skill-xhs-comment-reply · Xiaohongshu Comment Engagement
List your notes, scrape comments under a note, reply to each in your persona's voice, and delete comments.
Playwright-based with a persistent login state (shared with the publisher skill); headless-capable. Reply text is generated from your profile persona — the script only sends it. Dry-run previews first; you confirm the final wording. Has replied-ID dedup and per-reply anti-flag intervals. Automated replies carry platform risk — start with a test account in small batches.
Example invocation: "Scrape the comments on my latest note and draft replies."
Full brief
Positioning
skill-xhs-comment-reply is the engagement assistant for your Xiaohongshu comment sections: scrape + reply. Engagement only — no publishing, no analysis.
Core capabilities
- List my notes (
notes): lists published notes from the creator dashboard (title + note_id + xsec_token) — use it to locate "that note", never hand-parse tokens. - Scrape comments (
fetch):--url '<note link>'auto-resolves (recommended); intercepts the comment/page API response into comment JSON (id/nickname/content/time/region/replies). - Reply (
reply): locates a comment by nickname → clicks reply → types character by character → sends; replied-ID dedup (--replied-file) and anti-flag intervals. - Delete (
delete): deletes comments under your notes (including your own replies); dry-run by default, real deletion only with--execafter explicit confirmation (irreversible). - Persona replies: with a Profile, drafts replies from style/audience/preferences; generic-friendly otherwise.
- Out of scope: likes/favorites/private messages.
Workflow
- Check environment → QR login if needed (shared login state with xhs-publisher, one-time)
- User says "that note" →
notesto locate;fetchscrapes comment JSON and shows it - Draft per-comment replies in persona voice (hot/asking comments first, no template reuse)
- Pre-publish persona check (skill-persona-check; warns below 80)
replydry-run for user confirmation, then--exec; replied dedup
Inputs & outputs
| Input | Notes |
|---|---|
| Note locator | Note link (recommended), or bare note-id + xsec_token |
| Reply scope | Selected comment id/nickname + drafted reply |
Output: scraped comment JSON; drafted replies; reply execution results.
Boundaries with adjacent skills
| Skill | Lane |
|---|---|
| skill-xhs-comment-reply (this) | Engagement: scrape + reply |
| xhs-publisher | Publishing: image/text and video notes |
| xhs-analyzer | Analysis: viral patterns / keyword matrix / throttling detection |
| comment-insights | Comment sentiment/frequency/intent analysis (feeds on this skill's fetch JSON) |
Fit
- Batch replies maintaining your note comment sections
- Scraping comments to feed comment-insights for quantitative analysis
- Cleaning spam comments (dry-run + confirmed real deletion)
Before you start
- Needs playwright + chromium; reuses the xhs-publisher login state — scan once.
- Risk: automated replies risk throttling or bans; default reminder is test account, small batches, human review; scraping/replying defaults to
--no-proxydirect connection (proxy exits get flagged); first replies should be--headedobserved before headless batching. - Deletion is irreversible: dry-run list plus explicit user confirmation before
--exec.
skill-xhs-comment-reply is part of the Aiglade Skill library. Invoke it from the Aiglade chat box in plain language.