHub stars. Today, I’m introducing /human-review, another free AI skill I think you’ll love. Let’s face it - giving feedback to AI in chat is painful. Asking it to “update the 3rd paragraph” or “update image size to... X” is annoying when I just want to edit the text myself or give feedback to AI directly on the page. /human-review opens HTML and Markdown files in a visual editor that lets you: → Edit and format text directly → Resize images → Leave comments for AI like a Google Doc I've been using it to edit PRDs, landing pages, product copy, and much more. The whole review loop runs locally, so your data is safe and nothing is uploaded to the cloud. 📌 Get /human-review for free on GitHub: If you find it useful, please ⭐ the repo so more people can find it. 📌 Read my full post here for more on how I use /human-review:show more

This is sick @petergyang! Funny enough I've been running almost exactly this with a skill i created called "/review-page" haha and was just working on a piece about how skills can basically be mini apps

nice great minds think alike :)

For word docs, I would just tell codex to make them html and then use the annotate feature to request my edits. Of course, this was before voice mode…

Yep the annotate feature is great but I also want to make direct edits to html and markdown.

omg built this for myself (also called it google docs for my text files). was limited to just text edits though! this seems like it also includes visuals i.e. UI so might give it a spin!

lemme know how i can make it better!

Why did I think I was the only one with this problem? Great solution. Can’t wait to try it!

AWESOME! I have been wanting something like this. After editing some pages a project I'm working on, the agent did exactly what it was told to do. I threw a star on there for you. Thanks!!

the music so hype

Nice one! Really enjoy the /no-ai-slop

You don’t use Claude’s native playground plug-in?

@petergyang love the idea of /human-review! seems like it’ll make AI feedback way less annoying. curious how it handles complex edits tho?

能否开放0.0.0.0？好像无法使用ip访问，想提供给其他同事修改

The real tell that AI-generated code needs a human-review layer: we built an entire cottage industry of skills whose whole job is undoing what the first skill did.

so fun to see a skill version! i have been using this markdown-html version, i made myself. i was so sick of changes while working with ai.

yep, added inline comments to md files in PrettyMD for exactly this. drop a comment, thread with your agent gonna try yours now

How is this different from annotating and changing setting from codex preview?

You can make direct edits to text but yeah ultimately this should just be a harness native feature.

Next phase: human-review skill for AI video creation process in addition to html page

for AI video creation just use @HyperFrames_

@HyperFrames_ I use hyperframe skill for making video too. But last mile fine tune still takes time . Color , music , size of the image , wording , adjust motions etc

@HyperFrames_ you using it with fable? any design / video from scratch task i think fable is clearly better at than sol

@kunchenguid looks like lavish was ahead of it's time!

Why not use Lavish?

>...I just want to edit the text myself... Wild idea... but why can't you just do that? Stop being lazy. Open the code and edit it.

Very helpful, will test, thanks man!

Uh this is good I build remark ( to comment on your markdown files. maybe I should extend it’s features 🤘

How does this differ from @kunchenguid lavish?

So it’s like the Claude Edit feature but for chat?!

这很实用，我最近刚好遇到这个痛点

i’ve been using claude html artifacts for a similar workflow where I can just edit annotate confirm reject anything from am editor injected but i think this will streamline much better. Nice one.

The "update the 3rd paragraph" loop is painfully real. Curious how human-review structures feedback instead of just re-prompting.

Hallelujah yes thank you

okay why havent i thought about this before

I've been using a comment sidecar. Editing the page directly is much cleaner. Nice thinking!

/no-ai-slop is one of my most-used skills. I use it almost every day！ thanks

The chat-feedback problem is the same one teams have with goals. "Make it better" and "ship faster" both fail because there's nothing concrete to check the work against. Structure beats vibes the second you need to know if you're actually done.

Really nice!!

I’ve been using Deep Skill Finder to look for skills like this, and this one would fit perfectly. If you upload it to the Meyo Skill Marketplace, it’ll be much easier for people to discover.

the review gate is the acceptance contract made visible: pause → human reads diff → approval before merge. that's boss-skill's BMAD shape, and the feedback loop is real: gate verdicts should persist to local state, so the next run starts from what was rejected.

Looks cool! You can have the skill detect if the user is running in and if so open your html file in toastty, right next to your agent sessions

A must-have skill