Run Xiaohongshu account operations via an MCP
- @xpzouying - ：https://www.haha.ai/xiaohongshu-mcp---# 1. Task: The author developed an open-source xiaohongshu-mcp tool to automate Xiaohongshu operations. Using Claude Code as the agent, the workflow involves content research or preparation followed by automated image-text post publishing via the MCP interface. The human role included developing the tool, selecting the test account, and monitoring the initial execution.; tools: Claude Code. Human role: The original post does not say how a person stayed involved.; limitations: Prerequisites include a Xiaohongshu account and the xiaohongshu-mcp setup. Key constraints are platform limits: titles should be under 20 characters, and daily posting frequency is capped to avoid penalties. There is a significant risk of account restriction or content suppression if the platform detects automated AI behavior. Integration requires configuring Claude Code with the MCP server..
Tools
Published on Agent Work Atlas:
Overview
- @xpzouying - ：https://www.haha.ai/xiaohongshu-mcp---# 1.
Workflow
The author developed an open-source xiaohongshu-mcp tool to automate Xiaohongshu operations. Using Claude Code as the agent, the workflow involves content research or preparation followed by automated image-text post publishing via the MCP interface. The human role included developing the tool, selecting the test account, and monitoring the initial execution.
Reported result
During a one-week test on a new account, the process achieved over 90% automation and accumulated more than 6,000 likes and saves. However, traffic was unstable, and the author noted risks of content being flagged as 'suspected AI-generated,' which could reduce visibility. The author subsequently paused full automation to explore other tools.
Human role
The original post does not say how a person stayed involved.
Limitations
Prerequisites include a Xiaohongshu account and the xiaohongshu-mcp setup. Key constraints are platform limits: titles should be under 20 characters, and daily posting frequency is capped to avoid penalties. There is a significant risk of account restriction or content suppression if the platform detects automated AI behavior. Integration requires configuring Claude Code with the MCP server.
Try it
Act as a social media automation engineer. I want to test the xiaohongshu-mcp workflow using Claude Code. Inputs: - Target Account: [replaceable: new/test Xiaohongshu account credentials] - Content Topic: [replaceable: e.g., 'AI productivity tips'] - Output Location: [replaceable: local directory for drafts] Steps: 1. Use Claude Code to research [Content Topic] and draft 3 image-text posts. Ensure titles are under 20 characters. 2. Configure xiaohongshu-mcp to connect to [Target Account]. 3. Perform a dry run: Generate one sample post and display the preview to me for approval. Do not publish yet. 4. Upon my confirmation, publish the sample post. 5. Monitor engagement for 24 hours. Deliverables: - A JSON log of the automation steps. - Screenshots of the published post and initial metrics. Success Metrics: - Post publishes without error. - Content is not immediately flagged as AI-generated. - Engagement (likes/saves) is recorded. Error Handling: - If the API returns a rate limit or 'suspected AI' warning, stop immediately and report the error code. - Ask clarifying questions if the MCP configuration path is missing.
Source reports are not independently verified. Publication date is the catalog publication date, not the original article date.