xiaohongshu

RedNote MCP Skill

Based on xiaohongshu-mcp wrapped shell script toolset.

Prerequisites

cd scripts/
./install-check.sh    # 检查依赖（xiaohongshu-mcp、jq、python3）
./start-mcp.sh        # 启动 MCP 服务（默认端口 18060）
./status.sh           # 确认已登录

When not logged in, scan the QR code:mcp-call.sh get_login_qrcode Get the QR code and scan it with the RedNote App.

The service port can be overridden via the MCP_URL environment variable (default http://localhost:18060/mcp）。

Core data flow

Important: Most operations require the feed_id + xsec_token pair. These two values are obtained from search/recommendation/user profile results,and cannot be constructed on your own。

search_feeds / list_feeds / user_profile
        │
        ▼
  返回 feeds 数组，每个 feed 包含:
  ├── id          → 用作 feed_id
  ├── xsecToken   → 用作 xsec_token
  └── noteCard    → 标题、作者、封面、互动数据
        │
        ▼
  get_feed_detail(feed_id, xsec_token)
        │
        ▼
  返回完整笔记: 正文、图片列表、评论列表
  评论中包含 comment_id、user_id（用于回复评论）

Script reference

Script	Purpose	Parameters
search.sh <关键词>	Search notes	Keyword
recommend.sh	Homepage recommendations	None
post-detail.sh <feed_id> <xsec_token>	Post details + comments	Obtained from search results
comment.sh <feed_id> <xsec_token> <内容>	Post a comment	Obtained from search results
user-profile.sh <user_id> <xsec_token>	User profile + notes	Obtained from search results
track-topic.sh <话题> [选项]	Hot topic analysis report	--limit N --output file --feishu
export-long-image.sh	Export post as long image	--posts-file json -o output.jpg
mcp-call.sh <tool> [json_args]	Generic MCP call	See the tool table below
start-mcp.sh	Start service	--headless=false --port=N
stop-mcp.sh	Stop service	None
status.sh	Check login	None
install-check.sh	Check dependencies	None

MCP tool detailed parameters

search_feeds — Search notes

{"keyword": "咖啡", "filters": {"sort_by": "最新", "note_type": "图文", "publish_time": "一周内"}}

Optional filters fields:

• sort_by: General | Latest | Most liked | Most commented | Most favorited
• note_type: All | Video | Image-and-text
• publish_time: All | Within a day | Within a week | Within half a year
• search_scope: All | Viewed | Not viewed | Following
• location: All | Same city | Nearby

get_feed_detail — Post details

{"feed_id": "...", "xsec_token": "...", "load_all_comments": true, "limit": 20}

• load_all_comments: false (default) returns the first 10; true scrolls to load more
• limit: Maximum comments to load (only effective when load_all_comments=true), default 20
• click_more_replies: Whether to expand second-level replies, default false
• reply_limit: Skip comments with more replies than this value, default 10
• scroll_speed: slow | normal | fast

post_comment_to_feed — Post a comment

{"feed_id": "...", "xsec_token": "...", "content": "写得真好！"}

reply_comment_in_feed — Reply to a comment

{"feed_id": "...", "xsec_token": "...", "content": "谢谢！", "comment_id": "...", "user_id": "..."}

comment_id and user_id are obtained from the comment list returned by get_feed_detail.

user_profile — User profile

{"user_id": "...", "xsec_token": "..."}

user_id is obtained from the feed's noteCard.user.userId ,xsec_token uses that feed's xsecToken。

like_feed — Like/Unlike

{"feed_id": "...", "xsec_token": "..."}
{"feed_id": "...", "xsec_token": "...", "unlike": true}

favorite_feed — Favorite/Unfavorite

{"feed_id": "...", "xsec_token": "..."}
{"feed_id": "...", "xsec_token": "...", "unfavorite": true}

publish_content — Publish image-and-text

{"title": "标题(≤20字)", "content": "正文(≤1000字)", "images": ["/path/to/img.jpg"], "tags": ["美食","旅行"]}

• images: At least 1, supports local paths or HTTP URLs
• tags: Optional, topic tags
• schedule_at: Optional, scheduled publishing (ISO8601, within 1 hour to 14 days)

publish_with_video — Publish video

{"title": "标题", "content": "正文", "video": "/path/to/video.mp4"}

Other tools

Tool	Parameters	Description
check_login_status	None	Check login status
list_feeds	None	Get homepage recommendations
get_login_qrcode	None	Get login QR code (Base64 PNG)
delete_cookies	None	Delete cookies, reset login

Hot topic tracking

Automatic search → fetch details → generate Markdown report.

./track-topic.sh "DeepSeek" --limit 5
./track-topic.sh "春节旅游" --limit 10 --output report.md
./track-topic.sh "iPhone 16" --limit 5 --feishu    # 导出飞书

The report includes: overview statistics, hot post details (content + top comments), comment keywords, trend analysis.

Long image export

Export the post as a JPG long image with black text on a white background.

./export-long-image.sh --posts-file posts.json -o output.jpg

posts.json format:

[{
  "title": "标题", "author": "作者", "stats": "1.3万赞",
  "desc": "正文摘要", "images": ["https://..."],
  "per_image_text": {"1": "第2张图的说明"}
}]

Dependencies: Python 3.10+, Pillow.

Notes

• Cookies are valid for about 30 days; after expiration, you need to scan the QR code again
• On first startup, a headless browser will be downloaded (~150MB)
• Avoid operating the same account from multiple clients at the same time
• Posting limits: title ≤20 characters, body ≤1000 characters, ≤50 posts per day
• Linux servers without a desktop environment need xvfb installed (apt-get install xvfb, managed automatically by the script)