Sign in Start free

Free post generator

AI social media post generator for X, LinkedIn and more

Paste release notes, a changelog entry or a pull request description below and get one post each for X, LinkedIn, Bluesky and Threads, free.

Free. No signup. Runs in your browser, nothing is sent.

Each one is written for that network's limits: 280 characters on X1, 3,000 on LinkedIn4, 300 on Bluesky7 and 500 on Threads8. The guide under the tool shows how those formats differ and what a good before-and-after looks like.

By Ilya VolodarskyChecked 12 sources

Sample data

What to paste for the best posts

A generator can only repeat what you give it, so the input decides the output. The strongest inputs have four things in them:

  1. What changed, in one sentence a customer would understand. "Diffs is now on iOS and Android" beats "mobile parity for review surfaces".
  2. Who it's for. Team leads, admins, people on the free plan. A named reader makes the post specific.
  3. One real detail. A number, a limit, a button name. "600 requests per minute" is a fact people remember.
  4. Where to read more. A changelog anchor or docs page, not your home page.

Leave out internal names, ticket numbers and anything you'd be unhappy to see quoted. If a release has five unrelated changes, paste them one at a time. Five small posts usually beat one crowded post.

Pasting a pull request description

A PR description is written for a reviewer, not a customer. It explains how the change works: files touched, tests added, a migration to run. A post needs the opposite end: what someone outside the repo can now do. Before you paste one, add a line at the top in customer words, such as "Admins can now cap AI spend per user." The generator will build on that line instead of the implementation notes.

Two parts of a PR are worth keeping. The title, if it names the feature. And any screenshot or screen recording, which you can attach to the post yourself. Strip the rest: reviewer checklists, issue numbers, test plans and anything that names a customer.

Each channel's format, side by side

The same news needs a different shape on each network because the rules are different. These limits come from each platform's own documentation.

Text limits, links and tags per channel
ChannelText limitLinksTagsWatch out for
X280 characters1; Premium "longer posts" up to 25,00023Every URL counts as 23 characters1No rule in the docs we readEmoji count as 2 characters1
LinkedIn3,000 characters; longer belongs in an article4Paste a URL and a preview card appears5No number set in LinkedIn HelpLong posts fold; LinkedIn doesn't publish where
Bluesky300 graphemes7A link card is an embed, stored apart from the text7Up to 8 extra tags beyond those in the text7An emoji is one grapheme
Threads500 characters85 unique links per post at most8One topic tag, 1 to 50 characters, no periods or ampersands8The API counts emoji by their UTF-8 bytes8

Checked 2026-11-20.

Two of these rules surprise people. On X, a long URL and a short one cost the same 23 characters, so there's no gain in shortening links yourself. On Threads, only one topic tag counts; if you type several, Meta's documentation says the first valid one is used.

X: one idea that survives a fast scroll

X gives most accounts 280 characters1, and the post has to make sense without the link. Lead with the change itself, then one detail. A URL costs 23 characters wherever it sits, and every emoji costs 21.

X Premium raises the ceiling to 25,000 characters with what X calls longer posts2. Its help center adds a catch: if you mention an account after the first 280 characters of a long post, that account isn't notified3. Keep tags and the main point near the top.

Links are a live argument. Nikita Bier, X's head of product at the time, posted in April 2026 that link clicks were higher than they had ever been and advised putting the link in the body with enough context about what readers will see9. Buffer's study of 18.8 million posts found link posts had the lowest engagement of any format, and a median of zero for non-Premium accounts after March 202510. Our tweet generator guide goes deeper on X.

LinkedIn: a short story with a point

LinkedIn allows 3,000 characters per post4. Its help center suggests writing an article if you need more. That much room tempts people to pad, but the feed folds long posts behind a "see more" link, and LinkedIn doesn't say where the fold falls. Write the first two lines as if nobody will click.

A structure that works for product news:

  1. A line about the problem, in the reader's words.
  2. What you changed, in one sentence.
  3. Three short bullets: what people can do now.
  4. Who should care, or a question for the comments.

For links, paste the URL and wait a few seconds; LinkedIn builds a preview card. If the card shows an old image or title, LinkedIn's Post Inspector refreshes it for new posts. Posts already published keep their old preview6. Our LinkedIn post generator guide has more on hooks and the fold.

Bluesky counts graphemes, the characters a reader sees. A flag emoji built from several code points still counts as one. The limit is 300, set in the app.bsky.feed.post record definition in Bluesky's open protocol code7.

A link preview isn't part of the text. It's a separate embed on the post, so you can attach a card and still use most of the 300 for words. The same record can also list up to 8 extra hashtags, in addition to any written in the text7.

Write it as a plain note from a person: what shipped, one detail, the link. Marketing tone stands out badly in a 300-grapheme box.

Threads: 500 characters and one topic

Threads allows 500 characters per text post, up to 5 unique links, and a single topic tag8. Meta's API documentation sets the tag rules: 1 to 50 characters, no periods or ampersands, and only the first valid tag counts8. Pick the one topic your post belongs to, not a list.

The API also counts emoji by their UTF-8 byte length8, so a post heavy with emoji can hit the limit sooner than the visible length suggests. Tone sits between X and LinkedIn: conversational, a little personal, and fine with a question at the end.

Before and after: Linear's Mobile Diffs

Here is a real public changelog entry and the four posts we'd write from it. The drafts are written for this page in Gin's format. Linear did not post them.

Linear changelog, Nov 5, 2026 (summary) · linear.app/changelog

Diffs, Linear's review experience, is now available on iOS and Android. From your phone you can understand changes with Guided Reviews, start a review on mobile and continue on desktop, see what needs your review in your inbox, view preview links, work with review agents and cloud coding agents, and merge approved changes. Open it from the Reviews tab in the Linear mobile app.

XDraft in Gin's format

Linear's code review now runs on your phone. Diffs is on iOS and Android: guided reviews, preview links, and a Merge button. Start a review on the train and finish it at your desk.

X gets the change and the one concrete action (merge) in under 200 characters, with room for the link.

LinkedInDraft in Gin's format

Code review has mostly been a desk job.

Linear just put Diffs, its review view, on iOS and Android. From the Reviews tab you can:
• read a Guided Review of what changed
• open the preview link
• check in on review agents and cloud coding agents
• merge what's approved

Start on your phone, pick it up on your laptop where you left off.

Do you review code on your phone, or does it wait until you're back at a screen?

BlueskyDraft in Gin's format

Linear Diffs is now on iOS and Android. Guided reviews, preview links, review agents and merge, all from the Reviews tab. Start a review on your phone and finish it on desktop.

ThreadsDraft in Gin's format

Linear just shipped Diffs on mobile. Read a guided review, open the preview link and merge, right from the Reviews tab on iOS or Android.

Genuine question: who actually reviews code on their phone? #CodeReview

Notice what changes between them. LinkedIn gets the list and a question. Bluesky gets the plain facts. Threads gets the one topic tag and a conversational ending. The facts never change, and nothing appears that isn't in the changelog.

Before and after: Vercel's skills.sh API

A developer-facing launch, where the detail that sells it is a security property, not a feature list.

Vercel changelog, Jun 5, 2026 (summary) · vercel.com/changelog

The skills.sh API is now available. Authenticate with your project's Vercel OIDC token and query more than 600,000 open-source skills: search them, get a skill's details, and check its security audit. Tokens are short-lived, scoped to a team and project, and rotate automatically. The rate limit is 600 requests per minute per team and project.

XDraft in Gin's format

skills.sh has an API now. Search 600,000+ open-source agent skills, pull a skill's details and check its security audit. Auth is your Vercel OIDC token, so there's no long-lived key to leak.

The no-long-lived-key detail is the part a developer would repeat to a teammate, so it closes the post.

LinkedInDraft in Gin's format

If your agents install skills from the internet, you want two answers first: what does this skill do, and has anyone audited it?

The new skills.sh API from Vercel answers both from code. You can search more than 600,000 open-source skills, fetch a skill's details and check its security audit.

Auth uses your project's Vercel OIDC token. Tokens are short-lived, scoped to a team and project, and rotate on their own. No long-lived secret to leak or rotate.

Rate limit: 600 requests a minute per team and project.

Mistakes generators make, and how to catch them

  • Invented numbers. If the source says "faster", the draft must not say "3x faster". Search each draft for digits and match them to the input.
  • Hashtag piles. Five tags on X or LinkedIn read as spam. Threads only uses one anyway.
  • The same post everywhere. A LinkedIn story pasted into X gets cut off; an X one-liner on LinkedIn looks lazy.
  • Links that go nowhere useful. Point to the changelog entry or docs page for that change.
  • Hype words. "Revolutionary", "game-changing", "excited to announce". Delete them and the post gets stronger.

Read the first line of each draft on its own. If it doesn't tell a stranger what changed, rewrite that line before anything else.

Where Gin goes further than a generator

The free tool fills in templates in your browser, one paste at a time. Gin, the agent behind it, writes with a model and needs no paste. When a coding-agent session ends, its summary, commit subjects and pull request links arrive as a message. Gin decides which channels each one deserves, writes a post for each, and puts them in a deck on your phone.

It learns from what you do with the cards. A right swipe posts; a left swipe passes and teaches it what to skip. Holding a card opens remix chips (More technical, Shorter, Funnier, More specific, Add numbers, Less salesy), plus Wrong channel and Not true. Over time the drafts sound more like the ones you approve. Our AI social media manager guide compares that approach with tools that schedule what you write.

From drafts to autopilot

Three right swipes in a row on one channel and Gin offers to post there on its own. It still pings you two hours before each autopilot post, and one tap pauses all of them. X, your LinkedIn profile, Bluesky and Dev.to get posted directly. For a LinkedIn page (until LinkedIn approves Gin for page posting) and Threads, Gin copies the text and opens the composer.

Gin is free for up to 3 products and 20 published posts a month; drafts, swipes and remixes are free on every plan. Autopilot comes with the paid plans. See pricing, or open the app at trygin.ai/app.

Questions

Is the post generator free?

Yes. Paste release notes, a changelog entry or a pull request description and you get posts for X, LinkedIn, Bluesky and Threads without signing up.

Why does each channel get a different post?

The limits and habits differ. X allows 280 characters for most accounts, LinkedIn 3,000, Bluesky 300 graphemes and Threads 500 characters with one topic tag. A post written for one usually reads badly on another.

How many characters does a link use?

On X every URL counts as 23 characters, however long it is. On Bluesky a link card is a separate embed and does not use the 300-grapheme text limit. Threads allows at most 5 unique links in a post.

How many hashtags should a post have?

Threads uses only one topic tag per post. X and LinkedIn set no number in the documentation we read, so use one relevant tag or none, and never a list.

How is Gin different from this generator?

The generator fills in templates for one paste at a time. Gin gets a message when a coding-agent session ends (its summary, commit subjects and pull request links), picks the channels, writes each post, learns from your swipes and remixes, and can post on autopilot after three approvals in a row.

Related guides

Sources

Sources checked . Prices and limits change; each link goes to the page we read, on the day shown.

  1. X: counting characters, read
  2. X Help: X Premium how to (longer posts), read
  3. X Help: types of posts (longer posts), read
  4. LinkedIn Help: post and share updates, read
  5. LinkedIn Help: share articles or links, read
  6. LinkedIn Help: Post Inspector, read
  7. Bluesky post record (app.bsky.feed.post), read
  8. Meta: Threads posts, read
  9. Nikita Bier on links, X, Apr 12, 2026, read
  10. Buffer: do posts with links affect performance on X?, read
  11. Linear changelog: Mobile Diffs, read
  12. Vercel changelog: the skills.sh API is now available, read

Work at one of these companies and see something wrong? Send us a correction.

Ilya Volodarsky

Ilya VolodarskyMaker of Gin, co-founder of Winding AI

Ilya makes Gin at Winding AI. Before that he co-founded Segment. His team ships close to 30 agent-written pull requests a day, which is why Gin exists: someone has to tell people what shipped.

GitHubX

You ship. Gin posts.

Get the app, paste one message into your coding agent, and swipe through your first cards tonight.