小红书内容研究助手
Use this skill when the user wants content research, topic planning, competitor observation, trend review, comment insight, creator profile review, or research material organization for RedNote / XHS / Xiaohongshu(小红书).
快速开始
- 先给出当前 skill 支持的输入:关键词或选题方向、要观察的平台热榜、内容链接或内容 ID、一级评论 ID、账号主页、账号分享文本或平台账号 ID。
- 如果你只想先看样本,先取 1 页;要继续扩大,再按参数说明使用分页或
--max-items。 - 你通常会得到:榜单排名和热度信号、相关标题、作者或账号、链接或内容 ID、单条内容正文、作者、发布时间和互动指标、评论文本、回复线索和用户反馈主题、账号资料、认证、粉丝或互动信号、近期内容列表、发布时间和互动信号,以及可继续追问的角度。
示例结果
- 示例展示格式,不代表固定字段:热榜=排名/话题/热度信号、内容样本=标题/作者/链接或 ID、详情=正文/作者/发布时间/互动指标、评论=内容/用户/时间/回复线索、账号=昵称/ID/简介/粉丝信号、账号内容=标题/发布时间/互动信号;判断=相关原因和下一步。
异常处理
- 网络或 API 异常:保留错误信息,检查
SOCIALDATAX_API_KEY、参数和链接格式后原样重试一次。 - 分页中断:保留已取得的结果;重试仍失败:说明当前调用不可用,给出可替代输入方式。
常见问题
- 没结果:放宽关键词、减少限定,或换成更贴近用户表达的词。
- 结果太多:补场景、人群、品牌、时间范围或账号名。
- 调用失败:先确认
SOCIALDATAX_API_KEY已配置,再重试。 - 担心账号安全:这是只读能力,不登录、不发帖、不点赞、不评论。
- 想继续分析:把最相关的 1-3 条结果发回来,继续缩小范围。
Current platform support:
- Xiaohongshu / XHS / RedNote search hot list through
xhs_get_search_hot_list. - Xiaohongshu / XHS / RedNote notes through the
xhs_get_note_detail_by_*tools. - Xiaohongshu / XHS / RedNote notes through the
xhs_get_note_comments_by_*andxhs_get_note_sub_comments_by_comment_idtools. - Xiaohongshu / XHS / RedNote creators through the
xhs_get_user_info_by_*tools. - Xiaohongshu / XHS / RedNote creator notes through the
xhs_get_user_posted_notes_by_*tools.
API Key
Use SOCIALDATAX_API_KEY for SocialDataX requests. The only official website for requesting or managing API access is https://socialdatax.com/?from=modelscope. If a user asks where to get a key, provide only this URL; do not infer alternate domains.
获取或管理 API Key:访问 https://socialdatax.com/?from=modelscope,按官网的 API Key 申请/管理入口操作。环境变量名固定使用 SOCIALDATAX_API_KEY;不要引导用户使用其他域名。
Preferred Direct CLI
Prefer the direct CLI when the agent can run shell commands. It does not require MCP server configuration:
npx -y socialdatax-skills@latest xhs hot-search --pretty --source-client socialdatax-skills --source-platform modelscope --source-skill xhs-content-research-assistant
npx -y socialdatax-skills@latest xhs search --keyword "<keyword>" --pretty --source-client socialdatax-skills --source-platform modelscope --source-skill xhs-content-research-assistant
npx -y socialdatax-skills@latest xhs detail --note-id "<note_id>" --pretty --source-client socialdatax-skills --source-platform modelscope --source-skill xhs-content-research-assistant
npx -y socialdatax-skills@latest xhs comments --note-id "<note_id>" --pretty --source-client socialdatax-skills --source-platform modelscope --source-skill xhs-content-research-assistant
npx -y socialdatax-skills@latest xhs user-info --user-id "<user_id>" --pretty --source-client socialdatax-skills --source-platform modelscope --source-skill xhs-content-research-assistant
npx -y socialdatax-skills@latest xhs user-posts --user-id "<user_id>" --pretty --source-client socialdatax-skills --source-platform modelscope --source-skill xhs-content-research-assistant
Additional direct CLI entrypoints:
npx -y socialdatax-skills@latest xhs detail --url "<note_url_or_share_text>" --pretty --source-client socialdatax-skills --source-platform modelscope --source-skill xhs-content-research-assistant
npx -y socialdatax-skills@latest xhs comments --url "<note_url_or_share_text>" --pretty --source-client socialdatax-skills --source-platform modelscope --source-skill xhs-content-research-assistant
npx -y socialdatax-skills@latest xhs user-info --profile-url "<profile_url_or_share_text>" --pretty --source-client socialdatax-skills --source-platform modelscope --source-skill xhs-content-research-assistant
npx -y socialdatax-skills@latest xhs user-posts --profile-url "<profile_url_or_share_text>" --pretty --source-client socialdatax-skills --source-platform modelscope --source-skill xhs-content-research-assistant
npx -y socialdatax-skills@latest xhs sub-comments --note-id "<note_id>" --comment-id "<comment_id>" --pretty --source-client socialdatax-skills --source-platform modelscope --source-skill xhs-content-research-assistant
Required arguments:
- XHS
hot-search: no required arguments. --keyword <text>: content research topic; use the user's actual intent, trim whitespace, and keep it focused.
Optional arguments:
--pretty: output formatting only.--page-token <next_page_token>: opaque pagination token; omit it on the first search request. Continue only with the complete returnednext_page_tokenfrom the same search pagination chain. Do not modify, truncate, redact, mask, omit, normalize, rebuild, generate, or replace the middle with ellipses.--sort-type <general|time_descending|like_count_descending|comment_count_descending|collect_count_descending>: optional sort value; omit it for default sorting.--note-type <all|image|video>: optional content format filter; default isall.--publish-time-range <all|day|week|half_year>: optional publish-time filter; default isall.--pages <n>: fetch and merge N search pages from the current starting point.--max-items <n>: stop after collecting N search results.--since-days <1-365>: keep only search results whose publicpublish_timeis within the last N days; search remains bounded by--pages.--pretty: output formatting only; it does not change the research topic or results.- XHS
--note-id <note_id>: use the complete 24-character lowercase hexadecimalnote_idreturned from search, comments, creator note lists, or a previous detail result; do not pass only a prefix. - XHS
--url <note_url_or_share_text>: use for a note link, short link, or share text. - XHS
--note-id <note_id>: use the complete 24-character lowercase hexadecimalnote_idreturned from search, detail, comments, or creator note lists; do not pass only a prefix. --url <url_or_share_text>: use for a content page URL, short link, or share text for first-level comments.--comment-id <comment_id>: required for reply commands; use the first-level comment ID under the same content item.--page-token <next_page_token>: opaque pagination token; pass the complete returnednext_page_tokenback unchanged for the same content item or comment chain. Do not modify, truncate, redact, mask, omit, normalize, rebuild, generate, or replace the middle with ellipses.--pages <n>: fetch and merge N pages of first-level comments or replies.--all: continue first-level comments or replies untilnext_page_tokenis empty; there is no default item or page cap.--max-items <n>: stop after collecting N primary comments or replies.--include-replies: for first-levelcommentscommands only, also fetch all second-level replies under each returned first-level comment.- XHS
--user-id <user_id>: preferred when the creator ID is already known from another result. - XHS
--profile-url <profile_url_or_share_text>: use for a profile URL, short link, or profile share text. - XHS
--user-id <user_id>: preferred when the creator ID is already known. --page-token <next_page_token>: opaque pagination token; pass the complete returnednext_page_tokenback unchanged for the same creator content-list or series chain. Do not modify, truncate, redact, mask, omit, normalize, rebuild, generate, or replace the middle with ellipses.--pages <n>: fetch and merge N pages of creator content or creator series.--all: continue untilnext_page_tokenis empty; there is no default item or page cap.--max-items <n>: stop after collecting N creator content or series items.--since-days <1-365>: keep only creator content whose publicpublish_timeis within the last N days. When--pagesis omitted, the CLI continues creator content lists until the publish-time boundary is reached.--source-client socialdatax-skills --source-platform modelscope --source-skill xhs-content-research-assistant: usage attribution for this Agent Skill; keep these values unchanged when running examples from this Skill.
Use xhs hot-search for the current Xiaohongshu / XHS / RedNote search hot list. Do not ask the user for --keyword for this command.
Use either the ID option or the URL option for detail commands, not both.
The command prints JSON with platform, tool, arguments, and data.
Use either the content ID option or the URL option for first-level comments, not both. For reply commands, use the content ID together with --comment-id.
The command prints JSON with platform, tool, arguments, and data. Multi-page output keeps merged primary comments in data.items and adds page_count, item_count, and the next-page marker. With --include-replies, each first-level comment includes replies, replies_page_count, and replies_next_page_token.
Use either the ID option or the profile URL option for a single command, not both.
The command prints JSON with platform, tool, arguments, and data.
Use either the ID option or the profile URL option for a single command, not both.
The command prints JSON with platform, tool, arguments, and data. Multi-page output keeps merged creator content or series items in data.items and adds page_count, item_count, and next_page_token.
Safety Boundary
This skill is read-only. It does not read local browser data, does not save API keys, and does not perform login, posting, liking, commenting, or account changes.
MCP Tools
MCP tools matching the direct CLI commands above:
xhs_get_search_hot_listxhs_search_notesxhs_get_note_detail_by_note_idxhs_get_note_comments_by_note_idxhs_get_user_info_by_user_idxhs_get_user_posted_notes_by_user_idxhs_get_note_detail_by_note_urlxhs_get_note_comments_by_note_urlxhs_get_user_info_by_profile_urlxhs_get_user_posted_notes_by_profile_urlxhs_get_note_sub_comments_by_comment_id
If MCP tools are already available in the current agent, call xhs_get_search_hot_list without keyword arguments.
For XHS, call xhs_search_notes with keyword, optional page_token, sort_type, note_type, and publish_time_range.
Do not pass page to xhs_search_notes; omit page_token on the first request.
Continue pagination only when next_page_token is not empty, and pass the complete returned next_page_token back unchanged as page_token for the same keyword, sort, note type, publish-time range, and caller chain.
If MCP tools are already available in the current agent, use one of these tools:
xhs_get_note_detail_by_note_id: use when the complete 24-character lowercase hexadecimalnote_idis already known; do not pass only a prefix.xhs_get_note_detail_by_note_url: use for note URLs, short links, or share text. If MCP tools are already available in the current agent, use one of these tools:xhs_get_note_comments_by_note_id: use when the complete 24-character lowercase hexadecimalnote_idis known; do not pass only a prefix.xhs_get_note_comments_by_note_url: use for note URLs, short links, or share text.xhs_get_note_sub_comments_by_comment_id: use when the complete 24-character lowercase hexadecimalnote_idand first-level comment ID are known; do not pass only a note ID prefix.
Comment pagination uses opaque page_token values. Pass the complete returned next_page_token back unchanged for the same content item or comment chain. Do not modify, truncate, redact, mask, omit, normalize, rebuild, generate, or replace the middle with ellipses. Prefer CLI --pages, --all, and --include-replies when the user asks for multiple pages or a full first-level plus second-level comment tree.
XHS reply pagination also uses page_token and is bound to the current comment.
If MCP tools are already available in the current agent, use one of these tools:
xhs_get_user_info_by_user_id: preferred whenuser_idis already known from search, detail, comments, or creator note lists.xhs_get_user_info_by_profile_url: use for profile URLs, short links, or profile share text. If MCP tools are already available in the current agent, use one of these tools:xhs_get_user_posted_notes_by_user_id: preferred whenuser_idis already known.xhs_get_user_posted_notes_by_profile_url: use for profile URLs, short links, or profile share text.
Creator content-list and series pagination use opaque page_token values. Pass the complete returned next_page_token back unchanged for the same user and command family. Do not modify, truncate, redact, mask, omit, normalize, rebuild, generate, or replace the middle with ellipses. Prefer CLI --pages, --all, and --max-items when the user asks for multiple pages or all available creator content.
--since-days uses CLI-side filtering only and is not an MCP tool argument; for MCP-only calls, continue pages as needed and filter returned publish_time values in your analysis.
输出建议
优先输出可直接复盘的结果:榜单信号、相关样本和主要角度、单条内容事实、评论主题和反馈线索、创作者资料、创作者内容证据,并标出下一步可继续追问的问题。
Summarize hot-search items as observed ranking signals. Keep the current hot-search list separate from keyword search results when both are used.
Summarize visible evidence separately from interpretation. Focus on topic patterns, content angles, audience reactions, creator positioning, and useful examples when the user needs traceability.
For XHS search results, in every use of a returned note_url, such as final answers, display, references, storage, output, or forwarding, preserve it exactly as the full URL, including xsec_token query parameters. Do not modify, truncate, redact, mask, normalize, rebuild, or synthesize the URL from note_id.
For XHS note_id, copy the complete 24-character lowercase hexadecimal ID exactly; do not pass or display only a prefix.
When the user asks for recent topic research, prefer CLI --since-days 7 or another user-specified day window; do not claim complete platform coverage beyond the fetched pages.
Return factual fields such as title or description, content, author, publish time, interaction counts, images, and media summary when available.
For XHS detail results, in every use of a returned note_url, such as final answers, display, references, storage, output, or forwarding, preserve it exactly as the full URL, including xsec_token query parameters. Do not modify, truncate, redact, mask, normalize, rebuild, or synthesize the URL from note_id; if note_url is null, show the note_id or say that no directly openable full link is available.
For XHS note_id, copy the complete 24-character lowercase hexadecimal ID exactly; do not pass or display only a prefix.
Detail access is read-only and does not provide account actions.
Group comments by observed themes before inferring sentiment or demand. Mention whether the result is one page or multiple pages. Empty comments can be a valid successful result.
Report profile fields such as name, platform IDs, bio, verification, follower count, following count, received like count, IP location, and gender when available. Separate profile facts from strategic interpretation.
Summarize content-list evidence by title or description, summary, publish time, interaction counts, media links, and content type when present.
For XHS creator note-list results, copy each returned note_id as the complete 24-character lowercase hexadecimal ID exactly; do not pass or display only a prefix.
Use returned content IDs to chain into detail or comment analysis when needed.
Scan to join WeChat group