跳到正文
cccskills
免费在 GitHub 公开

comment-mining

Use when the user wants to mine comments and replies for audience reactions, customer language, questions, objections, complaints, product ideas, buying intent, sentiment, or voice-of-customer insights from public social posts and videos.

中文简介正在准备中。当前显示原文说明。

查看安装方法

包含的文件(1)

  • SKILL.md4.0 KB

SKILL.md(原文)

安装前可以查看交给 AI 代理的指令内容。

Comment Mining

Overview

Mine public comments for what people actually ask, complain about, want, misunderstand, or repeat. The output should help with product research, content ideas, copywriting, objection handling, and audience understanding.

When to Use

Use this skill when the user asks to:

  • analyze comments on a TikTok, YouTube video, Instagram Reel, Facebook post, Reddit post, or Rumble video
  • find audience questions, objections, complaints, or buying intent
  • extract voice-of-customer language
  • find content ideas from comments
  • understand sentiment around a post, creator, product, or topic

Comment Sources

PlatformEndpoint
TikTok comments/v1/tiktok/video/comments
TikTok replies/v1/tiktok/video/comment/replies
YouTube comments/v1/youtube/video/comments
YouTube replies/v1/youtube/video/comment/replies
Instagram comments/v2/instagram/post/comments
Facebook comments/v1/facebook/post/comments
Facebook replies/v1/facebook/post/comment/replies
Reddit comments/v1/reddit/post/comments
Rumble comments/v1/rumble/video/comments

Workflow

  1. Fetch comments

    • Use the post/video URL whenever possible.
    • Paginate when the endpoint supports it and the user wants depth.
    • Preserve comment text, author if public, like/upvote count, timestamp, and source URL.
  2. Clean lightly

    • Remove obvious spam/duplicates.
    • Keep slang, misspellings, and emotional wording if it is useful customer language.
    • Do not over-normalize exact quotes.
  3. Classify each useful comment Use these buckets:

    • questions
    • objections
    • complaints/pain points
    • praise
    • confusion
    • requests/feature ideas
    • buying intent
    • controversy/debate
    • jokes/memes/culture signals
  4. Cluster themes

    • Group similar comments.
    • Score themes by frequency and intensity.
    • Highlight exact quotes for each theme.
  5. Turn insights into actions Depending on the user's goal, produce:

    • content ideas
    • FAQ ideas
    • landing page copy angles
    • product ideas
    • objection-handling bullets
    • sales/support notes

Output Format

# Comment Mining Report

## Summary
- Source(s): {urls}
- Comments analyzed: {count}
- Confidence: High/Medium/Low

## Top Themes
| Theme | Type | Frequency | Intensity | Representative quote |
|---|---|---:|---|---|

## Audience Questions
- "..."

## Objections and Concerns
- **Objection:** ...
  - Evidence: "..."
  - Response angle: ...

## Buying Intent / Demand Signals
- "..."

## Exact Language to Reuse
- "..."
- "..."

## Content Ideas From Comments
1. ...
2. ...

Quality Guardrails

  • Label sample size and confidence.
  • Separate one loud comment from a repeated pattern.
  • Preserve exact quotes for useful language.
  • Avoid claiming broad market sentiment from one post's comments.
  • Call out moderation/platform bias when relevant.

Common Pitfalls

  • Do not flatten comments into generic sentiment. The value is in questions, objections, and exact wording.
  • Do not include personally identifying details unless they are already public and necessary.
  • Do not treat bot/spam comments as audience signal.
  • Do not skip Reddit post context. For Reddit, read both the original post and comments.

评价

还没有评价。欢迎分享你的使用感受。

评论

无需登录即可发表。欢迎分享使用感受、提问,或说明在哪些环境下无法运行。

还没有评论。来写第一条吧。

提交后会立即公开。请勿填写邮箱、电话等个人信息。URL 最多 2 个。

0 / 1000 字

此表单受 reCAPTCHA 保护,适用 Google 的隐私权政策和服务条款。

同一仓库的技能

Use when the user wants to analyze active ads from Meta/Facebook, Google, or LinkedIn ad libraries; tear down a competitor's messaging; extract hooks, offers, CTAs, video transcripts, landing page claims, and test ideas from public ads.

中文简介正在准备中。当前显示原文说明。

ScrapeCreators/social-media-research-skills41502026年8月27日 更新

Use when the user wants to evaluate a creator, influencer, or brand audience using public profile signals, TikTok audience demographics, follower/following data, comments, geography, language, and content fit. Helps judge sponsorship and market fit.

中文简介正在准备中。当前显示原文说明。

ScrapeCreators/social-media-research-skills41502026年8月27日 更新

Use when the user wants to research competitors' social media strategy, compare brands or creators, find what content is working in a niche, identify content gaps, or produce a practical social strategy brief from public social data.

中文简介正在准备中。当前显示原文说明。

ScrapeCreators/social-media-research-skills41502026年8月27日 更新

Use when the user wants to turn public social videos, transcripts, posts, or creator research into reusable content assets such as LinkedIn posts, X threads, short-form scripts, newsletters, blog outlines, carousels, or content calendars.

中文简介正在准备中。当前显示原文说明。

ScrapeCreators/social-media-research-skills41502026年8月27日 更新

Use when the user wants to analyze a creator, influencer, founder, or brand social account and understand positioning, content pillars, outlier posts, hooks, format choices, audience reaction, and what can be copied or tested.

中文简介正在准备中。当前显示原文说明。

ScrapeCreators/social-media-research-skills41502026年8月27日 更新

Use when the user wants to find creators, influencers, affiliates, or social accounts in a niche for outreach, partnerships, sponsorships, UGC, seeding, or competitive research. Produces scored prospect lists from public social data.

中文简介正在准备中。当前显示原文说明。

ScrapeCreators/social-media-research-skills41502026年8月27日 更新

查看 ScrapeCreators 的全部技能

报告此技能的问题