Automated LinkedIn posting without getting shadowbanned

"Shadowban" is a label people put on a post that is live and almost nobody sees. LinkedIn does not publish a status by that name. What it does publish, in 2026, is narrower and more useful: it limits distribution of content it classifies as generic or inauthentic, it removes spam, and it restricts accounts that use software to automate activity on the website.

Those are three different outcomes. A post that stays inside your network is a distribution result. A post that disappears was acted on under the content rules. An account that cannot post until a stated time was restricted, and LinkedIn's help page tells you the software to turn off. Mixing the three into one word makes the fix look like a ritual (post slower, change the hook, buy a "safe" scheduler) when the public record points at the mechanism and the content.

Boykli automates one job: it publishes a post or a carousel PDF you already approved, to the personal profile you connect, when the slot you saved arrives. It does not comment, like, connect, message, or scrape. That scope is the whole safety argument this page will make. It is not a promise that LinkedIn will distribute the post.

Three outcomes people call a shadowban

The impression split is the audit. LinkedIn's combined post analytics defines an impression as a display on screen and says the number is an estimate. Members reached counts distinct members and Pages. In network means people who follow you or are connected to you. Out of network means everyone else. A post with impressions and almost no out-of-network share was shown, mostly to people who already had a tie to you. How retrieval and ranking produce that split is in the LinkedIn algorithm guide. Boykli's Algorithm page is a reading of those notes. It does not score your posts, and it does not tell you that you were banned.

Free Creator Tool: Test your draft against LinkedIn's 2026 distribution rules using the Free LinkedIn Algorithm Checker (scores 140-char hook fold, link penalty, and dwell potential). Turn long posts into visual slides with the Free LinkedIn Carousel Generator.

What LinkedIn has prohibited by name

The help page on automated activity, updated in the weeks before this article, says LinkedIn does not allow third-party software or browser extensions that scrape, modify the appearance of, or automate activity on LinkedIn's website. It calls that a violation of the User Agreement. If the account is restricted for that reason, the instruction is operational: review and disable the software or extension, and the account is re-enabled at the time on the notice. The same page points members who use a tool for productivity toward a contact form, and it says that review path is only rolled out to some members.

The prohibited software page quotes User Agreement section 8.2. The clauses that matter for posting are the ones about scraping or copying the services with scripts, robots, crawlers, plugins, and add-ons, and about using bots or other unauthorized automated methods to access the services, add or download contacts, send messages, or create, comment on, like, share, or re-share posts, or otherwise drive inauthentic engagement. The page also covers fake accounts, fake engagement, and tools that try to manipulate content algorithms.

The word that does the legal work in that second clause is "unauthorized." LinkedIn's own composer clock is first-party scheduling. A member who approves an application on LinkedIn's OAuth screen is granting a permission LinkedIn issued. A tool that holds the password, replays a session cookie, or drives the website in a browser is the thing the help pages describe. Boykli is the OAuth case. Sign-in happens on LinkedIn. The scopes the app requests are openid, profile, email, and w_member_social. w_member_social is the member permission used to publish as that member. Boykli does not ask for the LinkedIn password, and it does not run inside the LinkedIn tab.

Two limits sit next to that distinction, and both are easy to skip in a product pitch.

Publishing through an approved grant is how the post leaves the app. It is not a distribution upgrade. The June and September 2026 notes apply to the text, whether you paste it, use LinkedIn's clock, or send it through an API. A polished generic post is still a polished generic post.

The API terms that govern developers are a separate document from the member User Agreement, and they restrict what an application may do with LinkedIn data. This page does not treat a competitor's paraphrase of those terms as LinkedIn's reading. The member-facing rule this article relies on is the one LinkedIn states on the help pages above: do not automate activity on the website with third-party software, bots, plugins, or extensions.

Spam is independent of the tool. The Professional Community Policies tell members not to post untargeted or gratuitously repetitive promotional content, not to use invitations as a promotion channel, and not to artificially increase engagement. Pre-agreeing to like or re-share each other's posts is named. The 13 March 2026 note describes engagement pods the same way: a coordinated group liking, commenting, and sharing to force visibility. LinkedIn said it had removed groups that showed pod behavior and had warned members whose actions showed pod participation or automation-tool use, including possible removal from programs such as Top Voices.

Automated comments are called out on their own. The March note defines them as comments posted with a browser extension, a script, or a third-party tool, without the person clicking Comment. They are not allowed. The September note says LinkedIn is catching hundreds of thousands of automated comment attempts a day and has blocked billions of automated attempts (posting at scale, and slop) over a couple of months. A scheduler that also "warms up" the account with comments is in that category. Boykli has no comment action, no first-comment field, and no engagement action at all.

Four ways a post can leave your drafts

The third row is a calendar. The fourth row is account automation. People shop for them with the same query. Meet Alfred’s pricing page sells cloud connects and messages on your LinkedIn account, and it turns on a LinkedIn, Facebook, Instagram, and X calendar only from the Business plan up. The seat math is in Meet Alfred alternative. The login screen is the test you can run without a lawyer: if the tool sends you to LinkedIn to approve an app, you are in the third row. If it asks for the LinkedIn password, a li_at cookie, or permission to act in the LinkedIn tab, you are in the fourth.

What Boykli's queue actually does on the day

The product description of the calendar is LinkedIn scheduling tool. The enforcement details below are the ones that matter when the fear is "automated posting."

You choose Monday through Sunday as toggles. A new queue starts on Monday, Wednesday, and Friday. Time of day starts at 10:00. The timezone starts as the device timezone, and it is stored only after you save. Every item shares that one clock. You cannot put Monday at 08:30 and Friday at 17:00 inside this queue. Order decides which draft gets which date. Drag to reorder.

When the slot time arrives, the publisher takes the earliest queued item for that profile and publishes it. It publishes one due item at a time. It does not empty the week in one burst because a cron woke up late. It skips the account when LinkedIn is not connected, when every weekday is off, when the timezone was never saved, or when there is no generatable plan. A cancellation that still has current_period_end in the future still has access. on_hold and an expired cancel do not, and due rows wait.

A paid seat is required before the app will write, create a carousel or image, publish, or add to the queue. Logged-in visitors can browse Radar and Algorithm without one. BYOK is $29 a month. Managed is $88 a month. Managed caps are 210 posts, 150 carousels, 150 images, 25 Voice Agent sessions, and 130 post refinements. BYOK uses your OpenRouter key for chat and images and your Google AI Studio key for Voice Agent, with no Boykli generation cap. You pay those providers for usage. Neither seat adds a company Page, another network, a scheduled first comment, or an analytics pick for the hour.

Text posts, including a cover image when the draft has one, and carousel PDFs with the caption you set, are what the queue can send. A series is 2 to 10 parts, each on its own slot. The queue does not invent a posting time from your history. You pick the hour. A reading of the public 2026 timing datasets is in best time to post on LinkedIn.

A week that stays inside the published rules

This is the workflow for someone who wants posts to leave while they are in meetings, and who does not want a tool acting on the LinkedIn website.

1. Decide the mechanism before you connect anything. If the draft is already finished and you want one custom timestamp, use LinkedIn's clock. If you want the writing, a carousel PDF, and one repeating local time on a personal profile, use the Boykli queue. If the pitch is auto-comments, auto-visits, or connection requests, stop. That is the March 2026 category, and Boykli will not do it.
2. Approve the app on LinkedIn, as the profile that should speak. The Schedule header shows Connect LinkedIn until that grant exists, then LinkedIn Active. A company Page admin role does not create a Page queue. If the token expires, queued rows stay in Boykli and do not publish until you reconnect.
3. Save at least one weekday, one time of day, and the timezone. Zero days pauses auto-publish. A missing timezone blocks new items. The publisher also skips a queue with no timezone stored.
4. Write a post you would say out loud. Create posts can start from a note, a YouTube video, an article, a document, or a Voice Agent session. Radar can start a draft from an article in the feed. The September 2026 definition of slop is content that feels generic, repetitive, recycled, or built to game attention. LinkedIn also said it was replacing "enhance your post" with proofreading that does not change the member's voice. Edit the draft until a colleague could tell it was yours. The reason the product was scoped around your own idea is in building Boykli. How the draft is produced is in LinkedIn post generator.
5. Add one copy to the queue. If that draft is already queued, the app asks before it adds another. Forcing it creates a second slot. It does not edit the first. A duplicate week of the same paragraph is the "gratuitously repetitive" case in the community policies, and the API will not save you from it.
6. Read the calendar. Posts ahead is what is still waiting. Scheduled through is the last slot. Remove anything you would not sign. A failed row can go back to queued so the next pass retries that item. Five failed attempts and the row stops retrying on its own. That cap is there so a broken token does not hammer LinkedIn.
7. After the post is live, open LinkedIn's analytics for that post. Compare impressions, members reached, and the in-network versus out-of-network share. A small out-of-network share on one generic post is the June pattern, not evidence that the scheduler was detected. Change the substance, then the hour. Changing both in the same week makes the impression delta unreadable.

A concrete week: queue on Monday, Wednesday, and Friday at 09:15 in America/New_York. On Sunday you add three edited text posts and one carousel, in that order. They take Monday 09:15, Wednesday 09:15, Friday 09:15, and the following Monday 09:15. Put the carousel first if it should lead. Each publish is one item, as that member, at that clock. Nothing in that week comments on other people's posts to "seed" the algorithm.

What still gets distribution cut, even on an official post

The tool is the wrong suspect when all four of these are true: you approved an app or you used LinkedIn's own clock, you did not run a second extension that clicks the feed, the post is still on your profile, and the analytics split is mostly in network.

LinkedIn's 4 June 2026 note says content that appears generated by AI and lacks a clear perspective is less likely to travel beyond the immediate network. The 15 September note says members can report a post or comment as "Seems like AI slop" from the three dots, that those reports inform classifiers, and that enough reports can surface a message in the author's post analytics. More than one million members had used the button since August 2026. Classifiers look for low-quality, generic, inauthentic content, including content that is repetitive or built to game attention. Early detection of generic content was reported at 94% in initial testing. Views of content classified as AI slop were reported down 40%.

None of those figures is a threshold you can write toward. LinkedIn has not published a dwell-time cutoff, a hashtag cap, or a "safe" number of posts per week that exempts a generic draft. Boykli will not invent one. The practical reading is dull on purpose: a scheduled post still has to be a post. Automation changes who presses the button at 09:15. It does not change what the ranker is scoring.

Engagement pods remain a content-side violation even when every like is typed by a person. The March note is about coordination, not about whether a scheduler was involved. Asking a Slack channel to like within the first ten minutes is the behavior that note describes.

FAQ

Does LinkedIn shadowban accounts that schedule posts?

LinkedIn does not use that word in the help center or the 2026 newsroom notes. It does restrict accounts for automated activity on the website, and it tells those members to disable the software or extension. It also limits how far generic and inauthentic posts travel. A scheduler that publishes through an OAuth grant you approved is a different mechanism from a browser bot. It is still subject to the content rules.

Is Boykli's queue the same as an auto-comment or auto-connect tool?

No. The queue publishes a text post or a carousel PDF to the personal profile that authorized the app, at the slot you saved, one due item at a time. It does not send invitations, messages, likes, or comments, and it does not visit other profiles.

If I use AI to draft, will the post be limited?

It can be, if the result reads as generic, repetitive, or built to game attention. That is the June and September 2026 standard, and it applies to the text. Editing for a specific experience, a number you can stand behind, and a point someone could disagree with is the work LinkedIn's own writing-tool change points at: assistance that does not replace the member's voice. Boykli does not label a draft as safe for distribution.

How many automated posts is too many?

LinkedIn has not published a personal-profile cap that this page can quote. What it has published is enforcement against posting at scale as an automation pattern, and against gratuitously repetitive content. Boykli's default queue is three weekdays and one clock, and each tick sends one item. A week of the same paragraph on seven days is a content problem the cadence will not fix.

What do I do if LinkedIn restricts the account?

Follow the notice. The automated activity page says to disable the software or extension that automates activity on the website, after which the account is re-enabled at the time on the suspension notification. Change the password if the notice or that page tells you to. Turn off any extension that clicks, comments, or connects. Reconnecting Boykli does not clear a restriction LinkedIn applied, and the queue will not publish until LinkedIn accepts the account again.

Can this post to a company Page, or leave a first comment?

No. OAuth is the personal profile you connect. There is no first-comment field. Page posts and a scheduled comment belong in LinkedIn's composer or in a tool built for those jobs.

Does the posting hour protect the account?

No. The hour changes who is awake to see a display. It is not an input in the prohibited-software rules, and Boykli does not pick it from your analytics. You save one time of day. The public timing notes are in best time to post on LinkedIn.