Quick Verdict
xiaohongshu-mcp is for creators and marketers who already generate Xiaohongshu (RedNote) content with AI but still publish and engage by hand. It is an open-source MCP server (14,700+ GitHub stars, the highest in its class) that exposes publishing image posts, uploading videos, searching notes, and handling comments to HTTP-MCP clients like Claude Code and Cursor — so one conversation can run the full loop from topic research and competitor analysis to drafting, scheduled publishing, and comment replies. Credentials stay local and never pass through third parties.
Know the risk boundary before adopting: this kind of automation sits in a gray zone of platform rules, the author labels it “for learning purposes,” and heavy automation risks account restrictions. The sane usage is as an efficiency aid — the content is still yours, you just skip the mechanical steps. Treat it as an unattended mass-poster and the account likely won’t last.
Best For
- Xiaohongshu creators: one-step publishing of AI-drafted posts
- Ops and marketing staff: topic research, competitor-note analysis, comment upkeep
- AI workflow builders: chaining Xiaohongshu into n8n/Claude Code pipelines
- Developers: building custom ops tooling on top of it
- Not ideal for: unattended bulk broadcasters (restriction risk is yours) or users who won’t touch a CLI/Docker
Key Features
- 11 MCP tools across login, publishing, discovery, engagement, and user info
- Publishing: image posts, video uploads, scheduled publishing
- Discovery: keyword search, recommendation feed, note detail retrieval
- Engagement automation: comment, reply, like, bookmark
- Session management: one-time QR login, credentials stored locally
- Broad client support: Claude Code, Cursor, VS Code, Cline, Gemini CLI, etc.
- Three deployment options: prebuilt binaries (three platforms), Go source, docker-compose
Use Cases
- Automated publishing: AI drafts straight to image/video posts, with scheduling
- Topic and competitor research: search and pull note details for analysis
- Comment-section operations: review and reply in bulk
- Single-account content pipeline: semi-automated from topic to publish
- Data observation: profile and note statistics queries
Pricing
| Edition | Price | Notes |
|---|---|---|
| Prebuilt binary | Free | Download and run, macOS/Windows/Linux |
| Docker | Free | One-command docker-compose, easiest path |
| Build from source | Free | Requires Go, fully customizable |
Fully free and open source; the only cost is the machine running it.
Pros
- Highest-starred, actively maintained Xiaohongshu MCP
- Tools cover the full publish + discover + engage loop
- Credentials stored locally, no third-party servers
- Complete deployment options; Docker keeps the bar low
- Broad compatibility with mainstream MCP clients
Cons
- Automation may violate platform rules at your own risk; heavy use can trigger restrictions
- Platform limits apply: titles ≤20 chars, body ≤1,000 chars
- Web sessions kick each other; no concurrent logins
- Non-verified accounts may hit frequent verification prompts
- Platform front-end changes can break it until the community patches
Alternatives
| Tool | Better for | Strengths | Weaknesses |
|---|---|---|---|
| RedNote-MCP | Trying a different implementation | Same-direction open source | An order of magnitude less traction |
| n8n | Multi-platform pipeline builders | General workflow orchestration | No official Xiaohongshu node; still needs a bridge |
| Browser Use | Developers letting agents operate general web interfaces | Broader browser-task coverage for supervised page workflows | Not a Xiaohongshu-specific connector and does not guarantee compatibility with login, publishing limits, or UI changes |
| Coze | Operations teams building conversational agents and plugin workflows | Convenient combination of agents, knowledge, and workflows | Not a Xiaohongshu-specific connector; publishing and engagement still need an external bridge |
| Dify | Teams self-hosting LLM applications and workflows | Flexible visual orchestration across models and tools | Not a Xiaohongshu-specific connector and cannot replace this server’s publishing and session capabilities |
| Manual operation | Low-frequency posters | Zero restriction risk | Fully manual, slow |
FAQ
Is it free?
Fully free and open source across all deployment options.
Will my account get banned?
There is risk. The author labels it for learning; consequences of rule-violating automation are on you. Throttle frequency, keep content quality up, and validate on a secondary account first.
Are my credentials safe?
They stay on your own machine and never pass through third-party servers — but protect the runtime environment like a password.
Which deployment options are available?
You can use a prebuilt binary for macOS, Windows, or Linux, run it with docker-compose, or build it from Go source.
Can it publish both image posts and videos?
Yes. Its publishing tools support image posts, video uploads, and scheduled publishing.
Bottom Line
For anyone posting daily on Xiaohongshu, xiaohongshu-mcp closes the last gap between AI-generated content and manual publishing, and it is the most mature open-source option for the job. The right posture: validate on a secondary account, run low and steady, keep a human on content review. Used as an efficiency tool rather than a broadcaster, it genuinely saves hours.