home / skills / inclusionai / aworld / xhs-publisher

xhs-publisher skill

safe

/examples/skill_agent/skills/xhs-publisher

This skill automates posting Xiaohongshu image notes by navigating creator center, uploading images, filling title and content, and publishing.

npx playbooks add skill inclusionai/aworld --skill xhs-publisher

Review the files below or copy the command above to add this skill to your agents.

Files (2)

SKILL.md

1.7 KB

---
name: xhs-publisher
description: 小红书发布 skill - 通过 agent-browser (CDP) 自动发布小红书图文笔记，支持多图上传、标题正文填写、一键发布。使用场景：自动化发布图文笔记到小红书创作中心。
---

# 小红书发布 (xhs-publisher)

## 概述

通过已连接 CDP 的浏览器（agent-browser）自动发布小红书图文笔记：导航到创作中心、上传图片、填写标题和正文、点击发布。

## 工具路径

- 脚本：`./publish_xhs.sh`
- 依赖：`agent-browser`（CDP 已连接且已登录小红书）、`python3`

## 用法

```bash
././publish_xhs.sh -t <title> -i <images> [-c <content> | -f <content_file>] [-p <cdp_port>]
```

### 参数

| 参数 | 说明 | 必填 | 默认 |
|------|------|------|------|
| `-t` | 标题（≤20 字符） | 是 | - |
| `-i` | 图片路径，逗号分隔或多次 `-i`（至少 1 张） | 是 | - |
| `-c` | 正文内容（与 `-f` 二选一） | 二选一 | - |
| `-f` | 从文件读取正文（与 `-c` 二选一） | 二选一 | - |
| `-p` | CDP 端口 | 否 | 9222 |

### 示例

```bash
# 单图 + 短正文
././publish_xhs.sh \
  -t "测试帖子" \
  -c "这是一条测试帖子" \
  -i /path/to/test.png

# 多图 + 文件正文
././publish_xhs.sh \
  -t "多图测试" \
  -f content.txt \
  -i img1.png,img2.png,img3.png

# 多次 -i 指定图片
././publish_xhs.sh \
  -t "分享日记" \
  -c "今天的风景真好" \
  -i photo1.jpg -i photo2.jpg
```

### 注意事项

- 浏览器需已登录小红书创作者中心
- 图片文件必须存在且为有效图片格式
- 标题不超过 20 个字符
- 正文通过 `-c` 直接传入或 `-f` 从文件读取，二者必选其一

Overview

This skill automates publishing image-and-text posts to Xiaohongshu (Red) using a browser connected via CDP (agent-browser). It navigates to the Creator Center, uploads one or more images, fills title and body, and triggers one-click publishing. The tool runs from a simple shell wrapper and requires the browser to be already logged into the Creator Center.

How this skill works

The script connects to a running browser via CDP, opens the Xiaohongshu creation page, programmatically interacts with upload controls to add images, populates the title and content fields, and executes the publish action. It accepts command-line arguments for title, images, content (inline or from file), and CDP port, validating inputs before performing the automated UI steps.

When to use it

Automate regular content publishing to Xiaohongshu to save manual effort.
Bulk post scheduling or one-off batch uploads with consistent formatting.
CI or server-side workflows that need to publish generated images and captions.
Integration with content pipelines that produce images and copy programmatically.
Testing or validation of publishing flows in a controlled environment.

Best practices

Ensure the browser session is already logged into Xiaohongshu Creator Center before running the script.
Keep titles under 20 characters as required by the platform and validated by the tool.
Provide valid image file paths; verify files exist and are supported image formats.
Use content file (-f) for longer posts or to avoid shell-escaping issues with special characters.
Run against a dedicated automation profile or headless browser instance to avoid interfering with manual use.

Example use cases

Publish a single-image test post from a CI job to verify image generation output.
Upload a carousel of photos from a travel blog pipeline with a prepared caption file.
Daily automated posts for social accounts that use templated titles and generated images.
Batch migration of archived image posts by scripting repeated publish commands.

FAQ

Do I need to log into Xiaohongshu in the browser?

Yes. The connected browser must be logged into the Xiaohongshu Creator Center before running the script because the tool drives the existing session via CDP.

How do I supply multiple images?

Pass a comma-separated list to the -i option or repeat -i multiple times; at least one image is required.