Recherche de contenu Xiaohongshu

# Xiaohongshu Content Research

Research Xiaohongshu notes, creators, comments, and topic patterns, then produce an evidence-linked content brief without publishing or interacting with accounts.

## Loadout capabilities

Use <integration-tag>api:tikhub_api</integration-tag> for Xiaohongshu data. Prefer the `api:tikhub_api:xiaohongshu_app_v2` Action for note search, note details, comments, user details, and creator-post lists. Use `api:tikhub_api:xiaohongshu_web_v3` only when its live operation is a better match.

Inspect the exact Action schema before every call. Group Actions require the exact `actionName` constant returned by the live schema.

## Scope

This is a read-only research Skill. It does not log in to Xiaohongshu, publish posts, like, favorite, comment, reply, or manage an account.

## Research setup

Clarify:

- Category or niche.
- Target audience and geography, if relevant.
- Goal: trend discovery, competitor review, creator research, content gap, or topic validation.
- Time window and desired sample size.
- Named creators, notes, products, or keywords that must be included.

## Collection workflow

1. Create three to six query variants covering the core term, audience language, use case, and adjacent problem.
2. Search notes with `xiaohongshu_app_v2` using the exact search operation from the live schema.
3. Keep returned note IDs and source metadata. Never fabricate missing identifiers.
4. Select a balanced sample based on relevance, recency, note type, and visible engagement. Do not call a post successful from one metric alone.
5. Fetch note details and comments for the most relevant examples.
6. For creator analysis, fetch user information and recent posted notes through the matching live operations.
7. Record unavailable or incomplete fields instead of inferring them.

See [research-framework.md](research-framework.md) for the analysis rubric and output table.

## Analysis

Separate observation from interpretation:

- Observation: title, hook, format, topic, publish time, engagement fields, repeated comment questions.
- Interpretation: likely audience need, content angle, differentiation, and content gap.

Use medians or ranges when comparing engagement across several notes. Flag that platform metrics are snapshots and may not be directly comparable across different ages, creators, and paid placements.

## Deliverable

Return:

1. Research scope and collection time.
2. Evidence table with note or creator identifiers and source links when available.
3. Repeated hooks, formats, topics, and audience questions.
4. Three to seven content opportunities tied to observed evidence.
5. Suggested title patterns and outline structures, clearly labeled as recommendations.
6. Data limitations and missing fields.

## Safety and integrity

- Analyze public content only and avoid exposing unnecessary personal information.
- Do not infer health, ethnicity, sexuality, political affiliation, or other sensitive traits from profiles or posts.
- Do not encourage harassment, impersonation, engagement manipulation, or copying a creator's distinctive work.
- Paraphrase source content and keep quotations short.

## Attribution

This is a native read-only Loadout re-authoring of selected research workflows from the pinned source described in [UPSTREAM.md](UPSTREAM.md). The preserved source license is in [LICENSE.txt](LICENSE.txt).