小红书内容发布服务 - MCP Streamable HTTP. Xiaohongshu (Little Red Book) automation server for content publishing, feed management, user interactions, and account operations.
16 tools with mixed quality. Naming is strong (verb_noun pattern, clear intent). Descriptions are present but vary: some are detailed (publish_content, get_feed_detail), others are minimal (check_login_status, delete_cookies). Schemas are visible for most tools with typed parameters and descriptions, but several tools lack depth in parameter constraints (enums, ranges). Error handling and output schemas are not documented. No evidence of idempotency hints or confirmation patterns for destructive operations.
检查小红书登录状态
删除 cookies 文件,重置登录状态。删除后需要重新登录。
收藏或取消收藏笔记
获取小红书笔记详情,返回笔记内容、图片、作者信息、互动数据(点赞/收藏/分享数)及评论列表。默认返回前10条一级评论,如需更多评论请设置load_all_comments=true
获取登录二维码(返回 Base64 图片和超时时间)
点赞或取消点赞笔记
获取首页 Feeds 列表
获取当前登录用户的主页信息
Minimal descriptions for simple tools (check_login_status, list_feeds, my_profile, user_profile). Descriptions under 50 chars lack context for LLM selection. Should explain WHEN to call and WHAT is returned.
No output schemas documented. Tools return data but LLMs cannot plan downstream calls without knowing response structure. E.g., get_feed_detail returns comments but structure is not specified.
Destructive operations (delete_cookies, publish_content, publish_with_video) lack confirmation or dry-run patterns. Agents can accidentally delete state or publish without review.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 65 | <=2025-11-25 | v2 |
发表评论
发布小红书图文内容
发布视频笔记到小红书 (仅本地文件)
回复评论
搜索小红书内容(需要已登录)
在本地图片素材库中搜索, 返回匹配的图片 (含本地绝对路径). 用户已经预先上传图片到素材库, 每张图都有 tag 和描述. 想发布图文笔记但用户没明确给图片路径时, 用此工具按关键词找候选, 选合适的把 storage_path 填进 publish_content.images.
获取指定用户的主页信息
联网搜索互联网获取实时信息 (搜狗, 国内直连, 免费). 用于查询当前热点 / 近期事件 / 产品信息 / 创作素材 / 模型训练数据未覆盖的内容. 中文 query 优先. 返回每条含 title/url/snippet (snippet 是摘要 ≤300 字).
Parameters like 'visibility' and 'sort_by' accept free-form strings instead of enums. LLMs may hallucinate invalid values (e.g., 'semi-public' instead of 'public|private|friends-only').
No error handling guidance. Tools do not document what errors are retryable, what requires user intervention, or how to recover. E.g., 'login required' vs 'network timeout' vs 'invalid input'.