Xiaohongshu
by xpzouying
This MCP server lets an AI assistant log into Xiaohongshu (RedNote), publish posts and videos, search content, and reply to comments and notifications.
Run with Docker
docker pull xpzouying/xiaohongshu-mcp && docker compose up -d
About
xiaohongshu-mcp gives an agent direct, authenticated control over a Xiaohongshu (RedNote) account: checking login state, generating a login QR code, publishing image-text posts or a single local video file, and browsing the home feed or search results. It's written in Go and ships as a Docker image or standalone binary for macOS, Linux, and Windows, serving the MCP endpoint over HTTP at localhost:18060/mcp once it's running.
Past posting, it covers the day-to-day work of running an account: liking and favoriting notes, reading a post's full detail with comments, replying to a specific comment, checking a user's profile, and working through unread notifications one by one. It's built for creator and operator workflows rather than generic scraping.
Key features
- Login via QR code, with cookie-based session persistence you can reset on demand
- Publish image-text posts or a single local video file, with scheduling support
- Search content and browse the home feed with filtering
- Like, favorite, comment on, and reply to specific notes and comments
- Read a user's profile and your own notification list, including unread counts
- Ships as a Docker image or prebuilt binary, no dev environment needed
Use cases
- Automating routine posting and scheduling for a Xiaohongshu content account
- Having an agent monitor and reply to new comments and @-mentions
- Pulling search results and feed recommendations for content research
- Batch-checking engagement (likes, favorites) across a set of published notes
Available tools
check_login_status
Reports whether the current session is authenticated.
publish_content
Publishes an image-text post to Xiaohongshu.
publish_with_video
Publishes a video post from a single local video file.
search_feeds
Searches Xiaohongshu content; requires an authenticated session.
post_comment_to_feed
Posts a comment on a specified note.
get_feed_detail
Returns a note's full content, images, author info, engagement metrics, and comments.
Frequently asked questions
Do I need to know Go to run this?
No. The Docker image and prebuilt binaries for macOS, Linux, and Windows need no build step; you only touch Go if you're modifying the source.
Can it post video, or just images?
Both. publish_content handles image-text posts and publish_with_video handles a single local video file.