Xiaohongshu (RED) Content Prompt Builder

A good prompt, not the note, is what this tool builds

Xiaohongshu (小红书) notes have a recognisable shape — a hooky title, a warm and specific body, a cluster of topic hashtags, and a sprinkle of emoji — and large language models are very good at producing that shape once you brief them properly. The catch is the brief. Tell a model "write me a 小红书 post about a face powder" and you get something generic; tell it the product, the exact selling points, who it is for, the voice to use, the keywords to cover and the length and format you want, and you get copy you can almost publish. This builder is a deterministic assembler for exactly that brief: you fill the fields, and it joins them into a single, well-structured prompt — a leading instruction that asks the model to write a Xiaohongshu note, followed by clearly headed sections (product, selling points, audience, voice, keywords, length & format), each prefixed with a Markdown-style heading the model can parse at a glance. It does not write the note and it does not call any model; it produces the prompt you copy into the LLM of your choice. Think of it as the difference between handing a copywriter a one-line request and handing them a proper brief — the second one comes back usable.

The two highest-leverage fields are the selling points and the voice. Selling points give the model concrete, believable hooks to build around — "controls oil for eight hours", "no caking over makeup", "under the price of a coffee" — instead of vague praise. Voice decides the entire texture of the note: a close-friend chat (闺蜜口吻) reads as casual and trustworthy, a professional review (专业测评) reads as measured and credible, and a "what to avoid" warning reads as candid and useful. Naming the audience then keeps the pain points and wording pitched correctly, whether you are writing for budget-conscious students or for new parents. A good rule of thumb is to make every field concrete and specific: instead of "make it appealing", write "warm, first-person, like recommending to a close friend, no hard-sell". The more precise the brief, the less the model has to guess, and the closer the first draft lands to what you actually wanted to post.

Voice, keywords and format are what make a Xiaohongshu note land

The fields people skip and regret are keywords, length and format. Keywords and topic hashtags matter because Xiaohongshu is as much a search engine as a feed — telling the model to weave in #平价彩妆 or #学生党好物 and to append them as hashtags helps the eventual note surface for the terms your readers actually type. Length and format matter because a wall of text reads nothing like a real note: asking for a title first, a scannable body, hashtags at the end, and a few well-placed emoji turns generic prose into something that looks native to the platform. None of this constrains the model in a bad way; it focuses it. Spending one extra line on format and one on keywords is consistently the cheapest way to lift the quality of the draft you get back.

Because the output is structured plain text, the same prompt is portable across every major Chinese model and works just as well on ChatGPT, Claude or Gemini — paste it into the chat box and ask for two or three title variants while you are at it. And because the whole tool runs locally in your browser, you can iterate freely: tweak the selling points, copy again, regenerate, and compare — without anything you type ever leaving your device, being sent to a model, or being stored. One important caveat: always review what the model writes before you post it. Model-generated copy can overstate efficacy or slip into absolute claims, so check the facts, trim exaggeration, and add your own genuine experience so the published note stays truthful and within the platform's rules. Treat the first draft as a starting point, not a finished post — a quick edit pass on top of a well-briefed draft is what turns this from a novelty into a real time-saver.