xiaohongshu-media-collector

XHS Media Collector

Follow shared public skill rules in:

• postplus-shared public skill rules

Legacy alias: xhs-media-collector.

Use this skill when the user wants to:

• download Xiaohongshu cover images from validated datasets
• turn normalized XHS research output into a local media manifest
• prepare local cover-image assets for XHS card composition or downstream review

Read these references before implementation:

• skills/20-research/xhs-media-collector/references/validated-surfaces.md
• skills/20-research/xhs-media-collector/references/manifest-schema.md

Default posture

Only collect asset URLs that are already exposed by a validated upstream dataset.

Supported by default:

• coverUrl from normalized XHS post datasets

Not supported by default:

• non-cover image extraction from note pages
• direct Xiaohongshu video downloader output

Do not pretend video collection works when the validated downloader returns 404 Not found data. Before accepting a media request, tell the user that the released path supports validated cover images from upstream datasets only. It does not support Xiaohongshu video downloads. The supported path is cover-image collection, not direct Xiaohongshu video downloads.

What this skill is for

• building a local manifest from normalized XHS datasets
• downloading cover images from remote URLs already present in normalized data
• verifying that downloaded files exist and are non-empty

What this skill is not for

• discovering note URLs
• extracting post metadata
• extracting arbitrary note images
• downloading videos from note URLs by default

Failure posture

• fail if the input dataset contains no downloadable coverUrl image URLs
• fail if the requested asset type is video
• fail if the requested asset role is not cover
• fail if a download returns a non-2xx response
• keep the manifest as the single source of truth for downloaded assets

Public Skill Execution Contract

• keep media manifests, download reports, and intermediate verification files under <work-folder>/.postplus/xiaohongshu-media-collector/
• keep only final downloaded user-facing assets outside .postplus/
• start with a bounded first pass, usually 3-10 cover images before broader pulls
• fail fast if a download fails or the requested asset surface is not validated instead of pretending collection succeeded

Main scripts

• scripts/build_xhs_media_manifest.mjs
• scripts/download_xhs_media_assets.mjs
• scripts/verify_xhs_media_manifest.mjs

Minimal workflow

node ${CLAUDE_SKILL_DIR}/scripts/build_xhs_media_manifest.mjs \
  --input <work-folder>/.postplus/xhs-normalized.json \
  --limit 3 \
  --output <work-folder>/.postplus/xhs-media-manifest.json

node ${CLAUDE_SKILL_DIR}/scripts/download_xhs_media_assets.mjs \
  --manifest <work-folder>/.postplus/xhs-media-manifest.json \
  --output-dir <work-folder>/.postplus/xhs-media-assets \
  --output <work-folder>/.postplus/xhs-media-download-report.json

node ${CLAUDE_SKILL_DIR}/scripts/verify_xhs_media_manifest.mjs \
  --manifest <work-folder>/.postplus/xhs-media-download-report.json

Good output

Return:

• manifest path
• downloaded file count
• failed asset count
• stable local file paths for downstream use