Comment Moderation
by
Social comment handling for ecommerce teams: sort each comment by what it actually is, then reply, hide, escalate, or pin, with the reasoning recorded.
Comment sections under paid social are where brand risk and unanswered demand sit in the same thread. A question about sizing that nobody answers costs a sale. A complaint left visible under a scaling ad costs more than the ad earns. Hiding the wrong comment costs the most of all, because the person who was hidden will say so publicly and be right. Volume makes this worse: the ads that work generate the most comments, so the risk concentrates exactly where the spend is. This skill sorts each comment by what it actually is and replies, hides, escalates, or pins.
What It Does
- Separates a buying question from a complaint from an attack on the brand.
- Prioritizes comments under the ads currently taking the most spend.
- Drafts a reply in your voice for the questions that have an answer.
- Hides only what is abusive or spam, never what is merely negative.
- Escalates the complaint that names a real order or a safety issue.
- Pins the comment worth putting at the top of the thread.
Use Cases
- Answer the sizing question under a scaling ad before it costs the sale.
- Catch a genuine product complaint before it sits under spend for three days.
- Stop a team from hiding criticism that should have been answered.
- Pin the customer answer that resolves the question everyone keeps asking.
Example Session
Go through the comments on this week's ads and tell me what needs doing.
Sixty-one comments. Four need a person, the rest are handled.
- Type | Count | Action | Who
- Sizing and fit questions | 22 | Reply drafted | Auto
- Shipping timing questions | 14 | Reply drafted | Auto
- Positive, answer-worthy | 9 | One pinned | Auto
- Negative, product complaint | 3 | Escalate | Support
- Names a specific order | 1 | Escalate | Support
- Spam and bot links | 12 | Hide | Auto
The three product complaints all describe the same seam failure on the navy jacket, which makes it a quality signal rather than three customers. It is escalated as one issue. Twelve spam comments are hidden and nothing negative is: the complaint under your highest spending ad is answered publicly, because hiding it is what turns one unhappy customer into a thread about censorship.
Why This Beats Prompting It Yourself
Sentiment vs What The Comment Is Asking For: The Hidden Risk
A generic prompt sorts comments by sentiment, which puts a sizing question and an insult in different buckets and a complaint and a criticism in the same one. Sentiment says the complaint is negative. It does not say that three of them describe the same seam and belong to quality, or that hiding it under a scaling ad is the most expensive option available.
When You Still Need The Line Between Negative And Abusive
The rule that protects you is narrow: hide spam and abuse, answer everything else. This skill applies that line explicitly and records why each comment was hidden, so a moderation decision can be reviewed later. Teams that hide on sentiment eventually hide a legitimate customer, and that is the screenshot that travels.
Known Limitations
Tone judgment is imperfect on short comments, sarcasm and languages outside your configured set, so borderline items are escalated rather than actioned. It reads the comment and the ad it sits under, not the commenter's history, so a repeat troll looks like a first-time critic. It drafts and queues the actions; publishing replies or hiding comments stays a step you take.
Install It
- Download the ZIP.
It is free and there is no account to create. - Unzip it into your agent's skills folder.
Claude Code reads ~/.claude/skills/, which is hidden by default: the command in the folder block opens it. Other agents scan their own directory, so drop the same folder there instead. - Ask your agent to use it.
Restart the agent if it was already running, then it picks the skill up with no config.
skills folder
Copy
~/.claude/skills/comment-moderation/
SKILL.md
references/moderation-policy.md
queue/
# one file per run, with every decision logged
# macOS: create the folder and open it in Finder
mkdir -p ~/.claude/skills && open ~/.claude/skills
# Windows: paste in the Explorer address bar
%USERPROFILE%\.claude\skills\Frequently asked questions
How is this different from the platform's automated moderation?
Platform moderation filters keywords and obvious spam. It does not know which ad is taking the most spend, cannot tell a buying question from a complaint, and will not notice that three separate comments describe the same product defect. This prioritizes by spend and routes by intent.
Which agents does the skill run in?
Any agent that supports the open SKILL.md format: Claude Code, Cursor, Codex CLI, GitHub Copilot, Gemini CLI, Manus, Grok and others load it unmodified. The format is portable, the location is not. Each agent scans its own skills directory, so you drop the same folder into whichever one yours uses. There is no config file to edit and no API key to provision.
What does it need connected to work?
Comments from your ad accounts and organic posts, plus current campaign spend so prioritization works. Order lookup is optional and only used to escalate comments naming a specific order.
Will it hide negative comments?
No. It hides spam and abuse only. A negative comment gets a reply or an escalation, because hiding criticism is both the most tempting option and the one that reliably makes things worse.
Does it post replies automatically?
It drafts them and queues them. Publishing is a step you take, which matters most on the comments where the draft is nearly right and the last sentence needs to be yours.
What is included with the skill?
The skill itself, the moderation policy reference, the decision log format, and the escalation routing rules. It is free and security scanned. We re-publish the ZIP when the policy changes, so download it again if your routing shifts.
Some comments are a brand risk, others are a demand signal.
Polar joins social engagement to product and revenue data on one definition, so escalating or pinning follows what the conversation is worth.
<script type="application/ld+json">{"@context":"https://schema.org","@type":"FAQPage","mainEntity":[{"@type":"Question","name":"How is this different from the platform's automated moderation?","acceptedAnswer":{"@type":"Answer","text":"Platform moderation filters keywords and obvious spam. It does not know which ad is taking the most spend, cannot tell a buying question from a complaint, and will not notice that three separate comments describe the same product defect. This prioritizes by spend and routes by intent."}},{"@type":"Question","name":"Which agents does the skill run in?","acceptedAnswer":{"@type":"Answer","text":"Any agent that supports the open SKILL.md format: Claude Code, Cursor, Codex CLI, GitHub Copilot, Gemini CLI, Manus, Grok and others load it unmodified. The format is portable, the location is not. Each agent scans its own skills directory, so you drop the same folder into whichever one yours uses. There is no config file to edit and no API key to provision."}},{"@type":"Question","name":"What does it need connected to work?","acceptedAnswer":{"@type":"Answer","text":"Comments from your ad accounts and organic posts, plus current campaign spend so prioritization works. Order lookup is optional and only used to escalate comments naming a specific order."}},{"@type":"Question","name":"Will it hide negative comments?","acceptedAnswer":{"@type":"Answer","text":"No. It hides spam and abuse only. A negative comment gets a reply or an escalation, because hiding criticism is both the most tempting option and the one that reliably makes things worse."}},{"@type":"Question","name":"Does it post replies automatically?","acceptedAnswer":{"@type":"Answer","text":"It drafts them and queues them. Publishing is a step you take, which matters most on the comments where the draft is nearly right and the last sentence needs to be yours."}},{"@type":"Question","name":"What is included with the skill?","acceptedAnswer":{"@type":"Answer","text":"The skill itself, the moderation policy reference, the decision log format, and the escalation routing rules. It is free and security scanned. We re-publish the ZIP when the policy changes, so download it again if your routing shifts."}}]}</script>