Auto-Hide Comments If The Author Has No Friends Or Followers

If you searched for how to auto-hide comments if the author does not have any friends or followers, you probably landed on a settings page that does not quite say that. Google, YouTube, Facebook and Instagram all talk about spam filtering. None of them give you a switch labeled “author has no followers.” So this post is about building that switch yourself. You can run it by hand in the Creator OS web app or iOS app, or have an agent run it through the hosted MCP server and the CLI.

  every new comment
        |
        v
  +---------------------+
  | score the author    |
  | friends? followers? |
  | account age?        |
  +---------------------+
     |            |
   weak         strong
     |            |
     v            v
  hide        leave alone
  (review)    (or reply)

  whitelist runs first, always

What Auto-Hide Comments If The Author Does Not Have Any Friends Or Followers Actually Means

Nobody ships a setting with that exact name. What exists is a family of features that get close, and people describe them in plain language until the phrase sticks. Here is the honest mapping.

Instagram and Facebook have a hidden words list and a set of comment filters. You add keywords or phrases, and comments containing them get hidden from everyone but the commenter. YouTube has “Hold potentially inappropriate comments for review” and “Increase strictness,” which pushes borderline comments into a review queue instead of publishing them. None of these look at follower count.

Follower count as a signal is something you assemble. The idea is simple: a brand new account with no posts, no followers and no followers-of-followers leaving a link-heavy comment is statistically more likely to be a drive-by spam account than a real fan. A real fan usually has some history. You cannot read that history through any platform setting, but you can read it through a tool that has the comment in hand and a place to store your rules.

So the feature you want is a rule, not a toggle: when a comment arrives, look at the author, score them, and hide the comment if the score is weak. Everything below is how to build that in Creator OS without writing a service from scratch.

Get started for $0

Why Google Shows This Setting and Where It Comes From

The search result that brought you here is probably a support page about comment moderation or spam filtering. Search engines map your phrase onto the closest page they have, which is usually a platform help article about hiding comments. That article is real and useful. It just does not do what your query asked.

The other source is older blog content about blog comment spam. On a WordPress site, comment spam filters have used signals like “author URL exists,” “author has commented before,” and “author email matches a known pattern” for years. Some of those filters do approximate a friends-and-followers check, because a spammer’s author profile is empty. That language leaked into social media advice, where the native tools do not support it.

If you run a WordPress blog alongside your socials, that older logic is still the right shape. Creator OS connects a WordPress site (WordPress.com or self-hosted, on your own domain) and can write SEO articles with the AI, drafting by default unless you set publish. You can read about that at the Creator OS docs, but for this post the important part is that the same author-scoring idea applies in both places.

The Signals You Can Check Before Hiding a Comment

You need signals you can actually read at the moment a comment arrives. Here is what is available on the social side, plus what you can infer.

• Account age. New accounts are the strongest single signal. A commenter whose account is two days old is not the same as one whose account is two years old.
• Follower count. Low on its own is meaningless. Lots of real people have twelve followers. Low combined with new account and zero posts is different.
• Following count versus followers. An account following 4,000 people with 8 followers is a follow-back farm pattern.
• Post count. An account with no posts and a profile photo pulled from somewhere is a shell.
• Comment text. Links, “check my page,” crypto words, generic praise that could apply to any post.
• Comment history on your own account. Has this person commented before? Repeated good comments should protect them permanently.

The last one is the one most people skip, and it is the most valuable. Your own comment history is the best whitelist source you will ever have. Creator OS keeps comments per post and per platform, so you can pull them and check.

Building the Rule: Thresholds for Friends, Followers, and Account Age

Pick numbers. Vague rules produce vague results. A starting point that holds up in practice:

Then decide how many weak signals trigger a hide. The version I would ship first: hide when the account is under 7 days old AND followers are under 10 AND there is a link in the comment. Three weak signals together, not one. That combination catches the obvious drive-by accounts and almost nothing else.

Once that runs clean for a week, loosen it. Two weak signals including account age under 7 days is the next step. Do not start there.

Setup Walkthrough for Instagram, TikTok, and YouTube

Comment replies work on Instagram, Facebook, X, Threads, YouTube and LinkedIn. They do not work on TikTok, so on TikTok the play is different. Here is the same workflow on three platforms.

Step 1: Pull the comments you need to judge

Start with the inbox. Look at what has arrived recently:

creatoros inbox:comments --platform instagram --since 2026-01-01T00:00:00Z --limit 100

Then narrow to one post once you have a specific one in mind:

creatoros inbox:post-comments post_01HX... --accountId acc_01HX...

The ids are typed tokens like post_, cmt_ and acc_, so you can tell at a glance what you are holding. If you would rather do this by hand, the web app shows the same comments in one inbox.

Step 2: Handle your own comments

If you have written your own comment moderation service in front of the API, you get a comment.received webhook, signed with X-CreatorOS-Signature, and you score the author in your own code. That is the fully custom path. Read the docs at https://www.creatoros.ca/docs for the base URL and the request shapes. There is no separate API host to guess at; the docs will tell you.

The faster path is to let your AI app do the scoring. Connect the hosted MCP server at mcp.creatoros.ca/mcp, which runs on every plan and installs nothing. In Claude: Settings, Connectors, Add custom connector, paste the URL, sign in, pick the workspace, choose read-only or read and write. In Claude Code:

claude mcp add --transport http creatoros https://mcp.creatoros.ca/mcp --header "Authorization: Bearer cos_live_..."

Then the tools you need are already there: list_comments to read, hide_comment to hide, reply_to_comment to reply. Your prompt becomes a rule. Something like: “Go through my new Instagram comments from the last 48 hours. For each one, tell me the account age, follower count and whether there is a link. If the account is under 7 days old, has under 10 followers and the comment contains a link, hide it. If the account has commented on my posts before across the last 90 days, never hide it. Give me the list of what you hid.”

The AI app will ask before running destructive tools. Hiding a comment is one of the marked ones, so you confirm the batch rather than discovering it later.

Step 3: TikTok

TikTok is the exception. Replying to TikTok comments is not supported. Hiding and deleting do work on TikTok for Business accounts. So on TikTok the rule runs backwards: you do not reply, you only score and hide. Keep the same thresholds. Do not expect a reply to appear in your TikTok inbox; it will not.

Step 4: YouTube

YouTube is where this rule earns the most. YouTube comment replies are supported, and you can also pin your own first comment on a video. So the workflow is: hide the drive-bys, reply to the good ones, and auto-pin your own first comment. That last part is exactly what auto-pinning your YouTube comment covers.

If you want this on autopilot rather than as a batch you run, the same logic is available through automatic spam comment hiding on Instagram and through comment-to-DM keyword funnels on Instagram and Facebook. The funnel path is worth reading next: auto replying to Instagram DMs with AI.

Whitelisting Exceptions So You Do Not Hide Real Fans

A hide rule without a whitelist will eventually hide a real person. Build the exception layer before you turn the rule on.

1. Anyone who has commented before. Pull your history, and treat anyone with a prior comment in the last 90 days as trusted. This single rule removes most of your false positives.
2. Anyone you have replied to. If you already answered them, they are not a spam account.
3. Anyone who has ever sent a DM. DMs land in the same inbox in Creator OS, so you can cross-check.
4. Your own accounts and your team’s. Obvious, and still forgotten.

On the tool side, the safest pattern is to keep every hide reversible. inbox:hide-comment takes --unhide, so a mistake is one command away:

creatoros inbox:hide-comment post_01HX... cmt_01HX... --accountId acc_01HX... --unhide

If you want replies written in your own tone rather than yours in a hurry, that is covered in AI agents replying to Instagram comments in your voice.

Measuring Results: False Positives, Spam Caught, and Reach Impact

Run the rule for two weeks and look at three numbers.

Spam caught. Count hides. If it is zero, your thresholds are too tight and you are paying for nothing. If it is hundreds a day on an account your size, they are too loose.

False positives. Count unhides divided by hides. Anything above about 5 percent means loosen the rule. The usual culprit is account age, because real new users exist and they do comment. Add the “commented before” whitelist and this number usually drops hard.

Reach. Check daily account metrics and per-post analytics before and after. Hiding comments does not remove the post or the like, and you are not deleting anything, so the effect should be small. But measure it rather than assuming:

creatoros analytics:posts --platform instagram --from 2026-01-01 --to 2026-01-31 --sortBy engagement

For the wider picture of what is safe to hand to an agent and what is not, what to automate in autonomous social posting is the right companion read.

Common Mistakes That Get Legit Comments Hidden

• Using follower count alone. Low followers is normal. Low followers plus new account plus a link is a pattern.
• Ignoring your own history. The cheapest whitelist is the comments you already have.
• Running the rule at full strictness on day one. Start with three weak signals, then loosen.
• Deleting instead of hiding. Hiding is reversible and keeps the comment visible to the commenter, which avoids a public argument. Delete only after two confirmed spam hits from the same account.
• Forgetting that the rule is per workspace. An API key is pinned to one workspace, one set of socials. If you run several brands, the key for one cannot touch another, which is what you want, but it also means you configure each workspace separately.
• Expecting TikTok replies. They are not supported. Hide and move on.
• Never checking the hidden list. Open it weekly. Rules drift as your audience grows.

Get started with Creator OS

The Creator plan is $14.99/month or $59.99/year and covers 11 platforms plus Skool, one account on each. API keys, the MCP server, the CLI and the agent skills are included on every plan, so the rule in this post costs you nothing extra. If you want an agent harness that talks you through your brand and then runs your socials on Creator OS, start the open-source Social Agents harness with npm start creatoros social-agents. It does its own brand interview, posts and replies, runs automations and reports to a local dashboard, and you bring your own model.

Sign up at creatoros.ca/sign-up, connect Instagram, TikTok and YouTube, and set the rule up in an afternoon. If you would rather watch it done first, the walkthroughs are on the KevBuildsApps YouTube channel, and the launch video shows the agent workflow end to end. For the full tool list and every flag, the docs and the MCP docs are the reference.

Get started for $0
Free trial for new subscribers. Cancel anytime.