skill-community-ops · Comment Ops & Crisis Response
A comment-operations and crisis-response skill with three modes: A — reply strategy (layered reply templates for five comment types + a five-tier handling scale from must-reply to delete-and-block); B — topic mining from comments (clustering feedback into 3-5 new topic ideas); C — crisis response (three-tier crisis grading + statement drafts + unified talking points + red-line list + timing guidance).
Templates always use [placeholders], never hardcoded brands. It's a post-publish operations skill, distinct from pre-publish compliance checks.
Example invocation: "Help me reply to this batch of comments — and see if there's a topic worth mining."
Full brief
Positioning
skill-community-ops is a post-publish operations skill: replying to comments, mining comments for topics, and responding to negative events by tier. Three modes — whichever matches, runs; one request can trigger several.
Core capabilities
- Mode routing: A reply strategy / B topic mining / C crisis response; asks to clarify when the input is ambiguous.
- A — reply strategy: classifies comments into five types (ordinary praise / technical questions / purchase requests / trolls / malicious haters), gives 2-3 reusable templates per type with
[placeholders]for brand/product/links; five-tier handling: must-reply / guide to DM / pin / ignore / delete-and-block, with the rationale for each comment's tier; tone adapted per platform (Xiaohongshu warmth / Bilibili memes / Zhihu expertise / Douyin brevity / Weibo pace / Official Accounts restraint). - B — topic mining: clusters high-frequency demands, repeated questions, controversies, wishes, complaints; scores signal strength; converts high-value clusters into 3-5 topic cards (direction + source comments + why it's worth it + suggested format).
- C — crisis response: three-tier grading (🟢 ignorable / 🟡 respond / 🔴 formal statement — with the grading rationale always stated); response scripts or formal statement drafts (fact / accountability / measures / commitment, four parts) per tier; unified external talking points (one core stance + 3-5 Q&A); red-line list (no comment deletion, no blame-shifting, no emotional sparring); response-timing guidance.
- Profile-aware: with a profile, reply tone and crisis wording match the brand persona and compliance floor; without one, neutral and friendly.
- Hard rules: never recommends comment deletion, astroturfing, or malicious sparring; statements only promise deliverable measures.
Workflow
- Route to mode(s) (A / B / C, possibly multiple)
- A: classify comments → layered templates → handling tiers → platform tone
- B: comment clustering → signal scoring → topic-conversion formulas → 3-5 topic cards
- C: crisis grading (with rationale) → scripts/statement draft → unified talking points → red lines → timing
Inputs & outputs
| Input | Notes |
|---|---|
| Comment content | Required for A / B — a batch of real comments or a scenario description |
| Negative event description | Required for C — what happened / platform / spread / hard evidence |
| Target platform | Recommended — determines tone |
| Brand persona / red lines | Optional — can be provided manually without a profile |
Output: classified reply template table + handling list / comment-cluster summary + topic cards / crisis grading + scripts or statement draft + talking points + red lines + timing guidance.
Boundaries with adjacent skills
- skill-quality-gate: pre-publish compliance review; community-ops is post-publish comment engagement and crisis response.
Fit
- Batch-replying daily comment sections
- Mining the next episode's topics from comments
- Tiered response to bad-review waves or smear campaigns
- Unifying a team's external talking points
Before you start
- Prepare real comments (A/B) or a negative-event description (C: what / where / spread / evidence).
- With a profile, wording follows the brand persona and red lines; without one, generic safe wording applies.
- No comment-deletion, astroturfing, or malicious sparring is ever recommended; statement drafts never promise the undeliverable.
skill-community-ops is part of the Aiglade Skill library. Invoke it from the Aiglade chat box in plain language.