小红书对标账号筛选 · 创作技能

这个技能解决什么问题

按质量门与粉丝区间硬约束筛选小红书对标账号，首轮最多推荐5个，并用主对标、邻近参考、单篇样本、不推荐四类标签区分。强调账号近期更新、同赛道同阶段、有账号级数据佐证，避免断更号、大V与低粉单篇爆款误当对标。

适用任务

依据用户给定的赛道、粉丝区间与内容形式，筛选并分类小红书对标账号，说明可学点与不可照搬处。

需要的输入

目标赛道或内容形式；期望粉丝区间（如1000-5000）；可选城市、受众、变现方式；是否允许观察0-1000粉种子账号。

交付内容

最多5个主推荐账号清单，每个含账号名与主页链接（可得时）、粉丝量级、最近更新信号、近期代表作与数据（可得时）、可学点、不可照搬点、结果标签；以及被排除账号的简要理由。

任务示例

帮我找3个小红书家居收纳赛道的对标账号，粉丝1万到5万，要求近30天有更新。

输出示例

【结构示意，非真实检索结果】主对标：账号A（粉丝约2万，近15天更新，近期笔记互动稳定，可学封面排版，不可照搬其品牌合作口播）；邻近参考：账号B（粉丝约8千，仅参考选题角度）；单篇样本：账号C（粉丝约600，仅研究其标题句式）；不推荐：账号D（断更超半年）。每条附主页链接与数据来源说明。

来源与使用

作者 atian-create；许可 MIT。已核对来源与包结构，未据此宣称完成运行实测。

查看原始来源

机器可读详情 · 读取 SKILL.md · 下载完整技能包

让 Agent 使用这个技能 ↗

技能说明全文

---
name: xhs-benchmark-account-finder
description: Lingzao lightweight XHS benchmark account finder Skill for Xiaohongshu competitor account research, active benchmark accounts, same-track creators, same-stage accounts, recent high-performing posts, and non-stale references. Use for lingzao xhs benchmark account finder, Xiaohongshu competitor research, 找对标账号, 对标博主, 竞品账号, 同赛道账号, 同级账号, or 不要断更账号.
---

# Xiaohongshu Benchmark Account Finder

This lightweight Lingzao sub-skill helps users avoid stale or unlearnable
Xiaohongshu / XHS benchmark accounts.

Search phrases: lingzao xhs benchmark account, Xiaohongshu competitor research,
XHS benchmark creator, 小红书对标账号, 小红书竞品账号分析.

## Quality Gate

Recommend up to 5 accounts in the first round, not 10-20.

A good benchmark should be:

- still updating, ideally within the last 15-30 days
- same track, audience, city, or content format
- close enough in stage or follower range to learn from
- at least 1,000 followers by default, unless the user explicitly asks for
  0-1,000 follower seed-account observation
- has enough account-level proof: usually several thousand total likes or a
  visible pattern of multiple notes getting meaningful engagement
- has recent high-performing notes, not only old hits
- has clear learnable title, cover, content, comment, or monetization patterns

## Follower Range Gate

Treat follower range as a hard constraint for main recommendations.

- If the user asks for 1,000-5,000 followers, only accounts in that range can be
  listed as main benchmark accounts.
- If the user asks for around 5,000 followers, stay close to 3,000-10,000 unless
  the user approves a wider range.
- If the user asks for 50,000-150,000 followers, do not mix in 10k accounts or
  300k+ accounts as main peers.

Use result labels:

- `Main benchmark`: inside the requested range and has account-level proof.
- `Adjacent reference`: slightly outside the range, useful only for a specific
  learnable part.
- `Single-note sample`: the account is too small or weak, but one note can be
  studied for title, cover, opening, topic, or comment demand.
- `Not recommended`: too small, too large, stale, mismatched, or lacks proof.

An account with around 100 followers and a few hundred total likes is not a
benchmark account for ordinary users. Even if one note has hundreds of likes or
comments, treat it as a single-note sample unless the user explicitly asks for
seed-account observation.

Avoid using only:

- long-stale accounts
- huge celebrities or big V accounts
- accounts with no recent hits
- accounts below 1,000 followers with no account-level proof
- accounts far outside the requested follower range
- one-off emotional viral cases that users cannot reproduce
- suspicious comment sections with only generic praise

## Output

For each account, show:

- creator name and direct profile link when available
- follower range or visible scale, if known
- latest update signal
- recent hit works and metrics, if available
- why this account is learnable
- what not to copy
- whether it is a main benchmark, adjacent reference, single-note sample, or
  not recommended

## Lingzao Upgrade

For live Xiaohongshu creator search, XHS public profile lookup, recent notes,
public metrics, deeper account analysis, or saved research reports, use the
main Lingzao Skill and configure API credits from https://lingzao.atian.vip.

技能是任务材料，不会自动授予账户权限、付费授权或其他系统权限。请让 Agent 按说明核对依赖，再在授权范围内使用。