SocialDataX Content Insights

Problem

Topic selection, competitor observation, and brand research often stall not because posts are invisible, but because samples are scattered, fields are incomplete, and conclusions are hard to trace back to raw content. SocialDataX Content Insights targets XHS / Xiaohongshu content research and turns keyword search, sample extraction, trend observation, and follow-up analysis into a reusable agent workflow.

How it works

The skill is read-only and prefers direct CLI calls to the SocialDataX API, with xhs_search_notes as the MCP equivalent. The main input is keyword or a topic direction; optional controls include sort_type, note_type, publish_time_range, and result-scoping options such as --max-items and --since-days. For pagination, pass the full returned next_page_token back as page_token without truncation or rewriting. Output usually includes title, author or account, URL or content ID, and angles that can be pursued further.

Boundaries

It does not log in, post, like, comment, or read local browser data; it uses SOCIALDATAX_API_KEY from the environment. On insufficient_balance or a credit-shortage error, do not retry repeatedly; use the recharge link in the error. Returned samples are not a full-platform snapshot. Preserve full note_url and note_id, including the xsec_token parameter. If results are empty, loosen the keyword; if too many, add audience, brand, time range, or account constraints.